<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Liane Cafarate on Medium]]></title>
        <description><![CDATA[Stories by Liane Cafarate on Medium]]></description>
        <link>https://medium.com/@lianecafarate?source=rss-918acad7f898------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*e-3mbiRNmrAYZh6_fMh3ow.png</url>
            <title>Stories by Liane Cafarate on Medium</title>
            <link>https://medium.com/@lianecafarate?source=rss-918acad7f898------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 00:12:19 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@lianecafarate/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Real Bottleneck in AI Isn’t Building It, It’s Everything After]]></title>
            <link>https://lianecafarate.medium.com/the-real-bottleneck-in-ai-isnt-building-it-it-s-everything-after-c9bd3263532b?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/c9bd3263532b</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[customer-success]]></category>
            <category><![CDATA[change-management]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Sun, 13 Sep 2026 02:05:59 GMT</pubDate>
            <atom:updated>2026-09-13T15:54:59.920Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>Notes from the </em></strong><a href="https://www.oaiconference.com/virtual-conference"><strong><em>Optimized AI Global Virtual Summit</em></strong></a><strong><em> where infrastructure, memory, cost, and adoption data all pointed at the same gap</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YJX7Xw5EvA9tgtvoZH_WlQ.jpeg" /></figure><p>Eight speakers, one day, one live virtual summit. I took notes on the talks, and by the time the day’s closing speaker named the pattern directly, other speakers had already described the same gap without naming it. <a href="https://www.oaiconference.com/virtual-conference">Optimized AI’s Global Virtual Summit</a> on September 12 turned out to be less a set of separate talks and more one long argument, made from different angles, about the same problem: a lot of impressive AI capability, and the much bigger challenge of it actually landing as changed work inside a real organization.</p><p>This lands close to home. My job is, on paper, exactly this last mile: mapping how a customer actually works today, then figuring out what changes when AI enters the workflow. I went in expecting one useful idea. I came out with the same conclusion confirmed again and again, from angles I don’t usually see put together in one room: access, infrastructure, memory, cost measurement, and raw adoption numbers.</p><h3>AI Access Is Not AI Productivity</h3><p><a href="https://www.linkedin.com/in/carter-abdallah/">Carter Abdallah</a> (DevTech Engineer, NVIDIA) opened the day from the access side. His starting observation: AI has clearly improved productivity somewhere, but that jump hasn’t shown up yet for non-technical teams the way it has for engineers. He put it bluntly: companies keep telling people AI is incredible and that everyone should just try it and figure it out, then quietly blame the user when it doesn’t click, as if they just weren’t using it the right way.</p><p>To actually test where the gap was, NVIDIA embedded Forward Deployed Interns on non-technical teams, Creative, Startup, and Communications, and had them work from the inside instead of guessing from an engineering seat. What they found was a short, consistent list of blockers that had nothing to do with AI itself:</p><ul><li>How do I share my work with others?</li><li>I changed something and don’t like, how to go back?</li><li>I want the data to persist</li><li>If something breaks, how do I know? How do I fix it?</li><li>I need to give it sensitive data. how I do that securely?</li><li>I want it to update on its own, how?</li></ul><p>None of these are AI problems. They are the software-engineering primitives that decide whether an AI prototype ever becomes something a non-technical team can actually run day to day, and once a team has them in place, they fade into the background entirely, nobody thinks about their web host once it works.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fdatawrapper.dwcdn.net%2FJHU06%2F2%2F&amp;display_name=Datawrapper&amp;url=https%3A%2F%2Fdatawrapper.dwcdn.net%2FJHU06%2F2%2F&amp;image=https%3A%2F%2Fdatawrapper.dwcdn.net%2FJHU06%2Fplain-s.png%3Fv%3D2&amp;type=text%2Fhtml&amp;schema=dwcdn" width="600" height="359" frameborder="0" scrolling="no"><a href="https://medium.com/media/9baf2760344e738f8205e61842d4830e/href">https://medium.com/media/9baf2760344e738f8205e61842d4830e/href</a></iframe><p>There was also a real assumption sitting underneath a lot of this optimism: that once anyone can build software with AI, people would stop paying for tools like Salesforce. Carter’s point was that the software itself was never really the expensive part. What’s hard to replace is the service wrapped around it, because when something a non-technical person built with AI breaks, there’s no one to call. A SaaS subscription’s price includes exactly that safety net, and building it yourself doesn’t solve for it.</p><p>For now, the honest fixes are closer to workarounds than platforms: deploying MCP to an internal cluster, using Google Drive as a makeshift database, leaning on existing SaaS tools that already have secure access to sensitive data, or borrowing skills to manage versions on GitLab. His ask to the room was direct: find where AI hasn’t reached yet, and ask honestly whether that’s an access problem or a missing-primitives problem, before assuming it’s a model problem.</p><p>His CTA to the room was three-part:</p><ol><li>Identify the areas where AI hasn’t permeated yet, and ask honestly whether that’s an access problem or a missing-primitives problem before assuming it’s a model problem;</li><li>Break down those barriers, safely;</li><li>Forward deploy your technical teams, the same move NVIDIA had already tested with its own interns.</li></ol><p>He left the room with three open questions rather than answers:</p><ul><li>Should every team have an engineer?</li><li>Is this a temporary problem?</li><li>How do you actually balance upskilling your own people against hiring or buying the gap closed?</li></ul><h3>Building AI Systems You Can Trust</h3><p><a href="https://www.linkedin.com/in/muazmazahid/">Muazma Zahid</a> (Group Product Manager, Google Data Cloud) followed with a talk on building trustworthy AI systems across data, models, and agents in production, by every account a strong one, but I didn’t capture slides since she mentioned she’d share the deck afterward, and that link isn’t live yet. I’ll fold in her actual points once it is.</p><h3>Serving Agents Isn’t Serving Chat: AI Infrastructure At Scale For Forward-Deployed Engineers</h3><p><a href="https://www.linkedin.com/in/sanjeevg89/">Sanjeev Ganjihal</a> (Cloud Platform &amp; Infrastructure Engineer, Google) made the same argument a layer down, for infrastructure. His numbers were the sharpest:</p><blockquote>an agent produces roughly four times the tokens of a single chat exchange, and a multi-agent system produces roughly fifteen times as many (both cited on his slides to Anthropic’s 2025 research), with production agents running close to a 100 to 1 ratio of input to output tokens (Manus, 2025).</blockquote><p>On one Nvidia H100 GPU with 80GB of memory, that difference is the whole story: room for around 170 concurrent chat sessions, or only about 4 concurrent agent sessions, because an agent’s working memory is that much heavier. His line:</p><blockquote>“serving agents isn’t serving chat,”</blockquote><p>and most infrastructure teams are still sizing for chat.</p><h3>Agent Memory: The Cure for Agentic AI&#39;s Hidden Cost Problem</h3><p><a href="https://www.linkedin.com/in/ktalamad/">Kartik Talamadupula</a> (Distinguished Architect, Oracle) took the same gap into agent memory. His example: a coding agent fixes a bug on day one, real progress, files inspected, a failed approach ruled out, tests run. A session restart on day two makes it redo all of it, because nothing persisted. His slide’s line:</p><blockquote>“without memory, every turn pays to rediscover what the agent already learned.”</blockquote><p>His actual warning was sharper than that, though: a correct final answer can hide exactly where a memory system is failing, so he argued for judging memory as a system, whether the right evidence got captured, got found again, got used correctly, and actually improved the result, not just whether the agent happened to get this one answer right.</p><h3>Your AI Coding Bill Isn’t the Problem</h3><p><a href="https://www.linkedin.com/in/emilieschario/">Emilie Schario</a> (VP of Engineering, Anaconda) took the same gap into cost. She opened with the request she says she hears every week from engineering leaders: a spending cap on AI coding tools. Her argument was that a cap does not control cost, it moves it. A developer’s refactor freezes mid-task when the org hits its daily limit, the agent’s reasoning is lost, and someone has to manually untangle a half-finished migration, which raised the real question underneath the whole talk:</p><blockquote>did the company actually control costs, or destroy more value than it saved?</blockquote><p>Her line: production systems do not get shut down mid-task because a budget alert fired, and this should not either, especially now that AI spend behaves like a variable cost, it moves with model choice, task complexity, and retries, rather than a flat seat license.</p><p>Her proposed fix was a real metric, AI Cost to Ship: total AI spend divided by merged pull requests, counting the spend on abandoned attempts too, not just the ones that worked, because excluding failed work would make the org look more efficient than it actually is.</p><p>In her worked example, $12,000 in spend against 600 merged PRs gives a $20 unit cost, a number to track over time, not a target to optimize in isolation. Reading the trend matters more than the number itself: spend rising while PRs stay flat means AI activity isn’t becoming output; spend and PRs rising together means the team added capacity at a stable cost; PRs rising faster than spend means the unit cost of shipped work is genuinely falling; and adoption rising while unit cost worsens usually means rework or review congestion is quietly eating the gains. She was direct about the metric’s limits too: it can’t prove AI caused a PR to merge, that every PR carries equal value, or that a lower number isn’t hiding quality shortcuts, so she paired it with a warning against optimizing it in isolation, since teams can always shrink PRs to pad the denominator or defer real exploratory work to make the ratio look better. The same shape, she argued, generalizes past engineering:</p><blockquote>support needs AI spend paired with resolved issues and CSAT, sales with qualified or closed-won value, marketing with qualified demand created.</blockquote><p>She closed by naming where she thinks enterprise AI measurement actually is right now: access told leaders who could use AI, adoption told them who did, and now the real question is what the investment produces.</p><h3>A World Where Anyone Can Build and Discover Anything</h3><p><a href="https://www.linkedin.com/in/dynamicwebpaige/">Paige Bailey</a> (AI Developer Relations Lead, Google DeepMind) followed with a tour of recent Google launches, AlphaGenome and Gemma among them, more a look at what Google has been shipping than a single argument, so it sits outside the throughline here.</p><h3>Agent Harness For Self-Learning AI Agents</h3><p><a href="https://www.linkedin.com/in/reshmalaljagadheesh/">Reshma Lal Jagadheesh</a> (Senior Data Science Manager, The Home Depot) closed the gap out with the adoption data itself. She shared some numbers that stuck with me most: as of this year, an estimated 84% of people have never used AI at all, 16% use a free chatbot, and only about 0.04% use any kind of AI coding tool, the closest population to the frontier work the room was actually discussing, self-learning agents.</p><blockquote>“The gap is the opportunity,”</blockquote><p>she said, and her talk was about closing it technically: a three-role loop where one part of the system acts, a second critiques what happened, and a third folds the lesson back in as a small update instead of rewriting the agent’s whole memory. She built that design specifically to avoid what she called context collapse: in the data she cited, a routine step meant to compress an agent’s accumulated knowledge instead took it from 66.7% accuracy down to 57.1%, worse than not compressing at all.</p><h3>The Last Mile of AI is Longer Than You Think</h3><p><a href="https://www.linkedin.com/in/joelgrus/">Joel Grus</a> (Principal AI Engineer, CPS Energy) closed the day with the talk that named what the rest of the day had been circling.</p><p>He opened by sitting himself between the two camps rather than picking one, calling himself “an AI enthusiast and an AI skeptic,” and used real, dated social media reactions to frontier model demos to make the tension concrete: one post marveling that someone had built a full immersive 3D game from a single prompt, next to another wondering, given that models can now resolve Millennium Prize-level math problems, why annualized US GDP growth was still sitting around 1.5%. He didn’t resolve that tension. He used it to ask the question the whole day had been building toward: if the models are this capable, why isn’t it showing up anywhere near that scale?</p><p>His answer:</p><blockquote>the coding sprint, the part AI actually speeds up, was never the bottleneck.</blockquote><p>By his own estimate, writing the code is roughly 2% of the total time it takes to ship anything into a real organization. The other 98% is what he called the last mile: architectural review, security review, risk review, funding approval, project manager onboarding, change approval, QA, end-user training, and go-live communications, a sequence he described, only half joking, as “the Bureau of Eternal Sign-offs.”</p><blockquote>Speeding up the 2% doesn’t touch the 98%.</blockquote><p>He was blunt about companies not even understanding their own workflows well enough to know what to automate: an organization that doesn’t fully understand how it works internally is not going to figure out how its customers work either, and understanding customer work is the actual hard part. His sharpest line:</p><blockquote>“knowing how to automate doesn’t mean you’re automating the right thing,”</blockquote><p>illustrated with a real, dated example, an AI assistant confidently giving wrong guidance on a Stripe subscription API call, then having to be corrected mid-task by the person who actually knew the constraint. The capability is becoming a commodity. The deployment isn’t.</p><p>That’s where he placed the Forward Deployed Engineer, the same role Carter Abdallah’s NVIDIA interns were effectively playing:</p><blockquote>“how many FDEs does it take to change a workflow? Only one, but the workflow has to want to change.”</blockquote><p>Even when a team does everything right technically, the last obstacle is still whether people adopt it: plenty of engineers are openly skeptical of AI tools, one department using something well doesn’t mean the rest of the org moves at anywhere near the same speed, and a small proof of concept can bring skeptics along, but a team that simply doesn’t want to change is a much harder problem than a technical one. His closing argument was that augmentation, humans and AI working together with humans still catching the AI’s mistakes, is likely to last a lot longer than the industry expects, because the last mile is organizational, not technical, and organizations change slower than models do.</p><h3><strong>What I am taking back to my own work</strong></h3><p>This is not new to me, even if the AI part is. Years ago, working a migration for a government customer in Brazil, the product was never what slowed things down. Every change needed sign-off from teams that were not mine, most memorably the database team, whose approval could sit for weeks before anything moved. The last mile Joel described, access, approval, another team’s calendar, is the exact one I watched slow that migration, long before anyone was talking about AI.</p><p>Hearing the same conclusion from every direction, an access story, an infrastructure deep dive, a memory framework, a cost model, real adoption numbers, and a closing talk that named the pattern directly, confirmed something I already knew:</p><blockquote>the technology was never the real problem.</blockquote><p>The organizational one is: whether a workflow actually wants to change, whether the system around the model can hold state, whether anyone is measuring output instead of activity. That is, not coincidentally, most of what my own job actually is. Less engineering, more change management. The building is mostly the easy part. What many are still struggling with is a reliable playbook for everything that comes after, and that gap is where the next few years of this job will actually happen.</p><p>🙌 Thank you, <a href="https://www.linkedin.com/in/hamzafarooq/">Hamza Farooq</a> and <a href="https://www.linkedin.com/in/julie-koon-koon-251a06b2/">Julie Koon Koon</a>, who organized and hosted the whole day. Keeping me engaged for several hours on a Saturday is its own kind of achievement. If you want to catch the next one, <a href="https://www.oaiconference.com/">Optimized AI</a> is heading to Atlanta in March 2027, with early-bird tickets already available.</p><p>😍 One more thing worth naming: four of the eight speakers were women, Muazma Zahid, Emilie Schario, Paige Bailey, and Reshma Lal Jagadheesh, on a lineup built entirely around infrastructure, memory systems, and cost engineering. That should not be remarkable. It still is, often enough that it is worth saying when it is not.</p><p>💬 If you sat through this same conference, I would like to know which talk landed closest to what you are actually dealing with. And if you did not, I am curious whether this gap shows up the same way in your corner of the industry, or differently.👀</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c9bd3263532b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[6 Agentic Ways to Use AI as a CSM]]></title>
            <link>https://lianecafarate.medium.com/6-agentic-ways-to-use-ai-as-a-csm-999ee49666ab?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/999ee49666ab</guid>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[customer-success]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Tue, 25 Aug 2026 14:41:01 GMT</pubDate>
            <atom:updated>2026-08-25T14:41:01.234Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l-qWfdU8byPEylF0AiBxlg.jpeg" /><figcaption>6 Agentic Ways to Use AI as a CSM</figcaption></figure><p>Last Thursday, I attended “<em>6 Agentic Ways to Use AI as a CSM</em>,” hosted by <a href="https://www.linkedin.com/in/markstagi/">Mark Stagi</a>, VP of Customer Success at <a href="https://www.avoma.com/">Avoma</a>.</p><p>It was so great getting a room full of CS professionals together in real life, away from Zoom , for a couple of hours. We were a small, intimate group, so it felt less like a panel and more like friends talking shop: work, fears, challenges, and how to get better at it.</p><p>Mark opened with a short teaser video, built using Claude. Ten minutes. Something that used to take a motion designer hours, maybe days. That’s the world we’re operating in now.</p><h3>The 6 (plus 1) Agentic Use Cases</h3><p>The core of the night was six agentic ways to use AI as a CSM, plus a bonus seventh. So here they are:</p><ol><li><strong>Pre-Call Briefing:</strong> Instead of walking in with tactical notes (completed tasks, next steps), use AI to pull in relevant industry trends and surface something more strategic. Ideally you’d already be an industry expert, but if you aren’t yet, at least you show up well-versed in your customers’ fields.</li><li><strong>Churn &amp; Risk Detection: </strong>Most CS platforms track usage, but usage isn’t the whole story. A champion leaving, or a shift in who holds power on the account, can matter more than any dashboard metric.</li><li><strong>Automated Business Review: </strong>Ask your point of contact one question: “<em>How can I help you get promoted?</em>” Then find the three metrics that would prove their value to leadership. Use AI to pressure-test those metrics (sometimes what your customer thinks they need isn’t what they actually need), surface quick wins, and map out the next logical steps for growth. <a href="https://gamma.app/">Gamma</a> was the recommended tool here for designing your BR deck.</li><li><strong>Expansion Signal Surfacing: </strong>Look for the gap between what a customer has and what they could have. Avoma’s own example: if sales professionals are the target buyer and a client has half as many seats as their sales team size on LinkedIn, that gap is your expansion signal.</li><li><strong>Post-Call Task Automation</strong>: The follow-up email, the CRM update, the new task, all handled automatically after the call ends.</li><li><strong>Onboarding Health</strong>: AI that doesn’t just flag that onboarding stalled, but tells you why.</li><li><strong>Personal CSM Growth</strong>: Using AI to sharpen your own skills as a CSM, not just your output.</li></ol><h3>Shared Challenges</h3><p>This is the part I really loved about the evening: the honesty from the group. Everyone started sharing their challenges and ideas to solve them.</p><ol><li><strong>Tool sprawl</strong>: Everyone’s stacking new AI tools on top of an already bloated stack, and every addition needs a real justification. (If you’ve ever opened your tools menu and felt a small wave of panic, you know the feeling.)</li><li><strong>Sales-to-CSM handoffs</strong>: In some cases, years of context buried in tools like Gong get lost the moment an account changes hands. <a href="https://www.glean.com/">Glean</a> came up as one way to surface a clean timeline instead of starting from zero.</li><li><strong>Tracking external signals:</strong> Catching account intelligence and industry shifts outside your own tools. <a href="https://exa.ai/">EXA</a> and <a href="https://gojiberry.ai/">Gojiberry</a> were the two recommended here.</li><li><strong>Cross-department expansion</strong>: Growing an account into a new department relies on strong, proactive discovery. Ask targeted questions about current priorities and pain points, and use AI to surface industry-specific trends: <em>“What should other departments care about right now?” “Are you already running into problem X?”</em></li><li><strong>Feeling overwhelmed by AI:</strong> This was the most shared sentiment in the room. To top it off, with minimal organizational guidance, team members must take self-initiative to uncover workflow inefficiencies. One personal insight: after an AI session, ask it to summarize how you two worked together and where else you could automate.</li><li><strong>Siloed adoption:</strong> Teams building the same agent twice because nobody knew someone else already built it. The fix is unglamorous: regular “<em>AI show and tell</em>” sessions and a shared library people can actually search.</li><li><strong>AI-averse customers:</strong> Some clients won’t let you record or transcribe. The takeaway here was reassuring: this is a “<em>not now</em>,” not a permanent no. It’s the same trade-off as turning on location tracking in Google Maps, weighing privacy against utility. Self-hosted models can help with the risk-averse ones, but the real move is discovery: find out why they’re hesitant (like a missing internal policy) and help them build one. One creative workaround from the room: when a customer won’t disclose their use case directly, reverse-engineer it. Cross-reference file names or job titles against LinkedIn to piece the picture together instead.</li><li><strong>Skill atrophy:</strong> The quieter fear underneath all of this: what happens to our own skills if AI does too much of the thinking for us.</li></ol><h3><strong>What Zoo</strong>m Can’t Show You</h3><p>At the end, our discussion evolved to this topic and it was quite valuable.</p><ul><li><strong>Sit with your customer:</strong> Spend a full day watching how they actually work, not how they describe their workflow on a call. Even your happiest customers are usually only using a fraction of what you built. You don’t find that fraction on a Zoom call. You find it by being in the room.</li><li><strong>Speak their language: </strong>Pay attention to the specific words your customer uses, run it through AI, and mirror that vocabulary back in your own communications. Small thing, but it signals you’re paying attention on their terms, not yours.</li><li><strong>Chase frustration, not just time:</strong> Not every time-consuming task is a problem, some users genuinely enjoy them. Separate what’s slow from what’s actually frustrating, and prioritize fixing the friction that makes people angry, not just the friction that takes a while.</li><li><strong>Observing customer interactions firsthand: </strong>A cross-functional team watched a recorded trial setup with an Account Executive, and it was uncomfortable. People questioned the AE’s prompts, and questioned how the user interpreted the interface. Uncomfortable, but exactly the kind of detail that never comes up in a normal debrief.</li><li><strong>Encourage continued exploration:</strong> Many times, customers stop exploring the moment their main pain point is solved. They won’t tell you about the next ten small frictions unless you’re there to notice them yourself, or ask the right question.</li><li><strong>Re-engaging customer pain</strong>: Effective discovery requires prompting customers to relive initial issues by asking specific contextual questions (when did the issue first occur, who was present). Re-evaluating these pain points reinforces the necessity of solving them and highlights product value.</li></ul><h3>Wrapping Up</h3><p>Here’s the thing that ties the whole night together: every single one of these agentic workflows exists to buy CSMs back the time and information to do this kind of work: the in-person, sit-with-the-customer, notice-what-they-don’t-say kind. AI can brief you, flag risk, and draft the follow-up. It can’t sit next to your customer and watch them struggle with a feature they never mentioned.</p><p>If you walk away with one takeaway from this post, I hope it’s this:</p><blockquote>Definitely use AI to improve and automate. But remember, a human in the loop remains essential.</blockquote><p>Thanks again to Mark and the whole Avoma team for putting this together, and to everyone in the room for making it worth showing up in person for.</p><p>Thanks for reading! 🙏 If you found this text helpful, please like and share it with your network 🔗.</p><p>What’s the one thing from your own CS work that AI still can’t replace? I’d love to hear it in the comments. 💬</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=999ee49666ab" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Customer Success in the Age of AI]]></title>
            <link>https://lianecafarate.medium.com/customer-success-in-the-age-of-ai-3c8f3d823a91?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/3c8f3d823a91</guid>
            <category><![CDATA[customer-success]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Tue, 16 Jun 2026 14:50:38 GMT</pubDate>
            <atom:updated>2026-06-16T14:50:38.555Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gWXtYBjBwpAKKKo1lQiE7Q.png" /><figcaption>Customer Success in the Age of AI</figcaption></figure><p>Two weeks ago, I attended an event called Customer Success in the Age of AI at <a href="https://www.mongodb.com/">MongoDB</a> (I was incredibly glad to join, thanks to a thoughtful invitation from <a href="https://www.linkedin.com/in/brookefasciana/">Brooke Fasciana</a>).</p><p>To be honest, I’ve been trying to be very intentional about the events I join lately. Everything is branded with AI now, and often these panels just don’t add real value. Some are too generic, repeating the same old speech about how “you just gotta use AI” without offering anything that actually moves the needle. Others are so highly technical and specific that they lose the broader context. So, I was genuinely interested when I heard about this one. It finally seemed relevant to my actual world.</p><p>One of the main topics brought up was the tension between feeling optimistic or anxious about AI replacing our jobs. I won’t go into every detail, but I will say two things:</p><ul><li>First, yes, AI will shift many roles. If you aren’t feeling at least a little bit of healthy pressure or uncertainty about it, you are missing the scale of this transition. 😬 Second, you shouldn’t waste energy on things you can’t control. Instead, focus on being optimistic about what AI can do for you, and use it to become a better version of yourself.✨</li><li>At this point, pretty much everyone in tech knows there is no way to avoid AI (well, maybe if you don’t need money, you could get away with it! 💸😂). Jokes aside, how can we <strong>REALLY</strong> use it? What does that even mean in our day-to-day?</li></ul><p>According to the <a href="https://www.mongodb.com/">MongoDB</a> CS leaders, specifically <a href="https://www.linkedin.com/in/darcieunderhill/">Darcie Underhill</a>, <a href="https://www.linkedin.com/in/sharpevan/">Evan Sharp</a>, and <a href="https://www.linkedin.com/in/kodyj/">Kody Jones</a>, AI is fundamentally shifting Customer Success in three ways:</p><ul><li>📉 <strong>Less manual work: </strong>Although some argue that new tech can occasionally create more work, we can pretty much all agree that automating repetitive tasks is a net positive.</li><li>🧠 <strong>More strategic work:</strong> To be fair, CS was always supposed to be strategic. I believe what they meant is, because you finally have less manual baggage, you actually have the bandwidth to step into that strategic advisor role.</li><li>📊 <strong>A higher skill bar: </strong>You can’t just teach customers how to navigate basic software features anymore. That won’t cut it.</li></ul><p>The speakers also emphasized that CS isn’t dead, it’s evolving. While I agree, this evolution is exactly what I want to discuss.🎯</p><p>There are CS influencers who believe a CSM will essentially need to become a “forward deployment engineer.” From my experience, finding a professional who truly excels at high-level technical architecture while possessing world-class soft skills has always been incredibly rare. Yet, that is exactly the intersection companies are now aggressively looking for, expecting AI to help bridge the gap.</p><p>The consensus from leaders at MongoDB and across the industry is clear: navigating corporate politics, relationships, and strategic nuances matters more than ever. But if you aren’t also structurally and technically fluent, you will struggle to survive.⚠️</p><p>This isn’t to say that Customer Success Managers were never technical before. Leaders like <a href="https://www.linkedin.com/in/janine-sneed-1aa2696/">Janine Sneed</a> have long pointed out that you simply cannot build genuine customer trust without deep product and technical knowledge[1]. But historically, a non-technical CSM could still deliver visible value. The baseline has shifted. Without technical fluency, you lose the leverage required to influence a customer’s strategy in the age of AI.</p><p>But let’s go a bit deeper, because I believe many of you (me included) are asking: How do I actually develop technical fluency? Isn’t that way too broad? Technology changes faster than ever. What exactly should we learn, and how deep do we need to go? 🤔</p><p><a href="https://www.linkedin.com/in/maxine-ohayon-316999ba/">Maxine Ohayon</a>, a CS Leader at Deel, recently posted about this shift, pointing out the changing expectations[2]. She admitted she used to hire “people-first,” focusing heavily on warmth and relationships, knowing candidates could learn the platform later. Now, she is looking for something else entirely. She wants candidates who understand data architecture, can map a workflow, spot an integration gap, and speak the language of an ops team.</p><p>But it was this specific quote from her that stuck with me:</p><blockquote>“And they do it while still reading the room better than anyone I’ve hired before.”👁️</blockquote><p>Now that AI handles the surface layer of customer engagement, operational credibility has become our new currency. We have to earn our seat at the table by understanding exactly how a customer’s business runs (which has always been true), where AI fits inside it (the new variable), and where it still desperately needs a human hand.🤝</p><p><a href="https://www.linkedin.com/in/gozdegorce/">Gozde Gorce</a>, an AI advisor at USF, echoed a similar perspective during a recent discussion on what happens when AI is embedded into every customer workflow[3]:</p><blockquote>Customer-facing roles, along with other roles, are becoming more about directing agents, not managing accounts.🤖</blockquote><p>The conclusion? The roles are compressing. Customer Success, Account Management, Support, and Ops are no longer four separate seats. They are becoming one unified professional who points AI agents at the right problems. We see this shift everywhere as teams are forced to do more with less. Post-sale work is moving rapidly toward product, data, and workflows. Again, being “good with people” is just table stakes now, not your primary pitch.</p><p>As Gozde put it:</p><blockquote>“You are being hired to direct the work and decide where human attention goes, not to babysit a queue. Pitch yourself as the person who runs the system, not the one who answers the tickets.”⚙️</blockquote><p>In conclusion, I agree with the MongoDB CS leaders that the CSM role won’t become the next forward deployment engineer (at least, not yet 😅). We still need world-class soft skills more than ever to navigate complex corporate politics and build genuine trust. The real challenge is that soft skills alone are no longer a shield against technological shifts. The future belongs to those who can protect the relationship while bringing enough technical fluency to the table to truly understand, direct, and maximize the systems their customers are running.</p><p>We don’t need to learn how to build the AI from scratch. We need to learn how to direct it. As the landscape continues to compress, the baseline has permanently changed.</p><p>The question is:</p><blockquote>Are we building the technical fluency required to lead this shift, or are we waiting to be swept up by it?</blockquote><h3>👋 Thanks for reading!</h3><p>If this resonated with you (or if you totally or partially disagree), I’d love to hear your thoughts.</p><p>💬 <strong>How are you navigating the balance between technical fluency and relationship building in your role? Drop a comment below!</strong></p><p>If you found this valuable, please give it a 👍 <strong>Like</strong>, <strong>Share</strong> it with your network, or 🔔 <strong>Follow</strong> for more unfiltered thoughts on the future of tech and Customer Success! Let’s keep the conversation going! 👇</p><h3>References</h3><p>[1] <a href="https://janinesneed.substack.com/p/15-the-technical-csm">Scaling Customer Success: The Technical CSM</a></p><p>[2] <a href="https://www.linkedin.com/posts/maxine-ohayon-316999ba_the-csm-role-isnt-evolving-its-being-reborn-activity-7470555491941806081-vXX0?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAAAEElXEBWagiuyEB17m7ooU2rh3pdJqOys0">LinkedIn Post from Maxine Ohayon</a></p><p>[3] <a href="https://www.linkedin.com/posts/gozdegorce_i-get-asked-this-a-lot-how-is-cs-changing-ugcPost-7470892820652806145-QF-w?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAAAEElXEBWagiuyEB17m7ooU2rh3pdJqOys0">LinkedIn Post from Gozde Corce</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3c8f3d823a91" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your APIs aren't ready for Agentic AI]]></title>
            <link>https://lianecafarate.medium.com/your-apis-arent-ready-for-agentic-ai-4d943756a2e0?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/4d943756a2e0</guid>
            <category><![CDATA[ai-agents-in-action]]></category>
            <category><![CDATA[wso2-api-manager]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[open-api]]></category>
            <category><![CDATA[wso2]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Thu, 11 Sep 2025 13:45:33 GMT</pubDate>
            <atom:updated>2025-09-11T13:45:33.695Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*buHgvm4c0gHyRD7P" /><figcaption>Photo by <a href="https://unsplash.com/@steve_j?utm_source=medium&amp;utm_medium=referral">Rubidium Beach</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>At last week&#39;s <a href="https://apiworld.co/">API World conference</a> 🌐 in Santa Clara, a session by <a href="https://www.linkedin.com/in/davidon/">David O&#39;Neill</a>, COO at <a href="https://www.linkedin.com/company/apicontext/">APIContext</a>, delivered a critical message for anyone in the API space. The talk, titled <strong><em>&quot;Your APIs aren&#39;t ready for Agentic AI — OpenAPI can fix that&quot;</em></strong>, argued that the rise of autonomous AI agents is fundamentally changing how APIs are consumed, and our current practices are not good enough.</p><p>The core takeaway was clear:</p><blockquote>In the era of agentic AI, your OpenAPI specification <em>is</em> your API, and it’s time to treat it that way.</blockquote><h3>The Agentic Shift is Here</h3><p>The talk highlighted a paradigm shift in API consumption.</p><blockquote>Increasingly, APIs are not called by human developers but by autonomous AI agents <em>🤖</em>.</blockquote><p>Unlike humans, these agents don’t read prose documentation; they rely entirely on machine-readable specifications to understand and interact with an API. <strong>This shift exposes significant gaps in API documentation and specification accuracy.</strong> The success of AI agent adoption hinges on stable and interoperable APIs, making the readiness of the API, not the AI itself, the primary challenge 💪.</p><h3>The Pervasive Problem of API “Drift” 😟</h3><p>A central theme of the presentation was “API drift&quot;: the divergence between an API’s actual behavior and its documented specification. According to an APIContext study, the problem is rampant 📊:</p><ul><li><strong>25%</strong> of public-facing APIs are non-conformant to their own spec. In a separate analysis, 75% of tested APIs had at least one non-conformant endpoint.</li><li><strong>89%</strong> of API specs had not been updated in the last six months.</li></ul><p>This drift creates serious issues because agents don’t guess; they parse. Errors in schemas, vague enums, or undefined errors result in breakage.</p><blockquote>Common problems observed in public APIs include poorly defined error codes, ambiguous paths, incorrect or unspecified data returns, and data in the wrong format. For an agent, this is not a minor inconvenience; it’s a critical failure <em>❌</em>.</blockquote><h3>OpenAPI as Critical Infrastructure 🏗️</h3><p>The talk stressed that OpenAPI specifications are not just documentation; they are the contract that your API consumers depend on 📜.</p><blockquote>A validated, versioned spec builds trust and enables automation. In this new landscape, robust API governance is no longer a “nice to have” but a necessity <em>✅</em>.</blockquote><p>To combat drift, specifications must be managed like code 💻. This involves several key practices:</p><ul><li>Enforcing CI/CD validation against production behavior.</li><li>Integrating OpenAPI into the full development lifecycle.</li><li>Tracking versions and maintaining clear changelogs.</li><li>Ensuring continuous enforcement of spec quality and production monitoring.</li></ul><h3>Are You Agent Ready? 🤔</h3><p>The presentation provided a clear “wishlist” for what AI agents require from an API:</p><ul><li>A discoverable and current OpenAPI Description.</li><li>Accurate types, enums, and status codes.</li><li>Clear error definitions for fallback logic.</li><li>Versioned, stable, and documented changes.</li></ul><p>A subsequent slide offered a checklist for teams to assess their readiness, asking if the spec reflects live behavior, if it’s running OAS 3.0 or higher, if common flows are documented, and if it’s updated alongside backend changes.</p><h3>Beyond the Spec: Advanced Infrastructure and Security 🛡️</h3><p>The accompanying white paper, “<a href="https://apicontext.com/resources/api-readiness-for-agentic-ai-white-paper/">Enterprise API Readiness in the Era of Agentic AI</a>,” delves deeper into the infrastructural and security adaptations required. Key recommendations include:</p><ul><li><strong>Agent Gateway Adoption:</strong> Deploying Model Context Protocol (MCP) servers to abstract APIs, enforce policies, and manage authentication for agents.</li><li><strong>Agent-Aware Controls:</strong> Implementing nuanced rate limiting, concurrency caps, and dynamic throttling to handle the high-volume, parallel patterns of AI traffic.</li><li><strong>Enhanced Security:</strong> Moving to OAuth 2.1 to enable fine-grained access control and creating special scopes for AI agents based on the principle of least privilege.</li></ul><h3>From Theory to Practice with WSO2 💡</h3><p>Listening to the talk, I couldn’t help but connect the challenges of the “Agentic AI” era directly to the solutions we build at <a href="https://wso2.com/">WSO2</a>. The white paper’s recommendations aren’t just theoretical; they are practical problems that can be solved today. WSO2 is already addressing the need for agent-ready infrastructure with its <a href="https://wso2.com/api-manager/usecases/ai-gateway/"><strong>AI Gateway</strong></a> capabilities, which are designed to securely manage and govern this new wave of AI-driven traffic.</p><p>The key recommendations from the talk map directly to core WSO2 features:</p><ul><li><strong>Agent Gateway Adoption:</strong> The paper calls for deploying <strong>Model Context Protocol (MCP) servers</strong> to abstract APIs and enforce policies for agents. WSO2 is tackling this head-on with <a href="https://wso2.com/bijira/"><strong>Bijira</strong></a>, our AI-native API management platform, which provides a centralized control plane to create, manage, and discover <a href="https://wso2.com/library/blogs/expose-discover-and-manage-mcp-servers-with-bijira/">MCP servers</a> from existing APIs. This allows organizations to make their current services securely available to AI agents without starting from scratch.</li><li><strong>Agent-Aware Controls:</strong> The need for nuanced rate limiting to handle high-volume AI traffic is a perfect use case for the <strong>WSO2 AI Gateway’s traffic management</strong> features. It allows for granular rate limiting based not just on requests, but on token counts, providing the sophisticated control needed to protect backend services from the intense, parallel patterns of AI agents.</li><li><strong>Enhanced Security:</strong> The talk emphasizes security and the principle of least privilege. The <strong>WSO2 AI Gateway</strong> delivers this through a suite of <strong>AI Guardrails</strong>, including a Semantic Prompt Guard to block harmful inputs and the ability to mask sensitive PII. This, combined with WSO2’s robust support for <strong>OAuth 2.1</strong>, provides the multi-layered security essential for safely exposing APIs to autonomous agents.</li></ul><h3>Wrapping Up</h3><p>The message from API World was unequivocal 📣: the era of autonomous agents is upon us, and it places an unprecedented demand on the quality and accuracy of our API specifications. As the final slide simply stated, “If the spec drifts, the agent walks” 🚶‍♂️.</p><blockquote>Organizations that treat their OpenAPI definitions as critical, code-managed contracts will unlock powerful AI-driven automation <em>🚀</em>. Those who don’t will find their APIs and their businesses left behind <em>⏳</em>.</blockquote><p>The challenges of API drift and the demands of agentic AI can seem daunting, but the path forward is clear. This is where a comprehensive platform like WSO2 becomes critical. By leveraging an <strong>AI Gateway</strong>, organizations can implement the necessary guardrails, traffic management, and security policies to make their APIs truly “agent-ready,” bridging the gap between today’s infrastructure and the future of AI-driven automation.</p><p>Thanks for reading! 🙏 If you found this recap helpful, please share it with your network 🔗. I’d love to hear your thoughts in the comments 💬: what did you find most interesting? What topics would you like to see next? 📝For example, would a step-by-step guide on how to implement these patterns with the WSO2 AI Gateway be useful? 🛠️ Let me know! 👇</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4d943756a2e0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[WSO2 API Manager 4.5 on Kubernetes: A Complete Local Deployment Guide]]></title>
            <link>https://lianecafarate.medium.com/wso2-api-manager-4-5-on-kubernetes-a-complete-local-deployment-guide-18e03fffd38c?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/18e03fffd38c</guid>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[api-manager]]></category>
            <category><![CDATA[wso2-api-manager]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[wso2]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Mon, 25 Aug 2025 15:18:54 GMT</pubDate>
            <atom:updated>2025-08-26T16:35:58.880Z</atom:updated>
            <content:encoded><![CDATA[<h4>This guide focuses on deploying WSO2 APIM locally on Kubernetes for testing and learning purposes.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ZH59IQWriBUaeINZ" /><figcaption>Photo by <a href="https://unsplash.com/@growtika?utm_source=medium&amp;utm_medium=referral">Growtika</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Before we dive into the tutorial, here are some key concepts that will help you better understand the instructions. If you’re already familiar with them, feel free to skip ahead.</p><h3>1) Key Concepts</h3><h4>1.1. WSO2 API Manager (APIM)</h4><p><a href="https://wso2.com/api-manager/">WSO2 APIM</a> is a market-leading full lifecycle API management open-source platform for building, integrating, securing, and exposing an enterprise’s digital services as managed APIs in the cloud, on-premises, and in hybrid architectures.</p><h4>1.2. Kubernetes (K8s)</h4><p><a href="https://kubernetes.io/">Kubernetes</a>, also known as K8s, is an open-source system for automating deployments, scaling, and the management of containerized applications.</p><p><strong>1.2.1. Why should you use them?</strong></p><ul><li><a href="https://www.docker.com/resources/what-container/">Containers</a> are a good way to bundle and run your applications.</li><li>In a production environment, you need to manage the containers that run the applications and ensure that there is no downtime.</li><li>For example, if a container goes down, another container needs to start. That’s how K8s can help you, providing a framework to run distributed systems resiliently. It takes care of scaling and failover for your application, provides deployment patterns, and more.</li></ul><p><strong>1.2.2. Cluster Architecture</strong></p><p>A <a href="https://kubernetes.io/docs/concepts/architecture/">K8s cluster</a> consists of a control plane plus a set of worker machines, called <a href="https://kubernetes.io/docs/concepts/architecture/nodes/">nodes</a>, that run containerized applications.</p><blockquote>A node may be a virtual or physical machine, depending on the cluster.</blockquote><ul><li>Each node is managed by the <a href="https://kubernetes.io/docs/reference/glossary/?all=true#term-control-plane">control plane</a> and contains the services necessary to run <a href="https://kubernetes.io/docs/concepts/workloads/pods/">Pods</a>.</li><li>Every cluster needs at least one worker node to run <a href="https://kubernetes.io/docs/concepts/workloads/pods/">Pods</a>.</li></ul><blockquote>A pod is the smallest deployable unit in Kubernetes and represents a single instance of a running process in the cluster. It can contain one or more containers and shares the same network namespace and storage volume.</blockquote><p><strong>1.2.3. Helm</strong></p><p><a href="https://helm.sh/docs/glossary/#helm-and-helm">Helm</a> is the package manager for K8s. Basically, Helm makes it easy to install applications and resources into K8s clusters.</p><p><strong>1.2.4. Chart</strong></p><p>A <a href="https://helm.sh/docs/glossary/#chart">Chart</a> is a Helm package that contains information sufficient for installing a set of K8s resources into a K8s cluster.</p><ul><li>Charts contain a Chart.yaml file as well as templates, default values (values.yaml), and dependencies.</li></ul><p><strong>1.2.5. Release</strong></p><p>A <a href="https://helm.sh/docs/intro/using_helm/#three-big-concepts">Release</a> is an instance of a chart running in a Kubernetes cluster.</p><ul><li>One chart can often be installed many times into the same cluster.</li><li>Each time it is installed, a new release is created.</li></ul><p><strong>1.2.6. Namespace</strong></p><p>In K8s, <a href="https://kubernetes.io/docs/concepts/overview/working-with-objects/namespaces/"><em>namespaces</em></a> provide a mechanism for isolating groups of resources within a single cluster.</p><h3>2) Architecture Overview</h3><p>Before diving into the setup, it helps to understand the architecture we’re building. At a high level, the WSO2 API Manager deployment on Kubernetes looks like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JI3guT6wMooEaK1RN649pA.png" /><figcaption>High-level architecture</figcaption></figure><ul><li><strong>Client</strong>: This could be a browser, Postman, or any consumer trying to access your APIs.</li><li><strong>Ingress: </strong>Acts as the entry point to the Kubernetes cluster. It handles TLS termination, routing requests to the correct service, and load balancing. The TLS certificate we created is applied here, ensuring HTTPS communication.</li><li><strong>Service</strong>: Provides a stable internal endpoint that forwards traffic to the APIM Pod(s).</li><li><strong>Pod &amp; Node: </strong>The actual containerized API Manager runs inside a Pod on a Node within the Kubernetes cluster. This is where the APIM logic executes.</li></ul><blockquote>In summary, Clients send HTTPS requests → Ingress handles TLS and routing → Services forward traffic → Pods execute the API Manager logic.</blockquote><h3>⚠️ For Testing &amp; Demo Purposes Only</h3><blockquote><em>This configuration is intended strictly for local testing, experimentation, and demonstration. It is NOT secured, optimized, or configured for production environments.</em></blockquote><h3>3) Prerequisites</h3><p>Before deploying APIM on K8s, make sure you have the following installed:</p><h4><strong>3.1. Linux OS</strong></h4><p>For this setup, I’m using Ubuntu 24.04.2 LTS. While the official <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/reference/product-compatibility/">WSO2 APIM compatibility matrix</a> lists Ubuntu 22.04 as tested, the platform generally supports most modern Linux distributions. Since this is a testing environment, the minor version difference shouldn’t impact functionality.</p><p>You can verify your OS version with:</p><pre>lsb_release -a</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/348/1*NzBXH9JHaCEgqAPZoIb22g.png" /><figcaption>Terminal: Ubuntu version</figcaption></figure><p><strong>3.1.1. System Architecture</strong><br>Next, check your system architecture, as this will be important later during the deployment. Run:</p><pre>uname -m</pre><p>If the output is x86_64, your system is running a 64-bit x86 architecture.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/154/1*rOB6qyV8c6QumWsZ2BGmTw.png" /><figcaption>Terminal: Linux Architecture</figcaption></figure><h4>3.2. Git</h4><p><a href="https://git-scm.com/">Git</a> is an open-source, distributed version control system. For this setup, I’m using <strong>version </strong>2.50.1.</p><p>Install Git on Ubuntu with:</p><pre>sudo apt install git-all💡 Note: It’s always a good idea to cross-check with the official installation guide for the latest instructions.</pre><p>Verify the installation:</p><pre>git --version</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/200/1*pCHu8w9LvNAk8UhMsf1zag.png" /><figcaption>Terminal: Git Version</figcaption></figure><blockquote><em>💡 </em>It’s always a good idea to cross-check with the <a href="https://git-scm.com/book/en/v2/Getting-Started-Installing-Git">official installation guide</a> for the latest instructions.</blockquote><h4>3.3. Helm</h4><p><a href="https://helm.sh/docs/glossary/#helm-and-helm">Helm</a> is the package manager for K8s (mentioned above 😉), enabling us to easily deploy and manage the WSO2 APIM charts. This guide uses Helm version 3.18.4.</p><p>Although the <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/kubernetes-deployment/kubernetes/kubernetes-overview/#helm">official compatibility matrix</a> lists Helm versions <strong>3.10–3.16.1</strong>, using 3.18.4 should not cause any issues for this testing environment, similar to how we discussed OS versions earlier.</p><p>Install Helm on Ubuntu with:</p><pre>curl https://baltocdn.com/helm/signing.asc | gpg --dearmor | sudo tee /usr/share/keyrings/helm.gpg &gt; /dev/null<br>sudo apt-get install apt-transport-https --yes<br>echo &quot;deb [arch=$(dpkg --print-architecture) signed-by=/usr/share/keyrings/helm.gpg] https://baltocdn.com/helm/stable/debian/ all main&quot; | sudo tee /etc/apt/sources.list.d/helm-stable-debian.list<br>sudo apt-get update<br>sudo apt-get install helm</pre><p>Verify the installation:</p><pre>helm version</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0h7Mfl2qo0Qzddy5TZ504Q.png" /><figcaption>Terminal: Helm version</figcaption></figure><blockquote><em>💡 </em>Always refer to the <a href="https://helm.sh/docs/intro/install/#from-apt-debianubuntu">official Helm installation guide</a> for the most current installation methods and troubleshooting.</blockquote><h4>3.4. Kubernetes Client (kubectl)</h4><p>kubectl is the <a href="https://kubernetes.io/docs/tasks/tools/">Kubernetes command-line tool</a> that allows you to interact with your cluster. You can use it to deploy applications, inspect and manage cluster resources, and view logs.</p><p>Since we already confirmed our <strong>system architecture</strong> earlier, that information helps us select the correct binary for installation. For the latest and most reliable installation steps, follow the <a href="https://kubernetes.io/docs/tasks/tools/install-kubectl-linux/#install-kubectl-binary-with-curl-on-linux">official documentation</a> and choose the instructions that match your system architecture.</p><p>After installation, you can verify that kubectl is working with:</p><pre>kubectl version --client</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/336/1*glyykQu_fd34-N5w7PLuiQ.png" /><figcaption>Terminal: kubectl version</figcaption></figure><h4>3.5. Docker</h4><p>If you’re not very familiar with Docker, you can check out my previous article on <a href="https://lianecafarate.medium.com/deploying-wso2-api-manager-4-5-with-docker-319c86303b4b">deploying WSO2 API Manager 4.5 using Docker</a>.</p><p>For this setup, you’ll need Docker (or a similar container runtime). Docker is an open platform for developing, shipping, and running applications inside containers, as mentioned earlier 😉.</p><blockquote><em>💡</em> <strong>Note:</strong> If you’re using <a href="https://rancherdesktop.io/">Rancher Desktop</a>, it already comes bundled with Docker, <em>kubectl</em>, <em>helm</em>, and several other essential tools.</blockquote><p>For this guide, I’m using Docker version28.3.2. To install Docker on Ubuntu, follow the official instructions: <a href="https://docs.docker.com/engine/install/ubuntu/#install-using-the-repository">Install Docker Engine on Ubuntu</a></p><p>After installation, verify that Docker is working correctly:</p><pre>docker version</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*Y_uVWdl2xwNVfZ9vVuYePQ.png" /><figcaption>Terminal: docker version</figcaption></figure><h4>3.6. Minikube (K8s Cluster)</h4><p><a href="https://minikube.sigs.k8s.io/docs/">Minikube</a> is a tool that quickly sets up a local Kubernetes cluster on macOS, Linux, and Windows. It’s ideal for development, testing, and learning Kubernetes without needing a cloud provider.</p><p>For this setup, I’m using Minikube<strong> </strong>1.36.0. To install Minikube, follow the <a href="https://minikube.sigs.k8s.io/docs/start/?arch=%2Flinux%2Fx86-64%2Fstable%2Fdebian+package">official installation guide</a>. I selected the following options to match my environment:</p><ul><li><strong>Operating System:</strong> Linux</li><li><strong>Architecture:</strong> x86_64</li><li><strong>Release Type:</strong> Stable</li><li><strong>Installer Package:</strong> Debian</li></ul><p>After installing Minikube, make sure it’s correctly installed by checking the version:</p><pre>minikube version</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/561/1*K5UZmJe1Re5v9kIejqEHOA.png" /><figcaption>Terminal: Minikube version</figcaption></figure><p>Once installation is verified, start your local Kubernetes cluster with:</p><pre>minikube start</pre><p>To increase the allocated resources, use the minikube start command with the --memory and --cpus flags. Adjust the values to suit your needs.</p><pre>minikube start --memory=8192 --cpus=4</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/687/1*8a_kpmkSgRNg5-sxs6ytYw.png" /><figcaption>Terminal: Minikube start</figcaption></figure><p>This initializes the cluster and configures kubectl to communicate with it.</p><h4>3.7. NGINX Ingress Controller</h4><p>In K8s, <a href="https://www.solo.io/topics/nginx/kubernetes-ingress-controller">Ingress</a> is a collection of rules that allow inbound connections to reach the cluster services.</p><p>The NGINX Ingress Controller is a software component that runs inside a K8s cluster and manages traffic coming into the cluster (inbound traffic).</p><p>It acts as a reverse proxy, routing traffic from external clients to the appropriate service within the cluster based on the Ingress rules configured in the cluster.</p><p>In short, it serves as the <strong>traffic manager</strong> for applications running in K8s.</p><p>In Minikube, enabling the NGINX Ingress Controller is simple using the built-in addon:</p><pre>minikube addons enable ingress</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wLpoFWX5rdsLyiacjHNhdA.png" /><figcaption>Terminal: Minikube Ingress Addon</figcaption></figure><p>To confirm that the Ingress Controller is running, check the pods in the ingress-nginx namespace:</p><pre>kubectl get pods --namespace=ingress-nginx</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/555/1*vi2_MbjaI343Xp76sS8fOQ.png" /><figcaption>Terminal: NGINX pods</figcaption></figure><h3>4) Deploying WSO2 APIM 4.5 on K8s Using Helm</h3><p>I know, that was quite a few prerequisites! 😅 But before we jump into the fun part — deploying APIM — I want to quickly explain an important concept: <strong>deployment patterns</strong>. Hang in there just a little longer, it’ll make the next steps much clearer!</p><h4>4.1. Deployment Patterns</h4><p>WSO2 APIM offers a flexible and componentized deployment architecture that supports diverse scalability, availability, and security requirements. The platform can be deployed in various patterns — from simple, all-in-one setups to fully distributed, production-grade environments — adapting seamlessly to organizational needs.</p><p>For this setup, we&#39;ll be using <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/kubernetes-deployment/kubernetes/am-pattern-0-all-in-one/">Pattern 0: APIM Deployment with All-in-One Setup</a>, which is ideal for introductory tutorials.</p><p><a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/single-node/all-in-one-deployment-overview/">All-in-One Deployment</a> means that all APIM components run in a single runtime instance. This approach is easy to set up, requires fewer resources, and is perfect for learning, testing, or small-scale environments, compared to a fully distributed deployment.</p><blockquote><em>✨</em> Want to learn more about APIM architecture and its components (Traffic Manager<em>🚦</em>, Key Manager<em>🔑</em>, Gateway <em>🌐</em>, etc.)? Check the <a href="https://apim.docs.wso2.com/en/latest/get-started/apim-architecture/">official documentation</a>. If you’d like a more detailed article on this, let me know in the comments <em>💬</em>!</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QX4vJmVc2z6NxIRbGqt8xQ.png" /><figcaption>All-in-one Single Node Deployment Pattern</figcaption></figure><p>Alright, we’re back on track! 😎 Now that deployment patterns are clear, it’s time to <strong>dive into the real fun: </strong>configuring the resources and getting WSO2 APIM up and running on K8s.</p><h4>4.2. Resource Requirements</h4><p>The <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/kubernetes-deployment/kubernetes/kubernetes-overview/#resource-requirements">official documentation</a> recommends that a single-node all-in-one deployment should have at least:</p><ul><li><strong>CPU:</strong> 2 cores</li><li><strong>Memory:</strong> 4 GB</li></ul><p>This should ensure that all APIM components run smoothly without resource contention.</p><h4>4.3. Add the WSO2 Helm Chart Repository</h4><p>Before deploying APIM on K8s, you need to add the WSO2 Helm Chart repository. Helm charts package all the K8s resources required for deploying WSO2 products. Adding the repository allows you to pull the latest charts and deploy them easily using Helm.</p><pre>helm repo add wso2 https://helm.wso2.com &amp;&amp; helm repo update</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/803/1*Gl6MVUjHsBYdpfJjir_2AA.png" /><figcaption>Terminal: Adding WSO2 helm chart repository</figcaption></figure><h4>4.4. Minimal Configuration</h4><p>To quickly try out WSO2 API Manager on Kubernetes, we’ll use the <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/kubernetes-deployment/kubernetes/am-pattern-0-all-in-one/#minimal-configuration">minimal configuration</a>, which is defined in the default_values.yaml file provided with the Helm chart.</p><p>This configuration includes:</p><ul><li>H2 database 🗄️ (embedded)</li><li>Default keystore and truststore 🔑</li><li>Basic settings ⚙️ suitable for testing and learning purposes</li></ul><blockquote><em>💡</em> <strong>Note:</strong> This configuration is meant for <strong>local testing and experimentation</strong>. For production use, you should deploy a proper external database, configure secure keystores, and tune the environment for performance and security.</blockquote><h4>4.5. APIM Deployment</h4><p>Now, let&#39;s deploy WSO2 APIM on K8s with minimal configuration using Helm. Run the following command:</p><pre>helm install apim wso2/wso2am-all-in-one --version 4.5.0-3 --namespace wso2 --create-namespace -f https://raw.githubusercontent.com/wso2/helm-apim/main/docs/am-pattern-0-all-in-one/default_values.yaml   </pre><ul><li>Install the all-in-one APIM chart (wso2am-all-in-one)</li><li>Deploy version 4.5.0–3</li><li>Create a new namespace called wso2</li><li>Apply the default minimal configuration from the Helm chart’s default_values.yaml</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UoEjM3hnEoRdbbOiXFbPqw.png" /><figcaption>Terminal: APIM Deployment</figcaption></figure><blockquote><em>💡</em> As I mentioned earlier, namespaces are a great way to keep groups of resources separated within a single Kubernetes cluster. That&#39;s why we created a dedicated space called wso2 where all the WSO2 API Manager components will live.</blockquote><h4>4.6. View Pods in the WSO2 Namespace</h4><p>Before interacting with your deployed APIM components, it’s important to check the status of your pods. This helps ensure that all components are running correctly and are ready to handle traffic. Use the following command to list all pods in the wso2 namespace:</p><pre>kubectl get pods -n wso2</pre><ul><li>Lists all <strong>pods</strong> running in the wso2 namespace.</li><li>Helps you check the status of each pod (e.g., Running, Pending, or CrashLoopBackOff).</li><li>Useful to verify that all APIM components are deployed successfully.</li></ul><blockquote><strong>Note:</strong> If the pod isn’t READY yet, don’t worry. It can take some time for all components to fully start. Additionally, a few adjustments may be needed before you can access the API Manager.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/716/1*1RAHLlj7Eo4GuovkO2JOFg.png" /><figcaption>Terminal: ContainerCreating status</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/647/1*AicxFy9Pn1Jm94n5E6iAWA.png" /><figcaption>Terminal: Running status</figcaption></figure><p>You can check WSO2 APIM logs using the following command:</p><pre>kubectl logs -n wso2 &lt;pod-name&gt;</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MJ3TaW8KRxoO977zl4xYKw.png" /><figcaption>Terminal: WSO2 APIM logs</figcaption></figure><h4>4.7. View Services in the WSO2 Namespace</h4><p>Let&#39;s also check the services running in the cluster. Services provide stable network endpoints for pods and expose the APIs inside your WSO2 namespace. Use the following command to list all services in the wso2 namespace:</p><pre>kubectl get svc -n wso2</pre><ul><li>Lists all <strong>services</strong> in the wso2 namespace.</li><li>Services expose your pods internally within the cluster or externally to the outside world.</li><li>This lets you see the cluster IPs, ports, and types (ClusterIP, NodePort, LoadBalancer) for APIM components.</li></ul><p>These are the internal services that expose the API Manager’s ports inside the cluster. Since the TYPE is ClusterIP, these services are not directly accessible from outside your cluster. We’ll configure access next using an Ingress.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gKPFsSQNREA6va_xL1KUwg.png" /><figcaption>Terminal: Services in the wso2 namespace</figcaption></figure><h4>4.8. Obtain the External IP</h4><p>Now that the WSO2 API Manager is deployed, let’s find out how to access it externally. We can do this by checking the K8s Ingress resources, which will show us the external IP address being used.</p><p>Run the following command:</p><pre>kubectl get ing -n wso2</pre><ul><li>Lists all Ingress resources in the wso2 namespace.</li><li>Ingress resources define how external traffic is routed to the services inside the cluster, typically via the NGINX Ingress Controller.</li><li>This helps you confirm that your APIM endpoints are properly exposed and ready to receive traffic.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_5u6kAXXQ4Ttpj22gS-7cQ.png" /><figcaption>Terminal: External IPs</figcaption></figure><p>In the ADDRESS column, you&#39;ll find the external IP. In this example, it’s 192.168.49.2. This is the IP you’ll use to access the API Manager through your browser (after setting up your /etc/hosts, which we’ll cover next).</p><p>If the ADDRESS is blank or &lt;pending&gt;, give it a moment, it might still be initializing.</p><h4>4.9. Update Your /etc/hosts File</h4><p>To access the WSO2 API Manager components through your browser using friendly domain names (like am.wso2.com or gw.wso2.com), you need to map those hostnames to the external IP address you just retrieved from the Ingress.</p><p>This is done by adding entries to your local machine’s /etc/hosts file.</p><p>Replace &lt;EXTERNAL-IP&gt; with the actual IP you saw in the kubectl get ing output (for example, 192.168.5.15):</p><pre>&lt;EXTERNAL-IP&gt; am.wso2.com<br>&lt;EXTERNAL-IP&gt; gw.wso2.com<br>&lt;EXTERNAL-IP&gt; websocket.wso2.com<br>&lt;EXTERNAL-IP&gt; websub.wso2.com</pre><p>For example, if your external IP was 192.168.49.2, your entries would look like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aHuqfyQjH22NUinOjvTPXw.png" /><figcaption>Terminal: /etc/hosts file</figcaption></figure><p>To edit the /etc/hosts file, open it with a text editor using sudo (since it requires admin privileges):</p><pre>sudo vim /etc/hosts</pre><blockquote>Don’t forget to save the file! If you’re using vim, press Esc, then type :wq! and hit Enter to write the changes and quit the editor.</blockquote><p>You can also use a different editor like nano if you prefer.</p><h4>4.10. Generate a Self-Signed TLS Certificate</h4><blockquote>🔎 Since we’re just testing locally, we can use a <strong>self-signed certificate</strong> to enable HTTPS without needing a certificate from a trusted authority.</blockquote><p>We’ll use <strong>OpenSSL</strong> to create both a self-signed certificate and a private key in one simple command.</p><p>Just run this in your terminal:</p><pre>openssl req -x509 -nodes -days 365 -newkey rsa:2048 \<br>  -keyout wso2.key -out wso2.crt \<br>  -subj &quot;/CN=am.wso2.com&quot; \<br>  -extensions v3_req \<br>  -config &lt;(cat &lt;&lt;EOF<br>[req]<br>distinguished_name = req<br>req_extensions = v3_req<br>[req_distinguished_name]<br>[v3_req]<br>keyUsage = digitalSignature, keyEncipherment<br>extendedKeyUsage = serverAuth<br>subjectAltName = DNS:am.wso2.com,DNS:gw.wso2.com,DNS:websocket.wso2.com,DNS:websub.wso2.com<br>EOF<br>)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Vrnrk_bBM7iX0SS4-hQ9Bw.png" /><figcaption>Terminal: Generating a self-signed TLS certificate</figcaption></figure><p>Here’s what each part of the command means:</p><ul><li>req -x509: Generate a self-signed certificate instead of a certificate signing request (CSR).</li><li>-nodes: Create an unencrypted private key (no passphrase needed).</li><li>-days 365: Make the certificate valid for 365 days.</li><li>-newkey rsa:2048: Generate a new 2048-bit RSA private key.</li><li>-keyout wso2.key: Save the private key to the file wso2.key.</li><li>-out wso2.crt: Save the certificate to the file wso2.crt.</li><li>-subj &quot;/CN=am.wso2.com&quot;: Set the certificate&#39;s Common Name (CN) to am.wso2.com. Replace this with the hostname you&#39;ll use to access the API Manager.</li><li>-extensions v3_req: Apply the v3_req section from the config for additional certificate options.</li><li>-config &lt;(...): Provide a temporary configuration inline for the certificate, defining extensions such as:</li><li>keyUsage = digitalSignature, keyEncipherment → Allows the key to be used for signing and encryption.</li><li>extendedKeyUsage = serverAuth → Indicates the certificate is for server authentication.</li><li>subjectAltName = DNS:am.wso2.com,DNS:gw.wso2.com,DNS:websocket.wso2.com,DNS:websub.wso2.com → Specifies all hostnames that this certificate will cover.</li></ul><p>This command will create two files:</p><ul><li>wso2.crt: Your self-signed certificate.</li><li>wso2.key: Your private key.</li></ul><p>You’ll use these files to configure TLS for your local APIM, enabling secure HTTPS communication for the hostnames listed in subjectAltName.</p><blockquote><em>💡</em> <strong>Note:</strong> For production deployments, always use a certificate issued by a trusted CA instead of a self-signed certificate.</blockquote><p><strong>(Optional)</strong> The following command inspects your TLS certificate (wso2.crt) and specifically extracts the Subject Alternative Name (SAN) extension, which lists all the hostnames or domains that the certificate is valid for.</p><pre>openssl x509 -in wso2.crt -noout -text | grep -A1 &quot;Subject Alternative Name&quot;</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/680/1*ocreYZNCpqtJlS1uOrbCFA.png" /><figcaption>Terminal: Inspecting the TLS certificate</figcaption></figure><h4>4.11. Enable HTTPS with a K8s TLS Secret</h4><p>Now that you have your self-signed certificate (wso2.crt) and private key (wso2.key), the next step is to make them available to Kubernetes so the NGINX Ingress Controller can use them for <strong>TLS termination. </strong>This means it will handle incoming HTTPS traffic securely on behalf of your WSO2 API Manager.</p><p>To do this, you need to create a <strong>Kubernetes TLS secret</strong> that stores your certificate and key. Run the following command, making sure to replace wso2.key and wso2.crt with your actual key and certificate files, and specify the correct namespace (wso2 in this example):</p><pre>kubectl create secret tls wso2-tls-secret --key wso2.key --cert wso2.crt -n wso2</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/952/1*21a08ak4OI0zcPN3ycWx3w.png" /><figcaption>Terminal: TLS secret created</figcaption></figure><p>Run the following command to restart the NGINX Ingress controller so it picks up the new TLS certificate:</p><pre>kubectl rollout restart deployment/ingress-nginx-controller -n ingress-nginx</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/672/1*1K83iamf2VZMDnQdspOerg.png" /><figcaption>Terminal: NGINX restart initiated</figcaption></figure><p>Next, wait for the rollout to complete and verify that the restart was successful by running:</p><pre>kubectl rollout status deployment/ingress-nginx-controller -n ingress-nginx</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/638/1*UhCY1a4mswnBYswYLTaiyQ.png" /><figcaption>Terminal: NGINX restart completed</figcaption></figure><p>This ensures that the Ingress controller is fully updated and ready to serve traffic using your new certificate.</p><h4>4.12. Verify Kubernetes Secrets</h4><p>After creating secrets, it’s a good idea to check that they exist and are correctly configured. Use the following command to list all secrets in the wso2 namespace:</p><pre>kubectl get secret -n wso2</pre><p>You should see your TLS secret (e.g., wso2-tls-secret) listed among the cluster secrets. This ensures that your Ingress and APIM components can reference the secret properly for HTTPS traffic.</p><p>💡 <strong>Tip:</strong> If you don’t see the secret, double-check the creation command or namespace. Secrets are <strong>namespace-scoped</strong>, so creating them in the wrong namespace will prevent APIM from accessing them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/433/1*aub_WNYqqXGoSjqO8KPWfQ.png" /><figcaption>Terminal: Getting secrets</figcaption></figure><h4>4.13. Inspect and Update the Ingress Resource</h4><p>After deploying WSO2 API Manager using Helm, the Helm chart automatically creates an <strong>Ingress</strong> resource to expose the service.</p><p><strong>(Optional)</strong> Run the following command to view the current Ingress configuration:</p><pre>kubectl get ingress apim-wso2am-all-in-one-am-ingress -n wso2 -o yaml</pre><p>You should see an output similar to:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/859/1*Wj2mxdqwvi9R0mDCEMBAjg.png" /><figcaption>Terminal: Ingress YAML file</figcaption></figure><p>This configuration:</p><ul><li>Routes all traffic for https://am.wso2.com to the internal WSO2 service on port 9443.</li><li>Uses the NGINX ingress controller.</li><li>Declares TLS, but <strong>does not yet include the certificate secret</strong> (secretName).</li></ul><p>To complete the TLS setup, you need to attach the TLS secret (created in a previous step) to this Ingress. Use the following command:</p><pre>kubectl edit ingress apim-wso2am-all-in-one-am-ingress -n wso2</pre><p>Find the tls section and update it like this:</p><pre>tls:<br>  - hosts:<br>      - am.wso2.com<br>    secretName: wso2-tls-secret</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/818/1*SLiqDCZs0gQsZGAxdsAN-Q.png" /><figcaption>Terminal: Updating the Ingress YAML file</figcaption></figure><p>Without specifying the secretName, the Ingress knows TLS is expected, but cannot serve HTTPS traffic correctly. Adding the secret ensures encrypted traffic is handled properly and lets you securely access the Publisher, Dev Portal, and Admin Portal via <a href="https://am.wso2.com./">https://am.wso2.com</a>.</p><h4>4.14. Adjust Configurations for default_values.yaml</h4><p>The WSO2 APIM Helm chart comes with a default configuration file (default_values.yaml) that defines how your deployment will behave. While the default settings are suitable for quick testing, customizing this file lets you tailor the deployment to your specific environment and needs, such as enabling HTTPS, configuring OAuth URLs, or specifying ingress settings.<br>Here’s how to adjust key configurations in the default_values.yaml for your local K8s setup:</p><p><strong>4.14.1. Download the Default Values File</strong></p><p>Start by downloading the <a href="https://github.com/wso2/helm-apim/blob/main/docs/am-pattern-0-all-in-one/default_values.yaml">official default values file</a> locally so you can edit it.</p><p><strong>4.14.2. Open the File for Editing</strong></p><p>Use your preferred text editor to open the file:</p><pre>vim default_values.yaml</pre><p><strong>4.14.3. Set the Ingress Class and TLS Secret</strong></p><p>Locate the kubernetes section and update the ingress-related settings to enable HTTPS with your TLS secret:</p><pre>kubernetes:<br>  ingressClass: &quot;nginx&quot;            # Use the NGINX ingress controller<br>  ingress:<br>    tlsSecret: &quot;wso2-tls-secret&quot;  # Name of the Kubernetes TLS secret for HTTPS</pre><p>This configuration ensures that the Ingress resource uses the NGINX controller and references the correct TLS certificate for secure HTTPS connections.</p><p><strong>4.14.4. Configure the JWKS URL</strong></p><p>If you’re using a hostname like am.wso2.com that is not globally routable (common in local testing), update the OAuth JWKS URL to point to localhost:</p><pre>wso2:<br>  apim:<br>    configurations:<br>      oauth_config:<br>        oauth2JWKSUrl: &quot;https://localhost:9443/oauth2/jwks&quot;</pre><p>This adjustment prevents authentication issues related to JWKS URL misconfiguration.</p><p><strong>4.14.5. Configure API Manager Resources</strong></p><p>Update the deployment resource settings to allocate sufficient memory and CPU for all WSO2 API Manager profiles:</p><pre>deployment:   <br>  ...<br>  resources:<br>      # These are the resource recommendations for running WSO2 API Management product profiles with profile optimization<br>      # Resource configurations defined here are applicable for all API Manager product profiles of this deployment<br>      requests:<br>        # -- Memory request for API Manager<br>        memory: &quot;4Gi&quot;<br>        # -- CPU request for API Manager<br>        cpu: &quot;2000m&quot;<br>      limits:<br>        # -- Memory limit for API Manager<br>        memory: &quot;6Gi&quot;<br>        # -- CPU limit for API Manager<br>        cpu: &quot;3000m&quot;</pre><p>These values ensure:</p><ul><li>The API Manager has enough resources to start and run reliably.</li><li>Memory and CPU limits prevent the container from consuming more than your available system resources.</li><li>Requests and limits apply to all API Manager product profiles in this deployment.</li></ul><p><strong>4.14.6. Save Your Changes</strong></p><p>After making these edits, save and close the file.</p><p><strong>4.14.7. Deploy or Upgrade Your Helm Release Using the Updated File</strong></p><p>Use your customized default_values.yaml when deploying or upgrading WSO2 API Manager:</p><pre>helm upgrade --install apim wso2/wso2am-all-in-one \<br>  --version 4.5.0-3 \<br>  --namespace wso2 \<br>  -f default_values.yaml</pre><ul><li>This command installs the release if it doesn’t exist, or upgrades it if it does.</li><li>Be patient! It may take a few minutes for all pods and components to start and become ready.</li></ul><p>You can monitor progress with:</p><pre>kubectl get pods -n wso2</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/656/1*u7lAaa8esoUc3k-6Nb67sg.png" /><figcaption>Terminal: Ready status</figcaption></figure><p><strong>4.14.8. Access the APIM</strong></p><p>Open your browser and navigate to one of the API Manager portals, for example:</p><pre>https://am.wso2.com/publisher</pre><blockquote><em>⚠️ You may see a warning like </em>“Your connection is not private”<em>. This happens because the certificate is self-signed and not trusted by your browser by default.</em></blockquote><p>Simply proceed to accept or trust the certificate for local testing. Once accepted, you should be able to access the Publisher and DevPortal consoles securely.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gzn9PUXBHMpJXa0ivGJ5Tw.png" /><figcaption>Browser: Connection not private</figcaption></figure><p>You should now see the login page. Use the default credentials:</p><ul><li>Username: admin</li><li>Password: admin</li></ul><p>Then click <strong>Continue</strong> to access the portal.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/719/1*6_2K8ti1PIvJZZbt-CFgGw.png" /><figcaption>Browser: APIM login</figcaption></figure><p>Voilà! You’re now ready to start using WSO2 API Manager. For guidance on creating your first API, check out this <a href="https://apim.docs.wso2.com/en/latest/manage-apis/design/create-api/create-rest-api/create-a-rest-api/">step-by-step guide</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dNKRGMXHhgNXORDTZRMz_g.png" /><figcaption>Browser: APIM Publisher portal</figcaption></figure><h3>🎬 <strong>Mission Accomplished</strong></h3><p>🙌 Congratulations! You’ve made it through what was definitely <strong>not an easy WSO2 API Manager + Kubernetes setup</strong>! 💪 It took multiple rounds of debugging and hunting for the right instructions to get everything working, so if you made it here, congratulations!</p><p>If this guide helped you navigate the twists and turns of pods, services, secrets, and ingress TLS, a little clap 👏 would mean a lot. It motivates me to tackle more tricky setups like this. Got questions, insights, or your own Kubernetes war stories? Drop them in the comments. I’d love to hear from you! 👂</p><p>Feel free to share 🔁 this guide with your fellow engineers. It might save them some headaches.</p><p>Here’s to successful deployments, smooth clusters, and happy coding in the cloud! ☸️✨</p><h3>📚 References</h3><p><a href="https://wso2.com/whitepapers/a-reference-architecture-for-deploying-wso2-middleware-on-kubernetes/">A Reference Architecture for Deploying WSO2 Middleware on Kubernetes</a></p><p><a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/kubernetes-deployment/kubernetes/am-pattern-0-all-in-one/#minimal-configuration">WSO2 APIM 4.5: Deploy on Kubernetes — Pattern 0: API-M Deployment with All-in-One Setup</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=18e03fffd38c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[No Message Left Behind: Transactional IBM MQ → Database Integration with WSO2 Ballerina]]></title>
            <link>https://lianecafarate.medium.com/no-message-left-behind-transactional-ibm-mq-database-integration-with-wso2-ballerina-6e6847a74d70?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/6e6847a74d70</guid>
            <category><![CDATA[ibm-mq]]></category>
            <category><![CDATA[mysql]]></category>
            <category><![CDATA[integration]]></category>
            <category><![CDATA[reliability]]></category>
            <category><![CDATA[wso2-ballerina]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Thu, 14 Aug 2025 14:23:01 GMT</pubDate>
            <atom:updated>2025-08-14T14:23:01.226Z</atom:updated>
            <content:encoded><![CDATA[<h4>Guaranteeing reliable data flow with IBM MQ, MySQL, and Ballerina’s transactional magic</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*xIne6UZJ_noBuiE3" /><figcaption>Photo by <a href="https://unsplash.com/@ikukevk?utm_source=medium&amp;utm_medium=referral">Kevin Ku</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>This article provides a step-by-step guide on configuring WSO2 Ballerina to integrate with IBM MQ, focusing on a transactional use case. We’ll explore how Ballerina can consume messages from an IBM MQ queue, process them, and then commit or roll back the transaction based on the outcome of a downstream operation, such as a database write.</p><h3>🩰What is Ballerina?</h3><p><a href="https://ballerina.io/">Ballerina</a> is an open-source, cloud-native programming language developed by <a href="https://wso2.com/">WSO2</a>. It’s built with the idea that modern applications need to talk to multiple systems, APIs, databases, messaging queues, SaaS services, and that this should be as easy as writing business logic.</p><p>Key features that make it a great fit for this project:</p><ul><li><strong>Service-oriented syntax</strong>: We can define network services (HTTP, WebSocket, JMS, gRPC, etc.) natively in the language.</li><li><strong>Connectors</strong>: prebuilt modules for systems like IBM MQ, MySQL, Kafka, Salesforce, and more.</li><li><strong>Built-in concurrency and transactions</strong>: manage message acknowledgments and database commits in the same code block.</li></ul><h3>🚶‍♂️🚶‍♂️🚶‍♂️What is IBM MQ?</h3><p><a href="https://www.ibm.com/products/mq">IBM MQ</a> is a robust and reliable message-oriented middleware that enables applications to communicate and exchange data securely across different platforms.</p><p>IBM MQ uses a simple model:</p><ul><li><strong>Messages:</strong> The packages of data that applications send to each other.</li><li><strong>Queues:</strong> The storage locations where messages are reliably held until a receiving application is ready to process them.</li><li><strong>Queue Managers:</strong> The core servers that manage the queues and the flow of messages.</li></ul><p>The key benefit of this approach is asynchronous messaging. It allows applications to send and receive data independently, meaning they don’t have to be available at the same time.</p><h3><strong>⛰️</strong>The Challenge: Ensuring Transactional Reliability</h3><p>Our use case involves processing critical appointment data from an IBM MQ queue. The messages are consumed by a Ballerina service and then persisted to a database. The key requirement is data integrity; if the database write fails, the message should not be acknowledged back to IBM MQ. This ensures the message remains in the queue to be re-processed later, preventing data loss.</p><ul><li><strong>Source</strong>: IBM MQ (APPOINTMENT.QUEUE) with critical appointment messages</li><li><strong>Processor</strong>: WSO2 Integrator (Ballerina service)</li><li><strong>Destination:</strong> Database persistence (MySQL)</li><li><strong>Transactional Model:</strong> Messages are acknowledged only after a successful database write. If the DB is down, MQ keeps the message for retry.</li></ul><h3>❓Why Ballerina and IBM MQ for Our Use Case?</h3><p>The combination of Ballerina and IBM MQ is a powerful solution for our transactional use case because they are designed to complement each other.</p><ul><li><strong>IBM MQ</strong> acts as a reliable buffer. It ensures message persistence, so even if the Ballerina service or the downstream database is temporarily unavailable, the appointment data is safely stored in the queue. This prevents data loss and decouples the systems, allowing them to operate independently.</li><li><strong>Ballerina</strong> then leverages its native transactional capabilities. Its transaction block ensures that the message is only removed from the IBM MQ queue <em>after</em> the database write has successfully completed. If the database operation fails for any reason, the transaction is automatically rolled back. This means the message is not acknowledged and remains in the queue, ready for a subsequent attempt.</li></ul><p>This symbiotic relationship guarantees end-to-end data integrity. IBM MQ handles the reliable queuing, while Ballerina provides the atomic processing logic.</p><h3>📂 Architecture Overview</h3><p>Our solution follows this comprehensive flow:</p><ol><li><strong>IBM MQ</strong> receives and queues critical appointment messages on APPOINTMENT.QUEUE</li><li><strong>Ballerina IBM Connector Listener </strong>consumes messages transactionally from the queue</li><li><strong>Appointment Processing Logic</strong> validates and transforms the appointment data</li><li><strong>MySQL Database</strong> persists structured appointment records across three tables</li><li><strong>Transaction Management</strong> ensures atomicity across the entire operation</li></ol><p>The system uses a database schema with three main entities:</p><ul><li><strong>Patient Table</strong>: Stores patient details (name, email, phone)</li><li><strong>Doctor Table</strong>: Contains doctor information (name, specialization)</li><li><strong>Appointment Table</strong>: Links patients and doctors with appointment details (time, hospital, status, notes)</li></ul><p>If the database becomes unavailable, messages remain unacknowledged in IBM MQ, ensuring zero data loss.</p><h3>⚠️ For Testing &amp; Demo Purposes Only</h3><blockquote><em>This configuration is intended strictly for local testing, experimentation, and demonstration. It is NOT secured, optimized, or configured for production environments.</em></blockquote><h3>🛠 Prerequisites</h3><ol><li><strong>IBM MQ</strong> (v9.4.1.0)</li><li><strong>MySQL </strong>(v8.4.0)</li><li><a href="https://marketplace.visualstudio.com/items?itemName=WSO2.ballerina-integrator"><strong>VS Code Extension — WSO2 Integrator BI</strong></a> (v2201.12.3)</li><li><a href="https://rancherdesktop.io/">Rancher Desktop</a> (v1.19.3) for containers</li><li>macOS Sequoia — Apple M1 Pro Arm64 (v15.5)</li></ol><blockquote><strong><em>▶️ Ready to dive in? Let’s get started!</em></strong></blockquote><h3>1) Setting up IBM MQ</h3><h4>1.1) Building IBM MQ Image</h4><p>If you’re on macOS with an Arm64 (Apple Silicon) processor, you might notice that IBM doesn’t provide an official Docker image for this architecture. The good news: you can build it yourself using IBM’s official MQ container repository <strong>[1]</strong>.</p><p>Follow these steps:</p><pre>git clone https://github.com/ibm-messaging/mq-container.git<br>cd mq-container<br>MQ_VERSION=9.4.1.0 make build-devserver COMMAND=docker</pre><p>The Makefile will check the system environment to determine if either the podman or docker command is available. Where both are available, then the default will be to build the container image using podman.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*FUDVw7xrReqin1MtPF5LkA.png" /><figcaption>Terminal: Building image</figcaption></figure><h4>1.1a) <strong>[Optional] Using Rancher Desktop with Docker</strong></h4><p>If you’re using Rancher Desktop and forget to specify the Docker context during the build (like I did 😅), the image might be built for Podman instead. To use this image with Docker, you can easily convert and load it with a simple command.</p><p>Use the following command to save the Podman image and load it into your Docker daemon:</p><pre>podman save ibm-mqadvanced-server-dev:9.4.1.0-arm64 | docker load</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/703/1*0knhd9MePtDHo9bE3wKycg.png" /><figcaption>Terminal Loading image into Docker</figcaption></figure><p>This ensures the image is available and ready to use within your Docker environment, regardless of how it was initially built.</p><h4>1.2) Running IBM MQ Image</h4><p>After building the Docker image, run the following command to start a new container. This command both creates and configures a running IBM MQ queue manager.</p><pre>docker run -p 9443:9443 -p 1414:1414 \<br>  --env LICENSE=accept \<br>  --env MQ_QMGR_NAME=QM1 \<br>  --env MQ_ADMIN_NAME=admin \<br>  --env MQ_ADMIN_PASSWORD=admin_password \<br>  --env MQ_APP_NAME=app \<br>  --env MQ_APP_PASSWORD=password \<br>  --name ibmmq-container \<br>  localhost/ibm-mqadvanced-server-dev:9.4.1.0-arm64</pre><p>Here’s a breakdown of what each part of the command does:</p><ul><li>docker run: This is the core Docker command to create and run a new container.</li><li>-p 9443:9443 -p 1414:1414: These flags map the container&#39;s ports to the host machine&#39;s ports.</li><li>1414: This is the default port for the <strong>IBM MQ listener</strong>. Ballerina clients will use this port to connect and send/receive messages.</li><li>9443: This is the port for the <strong>IBM MQ Web Console and REST APIs</strong>. You can use this port to access a web-based UI to administer the queue manager and its queues.</li><li>--env LICENSE=accept: This environment variable is mandatory to run the IBM MQ container. It accepts the IBM MQ license agreement.</li><li>--env MQ_QMGR_NAME=QM1: This variable sets the name of the <strong>queue manager</strong> to QM1. The queue manager is the primary server that hosts and manages the message queues.</li><li>--env MQ_ADMIN_NAME=admin<strong> &amp; </strong>--env MQ_ADMIN_PASSWORD=admin_password: These set the credentials for the <strong>administrator user</strong>, which has full control over the queue manager.</li><li>--env MQ_APP_NAME=app<strong> &amp; </strong>--env MQ_APP_PASSWORD=password: These set the credentials for a dedicated <strong>application user</strong>. This is the user that your Ballerina service will use to connect and interact with the message queues.</li><li>--name ibmmq-container: This gives the container a friendly name, making it easier to refer to in future commands (e.g., docker exec ibmmq-container ...).</li><li>localhost/ibm-mqadvanced-server-dev:9.4.1.0-arm64: This is the name of the Docker image you built in the previous step. It tells Docker which image to use for the container.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I6hIrUSMu4I5Q2ZpZI2pJA.png" /><figcaption>Terminal: Running the IBM container</figcaption></figure><h4>1.3) Accessing Our IBM MQ Instance</h4><p>The server is now ready, and we can access it at:</p><ul><li><strong>🌐 Web Console</strong>: <a href="https://localhost:9443/ibmmq/console/">https://localhost:9443/ibmmq/console/</a></li><li><strong>🔐 Username</strong>: admin</li><li><strong>🔑 Password</strong>: admin_password</li><li><strong>📡 MQ Port</strong>: 1414 (for applications to connect)</li><li><strong>📊 REST API</strong>: <a href="https://localhost:9443/ibmmq/rest/">https://localhost:9443/ibmmq/rest/</a></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/567/1*2gWvDZaHMIjszD8UBTlv7g.png" /><figcaption>IBM MQ Console: Login</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jTgSpcuanhZlvvpTfWF9SQ.png" /><figcaption>IBM MQ Console</figcaption></figure><h4><strong>1.4) Creating a Queue</strong></h4><p>After logging into the IBM MQ web console with the admin user, we will create a new local queue to hold the appointment messages.</p><p>Navigate to the <strong>Manage</strong> tab and click on <strong>Queue Managers</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TFl3WqhMVbZE76eOPlA6tg.png" /><figcaption>IBM MQ Console: Manage Queue Manager</figcaption></figure><p>Select our queue manager (QM1) and go to the <strong>Queues</strong> tab. Click the blue <strong>Create</strong> button.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gvO-iDhCqQAm9mNS033YUA.png" /><figcaption>IBM MQ Console: Create queue</figcaption></figure><p>A new window will appear. On the left, under <strong>Queue type</strong>, select <strong>Local</strong>. A local queue is the standard type for storing messages within the queue manager. Click the <strong>Next</strong> button.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EMNBgZ21vSf9v2As5ZJ3cg.png" /><figcaption>IBM MQ Console: Choose queue type</figcaption></figure><p>In the “Quick create” tab, enter APPOINTMENT.QUEUE as the <strong>Queue Name</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VVmj5H2nBiDRr2-TuulseQ.png" /><figcaption>IBM MQ Console: Quick queue creation</figcaption></figure><p>Click the <strong>Custom create</strong> button, then select the <strong>Extended</strong> tab, and set <strong>Default persistence</strong> to Persistent. This is a critical step to ensure messages are not lost if the queue manager is restarted. Then, return to the “Quick create” tab and click the blue <strong>Create</strong> button to finalize.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uj2wiD_tneN47wkjWgWPDg.png" /><figcaption>IBM MQ Console: Message persistence</figcaption></figure><p>We will see a confirmation message that APPOINTMENT.QUEUE was successfully created, and it will appear in our queue list.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mba0R6eBDvEzdu7lLUBLkA.png" /><figcaption>IBM MQ Console: APPOINTMENT.QUEUE created</figcaption></figure><h4>1.5) Granting Permissions to the Application User</h4><p>Before our Ballerina service can interact with the queue, we must give the app user the necessary permissions. We will use the setmqaut command inside the container to do this.</p><p><strong>1.5a) Accessing the container’s shell</strong>: Open a new terminal and run the following command to get a bash shell inside your IBM MQ container.</p><pre>docker exec -it ibmmq-container bash</pre><p><strong>1.5b) Setting queue permissions</strong>: Run the setmqaut command to grant the app user permissions to put, get, inquire, and browse messages on the APPOINTMENT.QUEUE.</p><pre>setmqaut -m QM1 -t q -n APPOINTMENT.QUEUE -p app +put +get +inq +browse</pre><ul><li>-m QM1: Specifies the queue manager.</li><li>-t q: Specifies the object type, in this case, a queue.</li><li>-n APPOINTMENT.QUEUE: Specifies the name of the queue.</li><li>-p app: Specifies the principal (user) to whom permissions are being granted.</li><li>+put +get +inq +browse: Specifies the permissions to grant.</li></ul><p><strong>1.5c) Verifying permissions</strong>: To confirm the permissions were set correctly, run the dspmqaut command.</p><pre>dspmqaut -m QM1 -t q -n APPOINTMENT.QUEUE -p app</pre><p>You should see an output similar to this, confirming the app user has the required permissions:</p><pre>Entity app has the following authorizations for object APPOINTMENT.QUEUE:<br> get<br> browse<br> put<br> inq</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/589/1*nVrwdnZhKkEyiMQVivgEtw.png" /><figcaption>Terminal Granting permissions to the app user</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/578/1*ruxdceRippmT-8IZbc7-Tg.png" /><figcaption>Terminal: Confirming permissions given to the app user</figcaption></figure><h3>2) Setting Up MySQL</h3><p>This section walks you through setting up a local MySQL database instance using Docker. This database will be the destination for the appointment data from IBM MQ.</p><p><strong>2.1) Pulling the MySQL Community Server Docker Image</strong></p><p>Let’s start by pulling the official MySQL Community Server image from Docker Hub. We’ll be using version8.4, as shown in the command below:</p><pre>docker pull mysql:8.4</pre><p><strong>2.2) Running the MySQL container</strong></p><p>Next, run the following command to create and start a new MySQL container. This command configures the container with a name, port mapping, and the root user’s password.</p><pre>docker run --name mysql-container -p 3306:3306 -e MYSQL_ROOT_PASSWORD=rootpass -d mysql:8.4</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*HIaT4patqpSe9o76YwyOxQ.png" /><figcaption>Terminal: Running MySQL container</figcaption></figure><ul><li>--name mysql-container: Gives the container a memorable name.</li><li>-p 3306:3306: Maps the default MySQL port (3306) inside the container to the same port on your host machine, allowing you to connect to it.</li><li>-e MYSQL_ROOT_PASSWORD=rootpass: Sets the root user&#39;s password.</li></ul><blockquote><em>⚠️ </em><strong>Note:</strong> For a real-world application, you would use a strong, secure password.</blockquote><ul><li>-d: Runs the container in &quot;detached&quot; mode, so it runs in the background.</li></ul><p><strong>2.3) Accessing the MySQL Client</strong></p><p>To interact with the running database, open the MySQL command-line client inside the container. This command allows you to execute SQL queries directly.</p><pre>docker exec -it mysql-container mysql -uroot -prootpass</pre><ul><li>docker exec -it mysql-container: Executes a command (mysql) inside the running mysql-container with an interactive terminal (-it).</li><li>mysql -uroot -prootpass: The command to start the MySQL client, logging in with the root user and the password you set.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/602/1*mGkYhigeeJAKEKHeZGyfFg.png" /><figcaption>Terminal: Accessing MySQL client</figcaption></figure><p><strong>2.4) Creating the Database and Tables</strong></p><p>Once you’re inside the MySQL client, simply copy and paste the following SQL script into the terminal. The script will be executed immediately, creating the blueriver database and the three tables (patient, doctor, and appointment) needed for our application</p><pre>CREATE DATABASE IF NOT EXISTS blueriver;<br><br>USE blueriver;<br><br>CREATE TABLE IF NOT EXISTS patient (<br>   patientId VARCHAR(255) PRIMARY KEY,<br>   patientName VARCHAR(255) NOT NULL,<br>   patientEmail VARCHAR(255) NOT NULL UNIQUE,<br>   patientPhoneNumber VARCHAR(20)<br>);<br><br>CREATE TABLE IF NOT EXISTS doctor (<br>   doctorId VARCHAR(255) PRIMARY KEY,<br>   doctorName VARCHAR(255) NOT NULL,<br>   specialization VARCHAR(255)<br>);<br><br>CREATE TABLE IF NOT EXISTS appointment (<br>   appointmentId VARCHAR(255) PRIMARY KEY,<br>   patientId VARCHAR(255) NOT NULL,<br>   doctorId VARCHAR(255) NOT NULL,<br>   hospital VARCHAR(255) NOT NULL,<br>   appointmentTime DATETIME NOT NULL,<br>   status VARCHAR(50),<br>   notes TEXT,<br>   FOREIGN KEY (patientId) REFERENCES patient(patientId),<br>   FOREIGN KEY (doctorId) REFERENCES doctor(doctorId)<br>);</pre><h3>3) Setting Up the BI Extension in VS Code</h3><p>This section guides you through setting up your development environment. You’ll need Visual Studio Code and the official Ballerina extension to build the integration project.</p><ol><li><strong>Install Visual Studio Code</strong>: If you don’t already have VS Code, download and install it from the official website <strong>[2]</strong>.</li><li><strong>Install the WSO2 BI Extension</strong>: Open VS Code, navigate to the Extensions view (Ctrl+Shift+X), and search for <strong>“WSO2 BI Integrator”</strong>. Install the official extension published by WSO2<strong>[3]</strong>.</li></ol><blockquote><em>💡</em><strong><em>Hint:</em></strong> If you prefer to skip the manual project creation steps, you can clone the completed project directly from the following GitHub repository: <a href="https://github.com/lianecafarate/critical-appointment-integration">https://github.com/lianecafarate/critical-appointment-integration</a></blockquote><p><strong>3.1) Creating the BI Project</strong></p><p>Now that your environment is ready, let’s create the integration project. On the left-hand menu, select <strong>WSO2 Integrator BI</strong>. Then, click on <strong>Create New Integration</strong>, provide a name for your integration, and choose a location to save your project.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RL4-gQYBn39Hjbo111NlJw.png" /><figcaption>VS Code: WSO2 Integrator: BI Extension</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7dhZB4CWkK6NzWVNU9gRWA.png" /><figcaption>VS Code: Create your integration</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vk7_F0IIoBIhmSQ43RpgjQ.png" /><figcaption>VS Code: Project created</figcaption></figure><h3>4) Writing the Code</h3><p>Now that our project is set up, we can start building the integration.</p><p>While the WSO2 Integrator BI provides a <strong>low-code designer</strong> and a powerful <strong>AI assistant (copilot)</strong> to accelerate development, we will walk through the code step-by-step (<strong>pro-code approach</strong>) for a better understanding of how the integration works.</p><p>Open the <strong>Explorer</strong> view by clicking the documents icon on the left-hand menu. Here, you’ll see a list of files that were created with your project, including config.bal and connections.bal.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/386/1*8Odf16W44YOlEhfyg2NhrQ.png" /><figcaption>VS Code: WSO2 Integrator BI — Explorer vie</figcaption></figure><p>We will begin by configuring the connections in the next section.</p><h4><strong>4.1) MySQL Connection</strong></h4><p>This section explains how to configure a MySQL database connection in your Ballerina project. The configuration is split across two files for clarity and best practices: connections.bal for client instantiation and config.bal for environment-specific variables.</p><p>connections.bal</p><pre>import ballerinax/mysql;<br>import ballerinax/mysql.driver as _;<br><br>// MySQL client initialization<br>final mysql:Client mysqlClient = check new (dbHost, dbUser, dbPassword, dbName, dbPort);</pre><p>This file initializes the MySQL client that your application will use to connect to the database.</p><p>config.bal</p><p>This file holds the configurable variables for your database connection. By keeping these details separate, you can easily change them for different environments (e.g., development, testing, production) without modifying your core application logic.</p><pre>// MySQL database configuration<br>configurable string dbHost = &quot;localhost&quot;;<br>configurable string dbUser = &quot;root&quot;;<br>configurable string dbPassword = &quot;rootpass&quot;;<br>configurable string dbName = &quot;blueriver&quot;;<br>configurable int dbPort = 3306;</pre><blockquote><em>⚠️</em> Remember to edit the values to match your specific setup.</blockquote><h4><strong>4.2) IBM MQ Connection</strong></h4><p>connections.bal</p><p>This file creates the QueueManager client, which acts as the main point of communication with your IBM MQ server.</p><pre>import ballerinax/ibm.ibmmq;<br><br>// IBM MQ client initialization<br>final ibmmq:QueueManager queueManager = check new (<br>    name = queueManagerName, host = queueHost, port = queuePort, channel = queueChannel, userID = queueUserID, password = queuePassword<br>);</pre><p>config.bal</p><p>This file holds all the configurable parameters for your IBM MQ connection. By using the configurable keyword, you can easily change these values for different environments without modifying your code.</p><pre>// IBM MQ configuration<br>configurable string queueManagerName = &quot;QM1&quot;;<br>configurable string queueHost = &quot;localhost&quot;;<br>configurable int queuePort = 1414;<br>configurable string queueChannel = &quot;DEV.APP.SVRCONN&quot;;<br>configurable string queueUserID = &quot;app&quot;;<br>configurable string queuePassword = &quot;password&quot;;<br>configurable string queueName = &quot;APPOINTMENT.QUEUE&quot;;<br>configurable decimal pollingInterval = 1.0;</pre><ul><li>queueManagerName = &quot;QM1&quot;: The name of the queue manager you created in your Docker container.</li><li>queueHost = &quot;localhost&quot; and queuePort = 1414: The host and port of the IBM MQ server. These values correspond to the port mapping you set up in the docker run command.</li><li>queueChannel = &quot;DEV.APP.SVRCONN&quot;: The name of the server-connection channel used by the client to connect to the queue manager.</li><li>queueUserID = &quot;app&quot; and queuePassword = &quot;password&quot;: The credentials for the application user, which you granted permissions to in a previous step.</li><li>queueName = &quot;APPOINTMENT.QUEUE&quot;: The name of the queue that the Ballerina service will listen to.</li><li>pollingInterval = 1.0: This sets how often the listener checks the queue for new messages, in seconds.</li></ul><h4>4.3) Consolidated Connection Configuration</h4><p>With both the MySQL and IBM MQ connections defined, let’s look at the final, consolidated connections.bal and config.bal files. These two files contain all the necessary configurations for your Ballerina service to communicate with both the database and the message queue.</p><p>connections.bal</p><p>This file now imports both the mysql and ibmmq modules and initializes both clients.</p><pre>import ballerinax/mysql;<br>import ballerinax/mysql.driver as _;<br>import ballerinax/ibm.ibmmq;<br><br>// MySQL client initialization<br>final mysql:Client mysqlClient = check new (dbHost, dbUser, dbPassword, dbName, dbPort);<br><br>// IBM MQ client initialization<br>final ibmmq:QueueManager queueManager = check new (<br>    name = queueManagerName, host = queueHost, port = queuePort, channel = queueChannel, userID = queueUserID, password = queuePassword<br>);</pre><p>config.bal</p><p>This file is a single source of truth for all your configurable environment variables. It includes the settings for both the database and IBM MQ.</p><pre>// MySQL database configuration<br>configurable string dbHost = &quot;localhost&quot;;<br>configurable string dbUser = &quot;root&quot;;<br>configurable string dbPassword = &quot;rootpass&quot;;<br>configurable string dbName = &quot;blueriver&quot;;<br>configurable int dbPort = 3306;<br><br>// IBM MQ configuration<br>configurable string queueManagerName = &quot;QM1&quot;;<br>configurable string queueHost = &quot;localhost&quot;;<br>configurable int queuePort = 1414;<br>configurable string queueChannel = &quot;DEV.APP.SVRCONN&quot;;<br>configurable string queueUserID = &quot;app&quot;;<br>configurable string queuePassword = &quot;password&quot;;<br>configurable string queueName = &quot;APPOINTMENT.QUEUE&quot;;<br>configurable decimal pollingInterval = 1.0;</pre><h4>4.4) Configuring the IBM MQ Dependency</h4><p>To use the IBM MQ connector<strong>[4]</strong>, you need to add the IBM MQ client libraries as a dependency in your project’s Ballerina.toml file. This ensures the necessary Java components are available at runtime.</p><p>Add the following to your Ballerina.toml:</p><pre>[[platform.java21.dependency]]<br>groupId = &quot;com.ibm.mq&quot;<br>artifactId = &quot;com.ibm.mq.allclient&quot;<br>version = &quot;9.4.1.0&quot;</pre><h4><strong>4.5) </strong>Defining Data Structures with Records</h4><p>In this section, we use Ballerina’s record<strong> </strong>type to define the structure of our data. Records are a powerful feature for modeling structured data, similar to a struct in C or a class in object-oriented languages, but simpler and more focused on data representation.</p><p>Here are the records we’ll use for our application’s data:</p><p>types.bal</p><pre>// Patient information record<br>public type Patient record {|<br>    string patientId;<br>    string patientName;<br>    string patientEmail;<br>    string patientPhoneNumber?;<br>|};<br><br>// Doctor information record<br>public type Doctor record {|<br>    string doctorId;<br>    string doctorName;<br>    string specialization?;<br>|};<br><br><br>// Appointment record<br>public type Appointment record {|<br>    string appointmentId;<br>    Patient patient;<br>    Doctor doctor;<br>    string hospital;<br>    //YYYY-MM-DD hh:mm:ss<br>    string appointmentTime;<br>    string status?;<br>    string notes?;<br>|};</pre><p>The ? symbol makes a variable or a field in a record <strong>optional. </strong>This means it can hold a value of the specified type, or it can be () (nil). This is useful when a value might not exist.</p><p>The | symbol indicates a close record, which means it can <strong>only</strong> contain the fields that are explicitly listed within the pipes. This provides strong type safety, which is a core benefit of Ballerina<strong>[5]</strong>.</p><h4>4.6) Defining Functions for Application Logic</h4><p>This section outlines the core functions of our application, which handle communication with both IBM MQ and the MySQL database. These functions are placed in functions.bal to keep the main service logic clean and organized.</p><p>AppointmentRow <strong>Record Type</strong></p><p>The AppointmentRow record is a specific data type used to model the results returned from a MySQL database query. It defines the structure for a single row from the appointment table, including optional fields like status and notes which might be nil.</p><pre>type AppointmentRow record {|<br>    string appointmentId;<br>    string patientId;<br>    string doctorId;<br>    string hospital;<br>    string appointmentTime;<br>    string? status;<br>    string? notes;<br>|};</pre><p>sendMessageToQueue()</p><p>This function demonstrates how to <strong>produce</strong> a message to the IBM MQ queue.</p><ul><li>ibmmq:Queue queue = check queueManager.accessQueue(...): This line uses the queueManager client to access the APPOINTMENT.QUEUE. The ibmmq:MQOO_OUTPUT flag specifies that the queue will be used for sending messages.</li><li>The function then constructs a sample JSON payload representing an appointment.</li><li>check queue-&gt;put(message): This is the core call that sends the message to the queue. The check keyword ensures any potential errors during this operation are handled.</li><li>check queue-&gt;close(): It&#39;s a best practice to close the queue after you&#39;re done with it to release resources.</li></ul><p>getAppointment()</p><p>This function queries the database to retrieve a single appointment record by its appointmentId.</p><ul><li>sql:ParameterizedQuery selectQuery = ...: A ParameterizedQuery is used to prevent <strong>SQL injection attacks</strong>. It safely embeds the appointmentId into the query.</li><li>mysqlClient-&gt;queryRow(...): This method executes the query and expects to retrieve a single row, which is then mapped to the AppointmentRow record type.</li><li>The function uses an if-else block to handle three possible outcomes: a successful result (AppointmentRow), no rows found (sql:NoRowsError), or a database error (sql:Error).</li></ul><p>scheduleAppointment() and insertAppointment()</p><p>These functions work together to handle the business logic of scheduling an appointment.</p><ul><li>scheduleAppointment(): This is the main function that coordinates the process. It first calls getAppointment() to check if the appointment already exists. This prevents duplicate entries.</li><li>insertAppointment(): If the appointment doesn&#39;t exist, this function is called to insert the new record into the database. It also uses a ParameterizedQuery to securely insert the data from the Appointment record.</li></ul><p>The use of log:printInfo throughout these functions provides clear feedback on the application&#39;s state, which is essential for debugging and monitoring.</p><p>functions.bal</p><pre>import ballerina/sql;<br>import ballerina/log;<br>import ballerinax/ibm.ibmmq;<br><br>// Define record type for appointment query results<br>type AppointmentRow record {|<br>    string appointmentId;<br>    string patientId;<br>    string doctorId;<br>    string hospital;<br>    string appointmentTime;<br>    string? status;<br>    string? notes;<br>|};<br><br>public function sendMessageToQueue() returns error? {<br>    ibmmq:Queue queue = check queueManager.accessQueue(queueName, ibmmq:MQOO_OUTPUT);<br>    string payloadString = &quot;{\n&quot; +<br>        &quot;    \&quot;appointmentId\&quot;: \&quot;APT001\&quot;,\n&quot; +<br>        &quot;    \&quot;patient\&quot;: {\n&quot; +<br>        &quot;        \&quot;patientId\&quot;: \&quot;PAT002\&quot;,\n&quot; +<br>        &quot;        \&quot;patientName\&quot;: \&quot;Amy Williams\&quot;,\n&quot; +<br>        &quot;        \&quot;patientEmail\&quot;: \&quot;amy.williams@gmail.com\&quot;,\n&quot; +<br>        &quot;        \&quot;patientPhoneNumber\&quot;: \&quot;+1-555-123-4567\&quot;\n&quot; +<br>        &quot;    },\n&quot; +<br>        &quot;    \&quot;doctor\&quot;: {\n&quot; +<br>        &quot;        \&quot;doctorId\&quot;: \&quot;DOC002\&quot;,\n&quot; +<br>        &quot;        \&quot;doctorName\&quot;: \&quot;Rafael Costa\&quot;,\n&quot; +<br>        &quot;        \&quot;specialization\&quot;: \&quot;Cardiology\&quot;\n&quot; +<br>        &quot;    },\n&quot; +<br>        &quot;    \&quot;hospital\&quot;: \&quot;Green Valley Hospital\&quot;,\n&quot; +<br>        &quot;    \&quot;appointmentTime\&quot;: \&quot;2025-08-15 10:30:00\&quot;,\n&quot; +<br>        &quot;    \&quot;status\&quot;: \&quot;scheduled\&quot;,\n&quot; +<br>        &quot;    \&quot;notes\&quot;: \&quot;Patient reported occasional chest pain. Follow-up required.\&quot;\n&quot; +<br>        &quot;}&quot;;<br><br>    ibmmq:Message message = { payload: payloadString.toBytes() };<br>    check queue-&gt;put(message);<br>    check queue-&gt;close();<br>    log:printInfo(&quot;Message sent to the queue! :)&quot;);<br>    return;<br>}<br><br>public function getAppointment(string appointmentId) returns string|AppointmentRow|sql:Error {<br>    sql:ParameterizedQuery selectQuery = `<br>        SELECT appointmentId, patientId, doctorId, hospital, appointmentTime, status, notes<br>        FROM appointment <br>        WHERE appointmentId = ${appointmentId}<br>    `;<br>    <br>    var result = mysqlClient-&gt;queryRow(selectQuery, AppointmentRow);<br><br>    if result is AppointmentRow {<br>        // Appointment exists, return the Appointment record<br>        return result;<br>    } else if result is sql:NoRowsError {<br>        // No appointment found with this ID<br>        return string `Appointment with ID &#39;${appointmentId}&#39; does not exist`;<br>    } else {<br>        // This is an actual database error<br>        return result;<br>    }<br>}<br><br>public function scheduleAppointment(Appointment appointment) returns error? {<br>    log:printInfo(&quot;==Schedule Appointment Function==&quot;);<br>    // Check if the appointment already exists<br>    var existingAppointment = getAppointment(appointment.appointmentId);<br><br>    if existingAppointment is AppointmentRow {<br>        // Appointment already exists<br>        return error(&quot;Appointment with ID &#39;&quot; + appointment.appointmentId + &quot;&#39; already exists.&quot;);<br>    } else if existingAppointment is string {<br>        // Appointment does not exist (getAppointment returned string message), proceed to schedule<br>        log:printInfo(&quot;No existing appointment found, proceeding to insert new appointment.&quot;);<br>        return insertAppointment(appointment);<br>    } else {<br>        // Handle database error (remaining type is sql:Error)<br>        return error(&quot;Database error: &quot; + existingAppointment.message());<br>    }<br>}<br><br>public function insertAppointment(Appointment appointment) returns error? {<br>    log:printInfo(&quot;==Insert Appointment Function==&quot;);<br>    sql:ParameterizedQuery insertQuery = `<br>        INSERT INTO appointment (appointmentId, patientId, doctorId, hospital, appointmentTime, status, notes)<br>        VALUES (${appointment.appointmentId}, ${appointment.patient.patientId}, ${appointment.doctor.doctorId}, ${appointment.hospital}, <br>                ${appointment.appointmentTime}, ${appointment.status}, ${appointment.notes})<br>    `;<br><br>    var result = mysqlClient-&gt;execute(insertQuery);<br><br>    if result is sql:Error {<br>        // Handle database error<br>        return error(&quot;Database error: &quot; + result.message());<br>    } else {<br>        // Appointment inserted successfully<br>        log:printInfo(&quot;Appointment inserted successfully with ID: &quot; + appointment.appointmentId);<br>        return;<br>    }<br>}</pre><h3>4.7) Implementing the Ballerina Service</h3><p>This is the core of our integration. The service.bal file contains the Ballerina service that listens for and processes messages from IBM MQ. The service is designed to be <strong>transactional</strong>, ensuring that messages are only removed from the queue if the database operation is successful.</p><p><strong>IBM MQ Listener</strong></p><p>The listener is configured to connect to your IBM MQ server. This code instantiates an ibmmq:Listener using the configurable variables from config.bal and connections.bal.</p><p><strong>Service Configuration</strong></p><p>The @ibmmq:ServiceConfig annotation is a crucial part of the service. It specifies how the service should behave when it receives messages.</p><ul><li>queueName: This specifies which queue the service should listen to.</li><li>sessionAckMode: ibmmq:SESSION_TRANSACTED: This is the most important part of our use case. It configures the session to be <strong>transactional</strong>. This means that a message is only removed from the queue after the service explicitly calls commit(). If an error occurs and the service callsrollback(), the message remains on the queue to be re-processed later.</li><li>pollingInterval: This specifies how often the listener polls the queue for new messages.</li></ul><p>onMessage <strong>Function</strong></p><p>This is the main function that executes every time a new message is received from the APPOINTMENT.QUEUE.</p><ul><li><strong>Decode Message</strong>: The payload of the IBM MQ message, which is a byte array, is converted into a string and then a json object. This JSON is then converted into a strongly-typed Appointment record using cloneWithType().</li><li><strong>Process Transactionally</strong>: The service calls scheduleAppointment() to insert the data into the database.</li></ul><p><strong>Conditional Commit/Rollback</strong>:</p><ul><li>If scheduleAppointment() returns an error, the service calls caller-&gt;&#39;rollback(). This tells IBM MQ not to acknowledge the message, and it will be made available for redelivery.</li><li>If the function completes successfully, the service calls caller-&gt;&#39;commit(). This acknowledges the message, and it is permanently removed from the IBM MQ queue.</li></ul><pre>import ballerinax/ibm.ibmmq;<br>import ballerina/log;<br><br>//IBM MQ Listener creation<br>listener ibmmq:Listener ibmmqListener = check new({<br>    name: queueManagerName,<br>    host: queueHost,<br>    port: queuePort,<br>    channel: queueChannel,<br>    userID: queueUserID,<br>    password: queuePassword<br>});<br><br>@ibmmq:ServiceConfig {<br>    queueName,<br>    sessionAckMode: ibmmq:SESSION_TRANSACTED,<br>    pollingInterval<br>}<br><br>//Service to handle incoming messages from IBM MQ<br>service on ibmmqListener {<br>    remote function onMessage(ibmmq:Message message, ibmmq:Caller caller) returns error? {<br>        string payloadString = check string:fromBytes(message.payload);<br>        log:printInfo(&quot;Received message from IBM MQ&quot;, message = payloadString);<br><br>        json payloadJson = check payloadString.fromJsonString();<br>        Appointment appointment = check payloadJson.cloneWithType(Appointment);<br>        log:printInfo(&quot;Processing appointment&quot;, appointment = appointment);<br><br>        // Try to schedule the appointment<br>        error? scheduleResult = scheduleAppointment(appointment);<br>        if scheduleResult is error {<br>            log:printError(&quot;Error scheduling appointment: &quot; + scheduleResult.message());<br>            check caller-&gt;&#39;rollback(); // rollback the message so it can be redelivered<br>        } else {<br>            log:printInfo(&quot;Appointment scheduled successfully&quot;);<br>            check caller-&gt;&#39;commit(); // acknowledge the message and commit the transaction<br>        }<br>    }<br>}</pre><h3>4.8) The Main Function</h3><p>The main function serves as the entry point for your Ballerina program. When you run your service, the code within this function is executed first. In our case, the main.bal file is used to demonstrate how to send a message to the IBM MQ queue.</p><pre>import ballerina/io;<br><br>public function main() returns error? {<br>    io:println(&quot;Appointment Service is starting...&quot;);<br><br>    check sendMessageToQueue();       <br>}</pre><ul><li>public function main() returns error?: This is the standard signature for a Ballerina executable. The returns error? part indicates that the function can either return () (nil) on success or an error on failure.</li><li>io:println(&quot;Appointment Service is starting...&quot;): This line simply prints a message to the console to confirm that the service has started.</li><li>check sendMessageToQueue(): This is the key line in this function. It calls the sendMessageToQueue() function that you defined in functions.bal. The check keyword handles any potential errors that the function might return.</li></ul><p>The purpose of this main function is to quickly <strong>produce a message</strong> to the queue. After this message is sent, the Ballerina service you defined in service.bal will automatically pick it up and process it. This setup allows you to test the entire flow from message production to transactional database persistence.</p><h3>In Conclusion: Zero Data Loss, Effortless Integration</h3><p>We’ve now seen how the powerful combination of WSO2 Ballerina and IBM MQ can build a robust, transactional integration that guarantees data integrity. By leveraging IBM MQ’s reliable messaging and Ballerina’s native transactional capabilities, we’ve created a system that:</p><ul><li><strong>Eliminates Data Loss</strong>: Messages are not lost even if the database is down, because they remain in the IBM MQ queue until they can be processed successfully.</li><li><strong>Ensures Atomicity</strong>: The entire operation, from consuming the message to writing to the database, is treated as a single, atomic unit. It either succeeds completely or fails entirely, leaving no incomplete records.</li><li><strong>Simplifies Development</strong>: Ballerina’s service-oriented syntax and strong typing make it easy to define the data structures, configure the connections, and write the processing logic.</li></ul><p>This architecture provides a scalable and resilient solution for handling mission-critical data flows. Whether you’re integrating with other messaging systems, databases, or APIs, Ballerina’s unique design and rich ecosystem of connectors provide the tools you need to build reliable, enterprise-grade applications.</p><h3>🎬 Wrapping Up</h3><p>🙌 Many thanks for sticking with me through it all.💪</p><p>If you found this article helpful, please <strong>give it a clap</strong> 👏. It really motivates me to create more content like this. Got questions, feedback, or your own tips? Drop a comment below. I’d love to hear from you 👂!</p><p>And don’t forget to <strong>share</strong> 🔁 this article with your network if you think others could benefit.</p><p>Here’s to smooth integrations and happy coding! ✨</p><h3>📚 References</h3><p>[<a href="https://community.ibm.com/community/user/blogs/richard-coppen/2023/06/30/ibm-mq-9330-container-image-now-available-for-appl">1] IBM MQ Container image for Apple Silicon (Arm64)</a></p><p>[2] <a href="https://code.visualstudio.com/docs/setup/mac">VS Code for macOS</a></p><p>[3] <a href="https://marketplace.visualstudio.com/items?itemName=WSO2.ballerina-integrator">VS Code Extension — WSO2 Integrator BI</a></p><p>[4] <a href="https://central.ballerina.io/ballerinax/ibm.ibmmq/latest">Ballerina IBM MQ Connector</a></p><p>[5] <a href="https://ballerina.io/learn/by-example/open-records/">Ballerina: Open records</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6e6847a74d70" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What Immigration Does to You]]></title>
            <link>https://lianecafarate.medium.com/what-immigration-does-to-you-13018842fc0e?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/13018842fc0e</guid>
            <category><![CDATA[women-empowering-women]]></category>
            <category><![CDATA[immigration]]></category>
            <category><![CDATA[women-in-tech]]></category>
            <category><![CDATA[expat-life]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Thu, 17 Jul 2025 03:10:36 GMT</pubDate>
            <atom:updated>2025-07-17T15:13:22.815Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Reflections sparked by “Passport to Growth” by Patricia East</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5GbkehziXa8zaLfj" /><figcaption>Photo by <a href="https://unsplash.com/@kylejglenn?utm_source=medium&amp;utm_medium=referral">Kyle Glenn</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>This reflection was inspired by Patricia East’s book, <a href="https://www.amazon.com/Passport-Growth-Leadership-Lessons-Expat-ebook/dp/B0CW1J2VXV"><em>Passport to Growth</em></a><em>.</em> I just want to share what the book made me feel, remember, and realize.</p><h3>📚How I Got the Book</h3><p>If I’m not mistaken, I met Patricia for the first time at one of <a href="https://www.theexpatwoman.com/">The Expat Woman</a> events. There was a holiday gift exchange 🎁, you know, the kind where you pick one gift from a table. I didn’t know her personally, just through a mutual friend. But I saw her book and, as a book lover, I naturally chose it as my gift.❤️📘</p><p>It took me a while to finally open it, you know, life gets busy (as it always does), but it sat patiently on my nightstand waiting for its moment. Last Sunday, that moment came. And when I started reading, I couldn’t stop. I devoured the entire book in just a few hours ⏳✨ and couldn’t wait to share my experience.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ftenor.com%2Fembed%2F16523529&amp;display_name=Tenor&amp;url=https%3A%2F%2Ftenor.com%2Fview%2Fbaby-excited-creepy-happy-birthday-best-present-ever-gif-16523529&amp;image=https%3A%2F%2Fmedia.tenor.com%2F5rH3uOCJxw8AAAAN%2Fbaby-excited.png&amp;type=text%2Fhtml&amp;schema=tenor" width="600" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/4210dac0c7266bc61c9f05aec279748e/href">https://medium.com/media/4210dac0c7266bc61c9f05aec279748e/href</a></iframe><h3>👩When You Know the Author</h3><p>Reading it felt like having a friend sit across from me and open up. I could hear Patricia’s voice in my head, her tone, her rhythm, her vulnerability.🫂</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*KmwNC51_mQG9-p8_zq4NSA.gif" /></figure><h3>💪Vulnerability Is a Superpower</h3><p>If you’ve heard of <a href="https://www.youtube.com/watch?v=iCvmsMzlF7o&amp;vl=en">Brené Brown</a>, you know this already: vulnerability is not weakness; it’s strength. Patricia’s stories struck me in that exact way. I found myself wanting to hug her and say, “<strong>I’m so sorry you had to go through this”</strong>, because she wrote with authenticity, with courage, and with heart.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*fqTv8KsZXy0_d0N1RBC-xA.gif" /></figure><h3>🙋🏽‍I Could Deeply Relate</h3><p>Throughout the book, I saw myself. Again and again. It was like time-traveling through my own journey, the highs, the lows, the resilience. I found myself scribbling notes in the margins, underlining moments that echoed my own experiences. It reminded me that the immigrant journey is unique, but also shared. 🌍</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*772qGlJ--qQ4xdnqnlZcmA.gif" /></figure><h3>🗣️I’m Proud of My Accent</h3><p>For a long time, I struggled with the idea that my accent made me seem less competent, a perception backed by research [1]📉, especially for Latinas. But I’ve learned that my accent tells a story. 🧳It’s a reminder of the diverse experiences I bring, perspectives shaped by living, working, and adapting in different cultures. That depth is something that can’t be taught, and it certainly shouldn’t be discounted.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*leeaEad81HnFocL4AVs5fg.gif" /></figure><h3>💁🏼‍♀️Be the Person Who Says “I’ll Show You”</h3><p>You remember <em>Legally Blonde</em>, right? That moment when Elle Woods decides to prove everyone wrong and goes full force into Harvard Law? Patricia has a moment like that in her book. (I don´t want to give away more details, but let’s just say someone gave her “feedback” that felt a lot more like prejudice.😒) What I loved was how she transformed that moment into motivation. 🔥It’s a powerful reminder: learn to separate judgment from real feedback, and let those experiences fuel your growth.🚀</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*L62wR9CiF6RfW8gXtO646Q.gif" /><figcaption>Elle Woods — Legally Blonde</figcaption></figure><h3>🌍What We Bring to the Table</h3><p>I don’t have hard data on this, but I’ve had several conversations with fellow immigrants to notice a pattern: many feel their professional experiences outside the U.S. are undervalued, even dismissed. That’s a shame. Working in environments with fewer resources or more instability teaches resilience, creativity, and grit. The world is bigger than just one country, and global experience <em>is</em> valuable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*gSbb-hfRBbJj7Pe8bU15bg.gif" /><figcaption>Elle Woods — Legally Blonde</figcaption></figure><h3>✨Share Your Story: It Matters</h3><p>Being an immigrant is hard. Sharing your journey, the wins, and the wounds is powerful. It helps others feel seen, understood, and less alone. It lights a path for those just beginning theirs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*N1SeuDLuWe-j3eG4dnQjzA.gif" /></figure><h3>🙌She Is Awesome</h3><p>Patricia East is a true inspiration for Latinas living abroad. She’s shown us that <em>yes, we can do it, </em>and she continues to share her knowledge and uplift others with honesty, strength, and heart. Through her book, she reminds us that while the journey isn’t always easy, it’s filled with growth, courage, and unforgettable adventures. She embodies the spirit of resilience and proves that our dreams are worth chasing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*zLvPhlFiHKeebDPRFAerng.gif" /><figcaption>The Devil Wears Prada</figcaption></figure><h3>📘Just Read the Book</h3><p>If you’re an immigrant, an expat, or someone navigating life outside your comfort zone, read this book. If you’ve ever felt like you didn’t belong, had to start over, or questioned whether your story mattered, read this book.</p><p>Passport to Growth isn’t just a memoir. It’s a reminder that you’re not alone. It’s honest, vulnerable, inspiring, and at times, beautifully raw. 💗You’ll laugh, nod in recognition, maybe even cry. 😄😭But most of all, you’ll finish it feeling seen, encouraged, and ready to keep going.</p><blockquote>Trust me, just read the book. You won’t regret it.<em>✨🙌</em></blockquote><blockquote><a href="https://www.amazon.com/Passport-Growth-Leadership-Lessons-Expat-ebook/dp/B0CW1J2VXV">Passport to Growth</a></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*u5NmO4DRXfv4GtHuduPLmA.gif" /></figure><h3>References</h3><p>[1] <a href="https://journals.sagepub.com/doi/10.1177/0739986316632319">Accent, Gender, and Perceived Competence</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=13018842fc0e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ Configuring MySQL on WSO2 API Manager 4.5 with Docker]]></title>
            <link>https://lianecafarate.medium.com/configuring-mysql-on-wso2-api-manager-4-5-with-docker-33b275d568a3?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/33b275d568a3</guid>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[api]]></category>
            <category><![CDATA[wso2]]></category>
            <category><![CDATA[wso2-api-manager]]></category>
            <category><![CDATA[mysql]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Fri, 11 Jul 2025 00:14:46 GMT</pubDate>
            <atom:updated>2025-07-11T00:23:09.350Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Xg0rPRqNaoY3nRQg" /><figcaption>Photo by <a href="https://unsplash.com/@rubaitulazad?utm_source=medium&amp;utm_medium=referral">Rubaitul Azad</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><blockquote><em>🔔</em> In this article, I’ll dive straight into configuring MySQL as the database for <strong>WSO2 API Manager 4.5</strong> using <strong>Docker</strong>. If you’re looking for an introduction to why WSO2 API Manager might be the right choice for your API management needs, I recommend checking out my previous article: <a href="https://lianecafarate.medium.com/deploying-wso2-api-manager-4-5-with-docker-319c86303b4b">Deploying WSO2 API Manager 4.5 with Docker</a>.</blockquote><h3>🐬 Why MySQL?</h3><p>WSO2 API Manager ships with an embedded <a href="https://www.h2database.com/html/main.html">H2 database</a>, which is ideal for development and testing purposes. However, for <strong>enterprise-grade production environments</strong>, WSO2 strongly recommends using a robust, industry-standard relational database management system (RDBMS), such as MySQL.</p><p>WSO2 API Manager 4.5 has been officially tested and verified with <strong>MySQL versions 5.7, 8.0, and 8.4</strong>. You can find more details in the <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/reference/product-compatibility/#tested-wso2-products">product compatibility guide</a>.</p><h3>🛠️ Prerequisites</h3><ol><li>🐳 Make sure you have <a href="https://rancherdesktop.io/"><strong>Rancher Desktop</strong></a> installed on your macOS system. For detailed steps, feel free to refer to my <a href="https://lianecafarate.medium.com/deploying-wso2-api-manager-4-5-with-docker-319c86303b4b">previous article</a>. Alternatively, you could use <a href="https://docs.docker.com/desktop/"><strong>Docker Desktop</strong></a> or <a href="https://podman.io/"><strong>Podman</strong></a><strong>.</strong></li><li>☕ Before getting started, make sure you meet the <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/install/installation-prerequisites/">official installation prerequisites for WSO2 API Manager</a>. One key requirement is having a <strong>JDK</strong> installed. I recommend using <a href="https://sdkman.io/"><strong>SDKMAN!</strong></a> for this. It makes it easy to install and switch between different JDK versions with just a few commands.</li><li>🐬 For this setup, we’ll be using MySQL 8.4 Community Server. Additionally, you’ll need to download the <a href="https://downloads.mysql.com/archives/c-j/">MySQL Connector/J (JDBC Driver) version 8.4.0</a>, which is required for WSO2 API Manager to connect to the MySQL database.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*anwzwRMzldSbIS8d1JEyWw.png" /><figcaption>MySQL Connector/J (Archived Versions)</figcaption></figure><h3>⚠️ For Testing &amp; Demo Purposes Only</h3><blockquote>This configuration is intended strictly for local testing, experimentation, and demonstration. It is NOT secured, optimized, or configured for production environments.</blockquote><h3>🧭 Step-by-Step Guide</h3><p>▶️ Ready to dive in? Let’s get started!</p><blockquote>For reference, this tutorial builds on the official WSO2 documentation on <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/setting-up-databases/changing-default-databases/changing-to-mysql/">switching from the default H2 database to MySQL</a>. I’ve adapted and extended it to show you how to do this specifically using <strong>Docker</strong> with <strong>WSO2 API Manager 4.5</strong>.</blockquote><h4>Default Databases in API Manager</h4><p>Before diving into the full configuration, it’s helpful to understand the default databases used by WSO2 API Manager. Each serves a distinct purpose within the product architecture. For a detailed overview, refer to the official documentation:<br> 👉 <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/setting-up-databases/overview/">WSO2 APIM Database Overview</a></p><ul><li><strong>AM database</strong>: Stores information related to the APIs along with the API subscription details. The API Manager database is also referred to as WSO2_AM_DB and apimgtdb.</li><li><strong>Shared database</strong>: Contains common data, including user management and the governance registry, used across WSO2 products. The Shared database is also referred to as WSO2_SHARED_DB and shareddb.</li><li><strong>Carbon database</strong>: Holds internal product-related data, such as configuration metadata. By default, this is stored in the embedded H2 database.</li></ul><p>Understanding these roles is key when migrating to an external database like MySQL, as each database needs to be configured separately.</p><blockquote><em>✅ </em>Alright, now that we’ve had a quick overview of the APIM databases, let’s start building our setup.<em>🛠️📦</em></blockquote><h4>Step 1: Pull the MySQL Community Server Docker Image</h4><p>Let’s start by pulling the official MySQL Community Server image from Docker Hub. We’ll be using version 8.4, as shown in the command below:</p><pre>docker pull mysql:8.4</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/575/1*fvn40aBJyDl855gL_5Up0A.png" /><figcaption>Terminal: Pulling the MySQL 8.4 Docker Image</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7QP8JafrnIWkG9SBcvDAOQ.png" /><figcaption>Rancher Desktop: MySQL 8.4 image pulled</figcaption></figure><p>You can find more details about the image on its official Docker Hub page: <a href="https://hub.docker.com/_/mysql">https://hub.docker.com/_/mysql</a>.</p><h4>Step 2: Create The Docker Compose File</h4><p>Start by creating a new directory to organize all your configuration files. In this example, we’ll call the folder mysqlconf:</p><pre>mkdir mysqlconf<br>cd mysqlconf</pre><p>Project Folder Structure:</p><pre>mysqlconf/<br>├── deployment.toml              # Custom WSO2 API Manager configuration<br>├── docker-compose.yml           # Main Docker Compose file<br>├── drivers/                     # MySQL JDBC driver<br>│   └── mysql-connector-j-8.4.0.jar<br>├── mysql-data/                  # MySQL data volume (optional; auto-created by Docker)<br>├── mysql-init/                  # SQL scripts to initialize databases<br>│   ├── 01-setup.sql             # Creates users and grants permissions<br>│   ├── 02-apimdb-mysql.sql      # Initializes the APIM database schema<br>│   └── 03-shareddb-mysql.sql    # Initializes the shared database schema</pre><ul><li>Your project folder will eventually look like this, but don’t worry, we’ll build each file step by step.</li><li>We’ll begin by creating the docker-compose.yml file and configure only the MySQL service. This allows us to validate that the database setup and initialization scripts are working correctly before introducing additional complexity.</li><li>Once everything is running smoothly on the database side, we’ll move on to adding the WSO2 API Manager service and wiring it up to use the MySQL databases we just configured.</li></ul><p>Now, let&#39;s create the docker-compose.ymlfile.</p><blockquote><em>🧱</em> If you’re new to Docker Compose or need a quick refresher, check out my earlier article <em>👉 </em><a href="https://lianecafarate.medium.com/configuring-choreo-analytics-on-wso2-api-manager-4-5-with-docker-53241881447f"><strong>Configuring Choreo Analytics on WSO2 API Manager 4.5 with Docker</strong></a> <em>🚀</em></blockquote><p>Create a file named docker-compose.yml inside the mysqlconf directory. You can use any text editor you like, even straight from the terminal with vim if you&#39;re feeling fancy (don’t forget: Esc + :wq! to save and exit. 😄).</p><pre>vim docker-compose.yml</pre><p>Then, copy and paste the following content into the file and save it.</p><pre>services:<br>  mysql:<br>      image: mysql:8.4<br>      container_name: mysql-server<br>      restart: unless-stopped<br>      environment:<br>        MYSQL_ROOT_PASSWORD: rootpass<br>      ports:<br>        - &quot;3306:3306&quot;<br>      volumes:<br>        - mysql-data:/var/lib/mysql<br>        - ./mysql-init:/docker-entrypoint-initdb.d<br>      healthcheck:<br>        test: [&quot;CMD&quot;, &quot;mysqladmin&quot;, &quot;ping&quot;, &quot;-uroot&quot;, &quot;-prootpass&quot;]<br>        interval: 10s<br>        timeout: 20s<br>        retries: 5<br>volumes:<br>  mysql-data:<br>    driver: local</pre><p>Let’s break down what this configuration does:</p><ul><li><strong>services:</strong> This is where we define the containers that will run. For now, we’re starting with only one: MySQL.</li><li><strong>mysql</strong>: This is the name of the service. We&#39;re using the official mysql:8.4 image and naming the container mysql-server.</li><li><strong>restart: unless-stopped</strong>: This ensures the container automatically restarts unless you explicitly stop it.</li><li><strong>environment</strong>: We&#39;re setting the root password for MySQL using MYSQL_ROOT_PASSWORD. (In a production setup, you&#39;d want to store this securely.)</li><li><strong>ports:</strong> This maps MySQL&#39;s default port 3306 inside the container to 3306 on your host machine, so you can connect to it locally.</li><li><strong>volumes:</strong><strong>mysql-data:/var/lib/mysql</strong>: It persists your database data outside the container.</li><li><strong>volumes:</strong><strong>./mysql-init:/docker-entrypoint-initdb.d</strong>: When you mount a folder into the MySQL container at /docker-entrypoint-initdb.d, any .sql, .sh, or .sql.gz files in that folder are automatically executed (in alphanumeric order) the very first time the MySQL container initializes a new data volume.</li><li><strong>healthcheck: </strong>Docker will check if the MySQL server is up by running a simple mysqladmin ping command every 10 seconds. This helps ensure other services only start once MySQL is ready.</li><li><strong>volumes:</strong><br> We define mysql-data as a named volume so Docker can store and reuse MySQL data even if the container is recreated.</li></ul><h4><strong>Step 3: Create The Folders </strong>mysql-data<strong> And </strong>mysql-init</h4><p>Next, let’s create two folders in the mysqlconf directory (if these folders already exist, you can skip this step):</p><pre>mkdir mysql-data<br>mkdir mysql-init</pre><ul><li>📁 mysql-data – This folder will be used by Docker to store MySQL data persistently. Even if you stop or recreate the container, your database files will remain intact.</li><li>📁 mysql-init – This folder will contain .sql initialization scripts. These scripts will automatically run the first time the MySQL container starts, allowing you to create databases, users, and grant permissions without manual steps.</li></ul><p>We’ll add the necessary SQL files in the next step.</p><h4>Step 4: Add SQL Scripts To The mysql-init Folder</h4><p>In this step, we’ll set up the required MySQL databases and users by following the official <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/setting-up-databases/changing-default-databases/changing-to-mysql/#setting-up-the-database-and-users">WSO2 API Manager documentation</a> 👈.</p><p>But instead of executing each command manually, we’ll automate the process by creating SQL scripts.</p><p>Let&#39;s create the following files inside the mysql-init directory:</p><p>01-setup.sql:The script below creates the initial users and grants them the necessary privileges for the two main databases used by WSO2 API Manager: apim_db and shared_db.</p><ul><li>First, it creates the apim_db database with the character setlatin1. Latin1 charset is recommended by WSO2 for these databases for compatibility reasons.</li><li>The USE apim_db; statement switches the context to the apim_db database, so subsequent commands (like creating users and granting privileges) apply specifically to this database.</li><li>It then creates the user apimadmin (if it doesn’t already exist) and grants this user all privileges on the apim_db database.</li><li>FLUSH PRIVILEGES; reloads the grant tables, ensuring that the privilege changes take effect immediately.</li><li>The same process is repeated for the shared_db database and its corresponding user sharedadmin.</li></ul><pre>CREATE DATABASE IF NOT EXISTS apim_db CHARACTER SET latin1;<br>USE apim_db;<br>CREATE USER IF NOT EXISTS &#39;apimadmin&#39;@&#39;%&#39; IDENTIFIED BY &#39;apimadmin&#39;;<br>GRANT ALL PRIVILEGES ON apim_db.* TO &#39;apimadmin&#39;@&#39;%&#39;;<br>FLUSH PRIVILEGES;<br><br>CREATE DATABASE IF NOT EXISTS shared_db CHARACTER SET latin1;<br>USE shared_db;<br>CREATE USER IF NOT EXISTS &#39;sharedadmin&#39;@&#39;%&#39; IDENTIFIED BY &#39;sharedadmin&#39;;<br>GRANT ALL PRIVILEGES ON shared_db.* TO &#39;sharedadmin&#39;@&#39;%&#39;;<br>FLUSH PRIVILEGES;</pre><ul><li>02-apimdb-mysql.sql: This script creates the required tables and schemas inside the apim_dbdatabase.</li><li>03-shareddb-mysql.sql: This script creates the required tables and schemas inside the shared_dbdatabase.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/942/1*m1VEwVxxexqIFx_8XxKiKg.png" /><figcaption>Terminal: Listing of files inside the mysql-init folder</figcaption></figure><p>When the MySQL container starts for the first time, it automatically runs these scripts in alphanumeric order (which is why we prefixed them with numbers) to set up everything needed for WSO2 API Manager to work properly.</p><p>These scripts are adapted from the official WSO2 tutorial for setting up MySQL databases: 👉 <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/distributed-deployment/configure-databases/working-with-databases-mysql/#setting-up-the-database-and-users">WSO2 APIM Docs: Changing to MySQL</a>. Since the scripts aren’t readily available for download, we’ll use a workaround to create and add them to the mysql-init folder.😅 Let’s dive into that in the next section!</p><h4><strong>Step 5: Extract the Database Scripts From The API Manager Container</strong></h4><blockquote>If you need a detailed walkthrough on how to set up WSO2 API Manager with Docker, feel free to check out my earlier tutorial:<br> 👉 <a href="https://medium.com/@lianecafarate/deploying-wso2-api-manager-4-5-with-docker-319c86303b4b">Deploying WSO2 API Manager 4.5 with Docker</a></blockquote><p>Since I already have the WSO2 API Manager Docker image pulled locally, I’ll go ahead and start the container directly using the command below:</p><pre>docker run -it -p 8280:8280 -p 8243:8243 -p 9443:9443 --name wso2apim wso2/wso2am:latest</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WTdH78Bb6ZhbHXhSLPrMgA.png" /><figcaption>Terminal: Running the APIM 4.5 container</figcaption></figure><p>Once the WSO2 API Manager container is up and running, we need to copy the database initialization scripts so we can run them on the MySQL databases we created earlier.</p><p>These scripts are located in the (&lt;APIM_HOME&gt;/dbscripts) directory inside the container. Here’s how to find the correct path and copy the files to your local machine:</p><p><strong>Step 5.1: Check The Path Inside The Container</strong></p><p>Start by accessing the running container:</p><pre>docker exec -it wso2apim bash</pre><p>Once inside, navigate to the dbscripts folder:</p><pre>cd wso2am-4.5.0/dbscripts<br>pwd</pre><p>You should see the following output:</p><pre>/home/wso2carbon/wso2am-4.5.0/dbscripts</pre><p>This confirms the full path to the database scripts inside the container.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o7Bi6YF6e2R_HNSqMKgJmw.png" /><figcaption>Terminal: Path to database scripts</figcaption></figure><p><strong>Step 5.2: Copy The </strong>dbscripts<strong> Folder To Your Local Machine</strong></p><p>Use the following command to copy the SQL script files from the running wso2apim container to your local machine:</p><pre>docker cp wso2apim:/home/wso2carbon/wso2am-4.5.0/dbscripts ./dbscripts</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*axNheBOGsVDN-ZyjjSBY4g.png" /><figcaption>Terminal: Scripts copied to local machine</figcaption></figure><p>This will create a dbscripts folder in your current working directory containing all the necessary SQL files for various databases, including:</p><ul><li>mysql.sql — for the Shared database (shared_db), which we&#39;ll rename to 03-shareddb-mysql.sql</li><li>apimgt/mysql.sql — for the API Manager database (apim_db), which we&#39;ll rename to 02-apimdb-mysql.sql</li></ul><p>Next, run these commands to rename and move the files into the mysql-init folder:</p><pre>mv dbscripts/mysql.sql mysql-init/03-shareddb-mysql.sql<br>mv dbscripts/apimgt/mysql.sql mysql-init/02-apimdb-mysql.sql</pre><p>Once you’ve done this, you can safely delete the original dbscripts folder with:</p><pre>rm -rf dbscripts</pre><h4><strong>Step 6: Edit The SQL Scripts To Specify The Database Context</strong></h4><p>For the 02-apimdb-mysql.sql script, add the following line at the very beginning of the file:</p><pre>USE apim_db;</pre><p>Similarly, for the 03-shareddb-mysql.sql script, add this line at the top:</p><pre>USE shared_db;</pre><p>Without specifying the database in the SQL scripts (using the USE &lt;database&gt;; statement), you will encounter this error when running the scripts:</p><pre>ERROR 1046 (3D000): No database selected</pre><p>Below are examples of the updated 02-apimdb-mysql.sql and 03-shareddb-mysql.sql files with the added USE statements at the top.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/797/1*2OQiK3DehEnqxQwC4HNm4Q.png" /><figcaption>VSCode: apimdb-mysql.sql</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/950/1*xO5lrcGBvhvbM7DmAeCJ6w.png" /><figcaption>VSCode: shareddb-mysql.sql</figcaption></figure><p>We’re almost done setting up the folder structure! What’s left are the deployment.toml file and the MySQL driver, both required for the API Manager service. We’ll come back to those in a bit.</p><p>👉 <strong>Quick note:</strong> If you already started the wso2apim container earlier, make sure to stop it for now. You can do that by running:</p><pre>docker stop wso2apim</pre><p>For now, let’s test if the MySQL container starts correctly and automatically runs the initialization scripts, creating the databases, users, and tables as defined.</p><h4>Step 7: Start The MySQL Container</h4><p>Run the following command:</p><pre>docker-compose up --build -d</pre><p>Here’s what this command does:</p><ul><li>docker-compose – Uses the docker-compose.yml file in your current directory.</li><li>up – Starts the services defined in the file.</li><li>--build – (Re)builds the images before starting the containers. This is useful if you&#39;ve made changes to the configuration or Dockerfile.</li><li>-d – Runs the containers in detached mode (in the background), so your terminal remains available for other commands.</li></ul><p>If everything is configured correctly, Docker will spin up the MySQL container, mount the volumes, and automatically run the SQL scripts inside the mysql-init folder.</p><p>✅ To verify that the MySQL container is running, use the command below:</p><pre>docker ps</pre><p>This will list all active containers and show details like container name, status, and ports.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cs3uJaZbc1FaJWHZ7M-rOA.png" /><figcaption>Terminal: MySQL container running</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9ROvhfyGNdjHG1knCYhmDg.png" /><figcaption>Rancher Desktop: MySQL container running</figcaption></figure><h4>Step 8: Verify That MySQL Was Set Up Correctly</h4><p>Let’s check if the databases, users, and tables were created as expected.</p><p><strong>Step 8.1: Access the running MySQL container</strong></p><pre>docker exec -it mysql-server mysql -u apimadmin -p</pre><p>When prompted, enter the MySQL root password (in our case: apimadmin).</p><p><strong>Step 8.2: Switch to apimdb databases and inspect the tables</strong>:</p><pre>USE apim_db;<br>SHOW TABLES;</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/315/1*A50oJXXW1ifUuvqmp5JTmg.png" /><figcaption>MySQL: apim_db tables</figcaption></figure><p>Repeat the same steps for the shared_db database using the sharedadmin user and password.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/573/1*mnDZpLLWoha9idLf7MBgqA.png" /><figcaption>MySQL: shared_db tables</figcaption></figure><p>✅ If the script ran correctly, you should see a list of tables like:</p><ul><li>AM_API</li><li>AM_SUBSCRIPTION</li><li>AM_APPLICATION</li><li>and many more…</li></ul><p>For shared_db, you might see tables like:</p><ul><li>REG_RESOURCE</li><li>UM_USER</li><li>UM_ROLE</li><li>etc.</li></ul><p>When you are done, you can exit the MySQL CLI.</p><pre>exit;</pre><p>And then:</p><pre>exit</pre><p>To leave the container shell.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/246/1*idV2IeIcH3uOq2d1GMeSfg.png" /><figcaption>Terminal: Exit</figcaption></figure><h4>Step 9: Check Logs and Clean Up</h4><p>Once you’ve confirmed the databases and users were created successfully, you can take the following optional steps:</p><p>📝 <strong>View the container logs</strong><br> To monitor the MySQL container logs in real-time:</p><pre>docker logs -f mysql-server</pre><p>This is helpful to double-check that the initialization scripts were executed without issues.</p><p>🧹 <strong>Clean up containers and volumes</strong><br> If you want to stop and remove all containers, networks, and <strong>volumes</strong> created by Docker Compose (including database data), run:</p><pre>docker-compose down -v</pre><blockquote><em>⚠️ </em><strong>Note:</strong> The -v flag tells Docker to remove the named volumes as well , in your case, the volume mysql-data where your MySQL data is persisted.</blockquote><h4>💡 (Optional) <strong>Access The MySQL Container Shell For Direct Interaction With The Database</strong></h4><pre>docker exec -it mysql-server bash</pre><p>This opens a shell inside the running container, allowing you to run MySQL commands or inspect logs.</p><p>Once inside the container shell, you can access the MySQL command-line interface (CLI) using the following command:</p><pre>mysql -u root -p</pre><p>You’ll be prompted to enter the root password you defined earlier (in this case, rootpass). After a successful login, you’ll see the MySQL welcome message and enter an interactive session where you can run SQL queries and manage your database.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/567/1*2vMPEw2ImRkbDFFjaoZ3dQ.png" /><figcaption>Terminal: MySQL connection</figcaption></figure><blockquote><em>⚠️ </em><strong>Quick Note:</strong> Earlier, we mentioned that before adding the API Manager service to our Docker Compose file, we need to prepare two key things: copying the MySQL driver and updating the deployment.toml configuration file. Now, we’ll start with copying the MySQL driver to our project folder. After that, we’ll move on to editing the deployment.toml file.</blockquote><h4>Step 10: Copy The MySQL Driver To Your Project</h4><p>To connect WSO2 API Manager to MySQL, we need the MySQL JDBC driver in our project. Let’s copy the driver JAR file into the drivers folder of our project.</p><p>If you haven’t created it yet, run this command to make the drivers folder inside your mysqlconf directory:</p><pre>mkdir drivers</pre><p>Refer to Step 3 of the Prerequisites section to download the MySQL Connector/J (JDBC driver). This driver enables WSO2 API Manager to communicate with your MySQL databases.</p><p>Make sure to download the platform-independent version and extract the .jar file (e.g., mysql-connector-j-8.4.0.jar) from the archive.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/404/1*xzO7QKxCuvdGE4tVQRRSwg.png" /><figcaption><strong>Terminal:</strong> Contents of the drivers folder showing the MySQL Connector JAR file</figcaption></figure><h4>Step 11: Edit The deployment.toml File To Use MySQL</h4><p>Next, we need to update the deployment.toml configuration file for WSO2 API Manager to connect it to our MySQL database.</p><p>Since the default deployment.toml lives inside the running wso2apim container, we&#39;ll copy it to our local machine just like we did with the SQL scripts.</p><p>If your WSO2 APIM container isn’t running yet, simply start it with this command:</p><pre>docker start wso2apim</pre><p>This will get your API Manager container up and running so you can proceed with the next steps.</p><p>Now, run this command to copy it to your project folder:</p><pre>docker cp wso2apim:/home/wso2carbon/wso2am-4.5.0/repository/conf/deployment.toml ./deployment.toml</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eNHwkkGQO-Ua27EuE9q_6Q.png" /><figcaption>Terminal: Copying the deployment.toml file from the APIM container into the local machine</figcaption></figure><p>Next, let’s edit the file locally. Open thedeployment.toml file with your favorite text editor (e.g., VSCode, etc.) and add the following at the end (or update if the section already exists):</p><pre>[database.apim_db]<br>type = &quot;mysql&quot;<br>url = &quot;jdbc:mysql://mysql-server:3306/apim_db?allowPublicKeyRetrieval=true&amp;amp;useSSL=false&quot;<br>username = &quot;apimadmin&quot;<br>password = &quot;apimadmin&quot;<br>driver=&quot;com.mysql.cj.jdbc.Driver&quot;<br><br>[database.shared_db]<br>type = &quot;mysql&quot;<br>url = &quot;jdbc:mysql://mysql-server:3306/shared_db?allowPublicKeyRetrieval=true&amp;amp;useSSL=false&quot;<br>username = &quot;sharedadmin&quot;<br>password = &quot;sharedadmin&quot;<br>driver=&quot;com.mysql.cj.jdbc.Driver&quot;</pre><p>Let&#39;s break down the configuration:</p><ul><li><strong>[database.apim_db]</strong> and <strong>[database.shared_db]: </strong>These sections define two separate datasources: one for the API Manager database (apim_db) and another for the shared database (shared_db).</li><li><strong>type = “mysql”: </strong>Specifies that the datasource uses a MySQL database.</li><li><strong>url: </strong>This is the JDBC connection string that tells WSO2 APIM how to connect to the MySQL server.</li><li>mysql-server refers to the hostname of the MySQL container/service (as defined in docker-compose.yml).</li><li>3306 is the default MySQL port.</li><li>apim_db and shared_db are the database names.</li><li>allowPublicKeyRetrieval=true enables the client to retrieve the public key from the server for authentication (required for some MySQL versions).</li><li>useSSL=false disables SSL connection for simplicity.</li><li><strong>username</strong> and <strong>password: </strong>Credentials used to connect to the respective databases. These must match the users created in your SQL scripts (apimadmin and sharedadmin).</li><li><strong>driver = “com.mysql.cj.jdbc.Driver”: </strong>Specifies the MySQL JDBC driver class that WSO2 APIM uses to connect to the database.</li></ul><blockquote><em>⚠️</em> <strong>Quick Note:</strong> To avoid <strong>“SSL”</strong> and <strong>“Public Key Retrieval”</strong> related errors on server startup we have added the following values to the url configuration. This is not recommended for Production environments for security reasons. Also note that if you are adding both allowPublicKeyRetrieval=true and useSSL=false to the url, it must be appended with <strong>“&amp;amp”.</strong></blockquote><p>📚 For more details, check the official WSO2 documentation here:<br>🔗 <a href="https://apim.docs.wso2.com/en/latest/install-and-setup/setup/setting-up-databases/changing-default-databases/changing-to-mysql/#creating-the-datasource-connection-to-mysql">Setting Up Datasource Connection to MySQL</a></p><blockquote><em>⚠️ </em><strong>P.S.1: </strong>Don’t forget to <strong>save</strong> your changes to the deployment.toml file after editing it! <em>💾</em></blockquote><blockquote><em>✨ </em><strong>P.S.2: </strong>To stop the WSO2 API Manager container for now, run docker stop wso2apim. We&#39;ll be using Docker Compose to start all services together shortly.</blockquote><h4>Step 12: Add the WSO2 API Manager Service to Docker Compose</h4><p>Now that MySQL is configured and working, let’s add the <strong>WSO2 API Manager (wso2apim)</strong> service to our docker-compose.yml. This will allow the API Manager to run alongside MySQL and connect to it seamlessly.</p><pre>services:<br>  mysql:<br>      image: mysql:8.4<br>      container_name: mysql-server<br>      restart: unless-stopped<br>      environment:<br>        MYSQL_ROOT_PASSWORD: rootpass<br>      ports:<br>        - &quot;3306:3306&quot;<br>      volumes:<br>        - mysql-data:/var/lib/mysql<br>        - ./mysql-init:/docker-entrypoint-initdb.d<br>      healthcheck:<br>        test: [&quot;CMD&quot;, &quot;mysqladmin&quot;, &quot;ping&quot;, &quot;-uroot&quot;, &quot;-prootpass&quot;]<br>        interval: 10s<br>        timeout: 20s<br>        retries: 5<br><br>  wso2apim:<br>    image: wso2/wso2am:latest<br>    container_name: wso2apim<br>    depends_on:<br>      mysql:<br>        condition: service_healthy<br>    ports:<br>      - &quot;8280:8280&quot;<br>      - &quot;8243:8243&quot;<br>      - &quot;9443:9443&quot;<br>    volumes:<br>      - ./deployment.toml:/home/wso2carbon/wso2am-4.5.0/repository/conf/deployment.toml:ro<br>      - ./drivers/mysql-connector-j-8.4.0.jar:/home/wso2carbon/wso2am-4.5.0/repository/components/lib/mysql-connector-j-8.4.0.jar:ro<br>    restart: unless-stopped<br><br>volumes:<br>  mysql-data:<br>    driver: local</pre><p>Here’s what the new service does:</p><ul><li><strong>Image:</strong> Uses the official WSO2 API Manager Docker image (wso2/wso2am:latest).</li><li><strong>Container Name:</strong> Names the container wso2apim for easy reference.</li><li><strong>Depends On:</strong> Waits for the MySQL container to be healthy before starting, ensuring the database is ready.</li><li><strong>Ports:</strong> Exposes API Manager’s key ports for HTTP (8280), HTTPS (8243), and the management console (9443).</li><li><strong>Volumes: </strong>1)<strong> </strong>Mounts the customized deployment.toml configuration file to tell API Manager how to connect to MySQL; 2) Mounts the MySQL JDBC driver (mysql-connector-j-8.4.0.jar) so API Manager can communicate with the MySQL database.</li><li><strong>Restart Policy:</strong> Automatically restarts the container unless it is manually stopped.</li></ul><h4>Step 13: Run Docker Compose with Both Services</h4><p>Now that we have configured the Docker Compose file with both <strong>MySQL</strong> and <strong>WSO2 API Manager</strong> services, it’s time to start them together.</p><p>Run the following command in your mysqlconf directory:</p><pre>docker-compose up --build -d</pre><p>The MySQL container will start first, and once its health check passes, Docker Compose will start the WSO2 API Manager container, ensuring proper service dependency management.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/329/1*8qwIycDKVxNGRNYT4hcgSQ.png" /><figcaption>Terminal: Docker Compose Command</figcaption></figure><p>You can verify both containers are running with:</p><pre>docker ps</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Gewsu1W3iXzOAmdIDT8vPw.png" /><figcaption>Rancher Desktop: mysql-server and wso2apim containers running</figcaption></figure><h4>Step 14: Check If The WSO2 API Manager Started Correctly</h4><p>You can check if the wso2apim container started correctly by running the following command in your terminal:</p><pre>docker logs -f wso2apim</pre><p>The -f flag in the command docker logs -f &lt;container_name&gt; stands for <strong>&quot;follow&quot;</strong>.</p><p>It means the command will show you the container’s logs <strong>in real time</strong>, streaming new log output as it happens, similar to tail -f for log files. This way, you can keep the terminal open and watch live updates from the container’s logs until you stop it (usually with Ctrl+C).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wvk4bY3RmeTIm4yVPzoPLA.png" /><figcaption>Terminal: docker logs -f command for the wso2apim container</figcaption></figure><p>Look for these lines in the logs to find the URLs for the WSO2 API Manager consoles:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/979/1*xtzyr5dtQEVoDothf8sSjw.png" /><figcaption>Terminal: APIM Console Links</figcaption></figure><blockquote><em>🔍</em> If you want to dive deeper into these portals, check out the <a href="https://apim.docs.wso2.com/en/latest/get-started/apim-architecture/#api-publisher">WSO2 official documentation</a> for more details.</blockquote><p>To access a portal, just open your browser and enter the URL. For example, the Publisher portal is:</p><pre>https://localhost:9443/publisher</pre><p>And voilà! You should be connected and ready to explore. 😊</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MYmQ-waNRefm2GujtpeT5g.png" /><figcaption>Webbrowser: WSO2 APIM Publisher Portal</figcaption></figure><h3>🎬 Let’s Close This Out</h3><p>Still here? WOW, kudos to you! 🎉 This was quite a journey, and I ran into plenty of challenges along the way, but it finally worked!</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ftenor.com%2Fembed%2F14701559&amp;display_name=Tenor&amp;url=https%3A%2F%2Ftenor.com%2Fview%2Fyatta-yes-happy-hiro-nakamura-gif-14701559&amp;image=https%3A%2F%2Fmedia.tenor.com%2FiSeJB4xRd6QAAAAN%2Fyatta-yes.png&amp;type=text%2Fhtml&amp;schema=tenor" width="600" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/a5c33583966748ef8d44e54cb7622f16/href">https://medium.com/media/a5c33583966748ef8d44e54cb7622f16/href</a></iframe><p>🙌 Many thanks for sticking with me through it all.❤️</p><p>If you found this article helpful, please <strong>give it a clap</strong> 👏. It really motivates me to create more content like this. Got questions, feedback, or your own tips? Drop a comment below. I’d love to hear from you 👂!</p><p>And don’t forget to <strong>share</strong> this article with your network if you think others could benefit.</p><p>Keep rocking those APIs! 🚀</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=33b275d568a3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ Tend, Befriend, and Thrive: A Celebration of Women Supporting Women]]></title>
            <link>https://lianecafarate.medium.com/tend-befriend-and-thrive-a-celebration-of-women-supporting-women-1d1376d7c9e7?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/1d1376d7c9e7</guid>
            <category><![CDATA[women-supporting-women]]></category>
            <category><![CDATA[life-abroad]]></category>
            <category><![CDATA[women-in-tech]]></category>
            <category><![CDATA[brazilian]]></category>
            <category><![CDATA[community]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Fri, 04 Jul 2025 20:08:45 GMT</pubDate>
            <atom:updated>2025-07-04T20:19:09.661Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RiB_tUcsHWXy-9KKL5g23A.jpeg" /><figcaption>Connect Club Fest — 06/27/2025 in Mountain View</figcaption></figure><p>I’m part of <a href="https://www.estefaniabarsante.com/connect-club-listadeespera">Connect Club</a>, a community of Brazilian women living abroad focused on personal development and connection. Our members come from diverse backgrounds and live in different regions, but we share a desire to grow and support one another.</p><p>We do all kinds of activities, some focused on growth (like workshops and speaker sessions), and others centered around connection (like our book club, brunches, and exercise meetups).</p><p>Last week, we hosted our very first Connect Club Fest: an afternoon of celebration, inspiration, and deep connection among women living in the U.S. Here are my main takeaways:</p><h3>1. 💡 Smart Women Support Women</h3><p>We began with an inspiring talk by our founder, <a href="https://www.linkedin.com/in/estefaniabarsante/">Estefania Barsante</a>, who spoke about the true meaning of community. One idea that really stuck with me is this:</p><blockquote>In stressful situations, women don’t typically follow the fight-or-flight model. Instead, we “tend and befriend.”</blockquote><p>This concept, backed by research, suggests that while both men and women experience the physiological symptoms of stress, women are more likely to respond behaviorally by nurturing and building social bonds. “Tending” means caring for others (and ourselves); “befriending” means forming social networks that offer mutual support and protection [1].</p><p>This is powerful. Society often teaches us that women compete and tear each other down, but that’s a <strong>myth rooted in patriarchy</strong>.</p><p>From a young age, girls are socialized to compare themselves, especially around appearance or popularity. These messages, reinforced by media and advertising, aren’t natural but learned.</p><p>Yet the reality is clear: <strong>Women supporting women leads to better outcomes</strong>. Here’s what research shows:</p><ul><li>📈 A 2019 <em>Harvard Business Review</em> study found that women who build strong networks with other women are more likely to advance in their careers [2].</li><li>🤝 Women-led mentorship programs outperform others in retention and satisfaction [3].</li><li>👩‍💼 Teams with strong female relationships perform better in collaboration, communication, and psychological safety, a critical thing for high-performing teams [4].</li></ul><p>So next time you hear “women are too competitive,” remember: <strong>that’s a myth.</strong> We’re stronger when we collaborate. 💪</p><h3>2. 🔍 Be Passionately Curious</h3><p>A few weeks ago, I was feeling overwhelmed at work (I wrote about it <a href="https://lianecafarate.medium.com/working-in-tech-and-feeling-overwhelmed-youre-not-alone-90b152eb691a">here</a>). What helped me bounce back was reconnecting with what energizes me, like staying curious and always learning. 📚</p><p>Little did I know, there’s actual research behind this. At Connect Club Fest, <a href="https://www.linkedin.com/in/patriciaeast/">Patricia East</a> beautifully reinforced the idea. She shared how it wasn’t perfection or innate talent that propelled her forward; it was her relentless curiosity.</p><blockquote>&quot;I have no special talent. I am only passionately curious&quot;. Albert Einsten</blockquote><p>Patricia also introduced us to the concept of the <strong>Mood Elevator</strong>, created by Larry Senn.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/498/1*_uGMOYLO5zUqXUJ25qLByg.png" /><figcaption>The Mood Elevator — Larry Senn</figcaption></figure><h4>🛗 The Mood Elevator</h4><p>Imagine your mood as an elevator moving up and down throughout the day:</p><ul><li><strong>⬆️ Top floors</strong>: gratitude, creativity, joy, patience (positive and productive states)</li><li><strong>⬇️ Lower floors</strong>: stress, anger, frustration, self-doubt (reactive states)</li></ul><p>✅ The goal isn’t to avoid bad moods but to <strong>recognize</strong> them and <strong>consciously shift</strong> upward.</p><h4>🎯 Curiosity helps us elevate</h4><ol><li>🔄 <strong>It breaks negative cycles:</strong> Instead of spiraling, curiosity makes us pause and reflect: “Why am I feeling this way?”</li><li>💭 <strong>It fosters empathy</strong>: Rather than judging, we ask, “What might be going on with them?”</li><li>🚀 <strong>It inspires growth:</strong> Curious people seek new ideas, perspectives, and solutions.</li></ol><p>Patricia also shared grounding techniques to return to a balanced emotional state. Her final message was unforgettable:</p><blockquote><em>“Everyone has the capacity to ask the questions no one else is asking.” 🌟</em></blockquote><blockquote>📘 <strong>Hot Tip:</strong> Patricia East is the author of <a href="https://www.amazon.com/Passport-Growth-Leadership-Lessons-Expat/dp/B0D98589R5"><em>Passport to Growth</em></a>, a powerful book where she shares her life-changing journey of self-discovery. Highly recommended , check it out!</blockquote><h3>3. ✈️ Panel: Women Abroad: Challenges and Learning</h3><p>I had the honor of joining <a href="https://www.linkedin.com/in/jessicastelmach/"><strong>Jessica Stelmach</strong></a> and <strong>Keren Hapuk </strong>on a panel about life abroad. Here are two big reflections:</p><h4>🌍 It’s hard to see the picture when you’re in the frame</h4><p>One question we discussed was: “What changed the most about you while living abroad?”</p><p>I remembered an experience from when I lived in Germany. We were deciding whether to attend a party but didn’t buy tickets early. When the day came, it was quiet due to a holiday weekend, so we tried to negotiate the price at the door. The German hosts said that it wouldn’t be fair to those who had already paid full price.</p><p>That moment, and many others, made me think about cultural differences and how Brazilians tend to value flexibility. Living abroad showed me the strengths and challenges of our culture and made me more open-minded.</p><h4>🎮 Playing life on hard mode</h4><p>Another question: “What have you learned about yourself?”</p><p>Living abroad is like playing a video game on hard mode. You’re probably not speaking your native language. You have to learn about new systems, healthcare, retirement, laws, and even grocery brands.</p><p>It’s exhausting sometimes. But it also <strong>accelerates growth</strong>. Learning how to move through these challenges with grace is an art. 🎨 And it’s worth it.</p><h3>4. 💪 Supported Women Are Stronger</h3><p>Leaving your comfort zone forces you to grow. I know I wouldn’t be the person I am today without these experiences.</p><p>But more than anything, I’m grateful for the women I’ve met on this journey. They’ve lifted me, shaped me, and reminded me of my strength. 💞</p><p>That’s why <strong>community matters</strong>. That’s why we say: <strong>We are stronger together.</strong> 🤝</p><h3>💬 Want to Join Us?</h3><p>If you’re curious about <strong>Connect Club</strong> and want to learn more, schedule a quick call here: <a href="https://calendly.com/estefaniabarsante/connect-club-interessadas?month=2025-07">Calendly link</a>.</p><p>Let’s grow together. 🌱</p><h3>References</h3><p>[1] <a href="https://scholar.harvard.edu/files/marianabockarova/files/tend-and-befriend.pdf">Biobehavioral Responses to Stress in Females:<br>Tend-and-Befriend, Not Fight-or-Flight</a></p><p>[2] <a href="https://hbr.org/2019/11/the-secrets-of-successful-female-networkers">The Secrets of Successful Female Networkers</a></p><p>[3] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC6997818">Successful mentoring of women</a></p><p>[4] <a href="https://www.glamour.com/story/woman-to-woman-mentoring-stem-subjects-study">Woman-to-Woman Mentoring May Keep Women From Dropping Out of STEM Subjects, Study Says</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1d1376d7c9e7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ Configuring Choreo Analytics on WSO2 API Manager 4.5 with Docker]]></title>
            <link>https://lianecafarate.medium.com/configuring-choreo-analytics-on-wso2-api-manager-4-5-with-docker-53241881447f?source=rss-918acad7f898------2</link>
            <guid isPermaLink="false">https://medium.com/p/53241881447f</guid>
            <category><![CDATA[wso2]]></category>
            <category><![CDATA[wso2-api-manager]]></category>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[api]]></category>
            <dc:creator><![CDATA[Liane Cafarate]]></dc:creator>
            <pubDate>Wed, 25 Jun 2025 18:43:02 GMT</pubDate>
            <atom:updated>2025-06-25T18:46:20.414Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*z441QBf6ly_DANrl" /><figcaption>Photo by <a href="https://unsplash.com/@dawson2406?utm_source=medium&amp;utm_medium=referral">Stephen Dawson</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>In my <a href="https://lianecafarate.medium.com/deploying-wso2-api-manager-4-5-with-docker-319c86303b4b">previous tutorial</a>, I showed how to deploy WSO2 API Manager 4.5 using a single docker run command.</p><pre>docker run -it -p 8280:8280 -p 8243:8243 -p 9443:9443 - name wso2apim wso2/wso2am:latest</pre><p>That was great for quick experimentation. But what if we need to customize the configuration, enable analytics, or scale to multiple containers?</p><p>That’s where <strong>Docker Compose</strong> shines.</p><h3>🐳 Why Docker Compose?</h3><blockquote><em>💡</em> Experienced users may wish to skip this section.</blockquote><p>Instead of typing long and repetitive docker run commands, Docker Compose lets us define our setup in a readable YAML file (docker-compose.yml). We can configure ports, volumes, and even link multiple services together.</p><h4>Example:</h4><p>Instead of this:</p><pre>docker run -d -p 8280:8280 --name service1 image1<br>docker run -d -p 9443:9443 --name service2 image2</pre><p>We can define both in one file:</p><pre>services:<br>  service1:<br>    image: image1<br>    ports:<br>      - &quot;8280:8280&quot;<br>  service2:<br>    image: image2<br>    ports:<br>      - &quot;9443:9443&quot;</pre><p>And start everything with:</p><pre>docker-compose up -d</pre><p>Much cleaner, right?</p><blockquote><em>💡</em> <strong>Hint:</strong> Flag -d, — detach. Detached mode: Run containers in the background</blockquote><h3>📊 What is Choreo Analytics?</h3><p><a href="https://apim.docs.wso2.com/en/latest/monitoring/api-analytics/choreo-analytics/getting-started-guide/">Choreo Analytics</a> is WSO2’s cloud-native API monitoring and insights solution, available out of the box for WSO2 API Manager.</p><p>It allows us to monitor API traffic, latency, and subscriptions in real time, all without hosting our own analytics stack.</p><h3>⚠️ For Testing &amp; Demo Purposes Only</h3><blockquote><em>This configuration is intended strictly for local testing, experimentation, and demonstration. It is NOT secured, optimized, or configured for production environments.</em></blockquote><h3>🛠 Step-by-Step: Enable Choreo Analytics on Dockerized API Manager 4.5</h3><h4>Extract the Default Configuration (deployment.toml file) from the Running Container</h4><blockquote><em>💡</em> A quick note: the deployment.toml file, located at &lt;API-M_HOME&gt;/repository/conf/deployment.toml, serves as the central configuration file for WSO2 API Manager. It acts as the single source of truth for enabling, tuning, and managing all key features of the platform.</blockquote><p>By providing a custom deployment.toml, we&#39;ll override the default configuration inside the Docker container, allowing us to enable Choreo Analytics and customize other settings as needed.</p><p>To ensure everything works smoothly, I copied the contents of the existing deployment.toml from my running container. Here’s how I did it:</p><p>🔹 <strong>Step 1: Identify the Running Container</strong></p><p>First, let’s confirm the name of the container currently running.</p><blockquote><em>💡</em> If you’re not sure, remember you need to have the container up and running (refer to the <a href="https://lianecafarate.medium.com/deploying-wso2-api-manager-4-5-with-docker-319c86303b4b">previous tutorial</a> if needed).</blockquote><pre>docker ps</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RsBeDSY9F2veFzF8olT3Fg.png" /><figcaption>Terminal: Running containers</figcaption></figure><p>This lists all running containers. In my case, the container name is wso2apim.</p><p>🔹 <strong>Step 2: Locate the Default </strong><strong>deployment.toml Path (Running Container)</strong></p><p>Now, let’s check the contents of the deployment.toml file. This file is located at &lt;API-M_HOME&gt;/repository/conf/.</p><p>To access it inside the running container, we’ll use the docker exec command, which lets us run commands interactively within the container.</p><p>Run the following to open a bash shell inside the container:</p><pre>docker exec -it wso2apim bash</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zVx3lYFl0PcDf6HTfYfbDw.png" /><figcaption>Terminal: Docker Exec</figcaption></figure><p>Let’s confirm the full absolute path of the deployment.toml file <strong>inside the running container</strong>, since this exact path will be important when we mount the file using Docker Compose.</p><p>When we run the following command pwdinside the container, we’ll see the directory, we’ll see the directory:/home/wso2carbon/wso2am-4.5.0/repository/conf.</p><p>Later, when configuring Docker Compose, we will map our local custom deployment.toml file to this exact path inside the container to override the default configuration.</p><p>🔹 <strong>Step 3: View and Copy the File Contents</strong></p><p>Let’s open the deployment.toml file and copy its contents using the following command:</p><pre>cat deployment.toml</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/540/1*6uaoYGAXkssIOPCgASbTJQ.png" /><figcaption>Terminal: deployment.toml</figcaption></figure><p>🔹 <strong>Step 4: Create a Local Version of the File</strong></p><p>We’ll organize everything in a directory (mine is called analyticsconf).</p><pre>mkdir analyticsconf<br>cd analyticsconf</pre><p>Now, let&#39;s create a local version of the deployment.toml and paste the copied content into the file created inside the conf folder (in my case, the project folder is named analyticsconf).</p><pre>touch deployment.toml</pre><p><strong>🔧 </strong>Your final file will be much longer, but it should start with something like this:</p><pre>[server]<br>hostname = &quot;localhost&quot;<br>#offset=0<br>base_path = &quot;${carbon.protocol}://${carbon.host}:${carbon.management.port}&quot;<br>#discard_empty_caches = false<br>server_role = &quot;default&quot;<br>...</pre><p>This is just a snippet. The actual deployment.toml contains many more settings that you can review or modify based on your needs.</p><blockquote><em>👉</em> <strong>Want to dive deeper?</strong> Check out the<a href="https://apim.docs.wso2.com/en/latest/reference/understanding-the-new-configuration-model/"> official WSO2 documentation on the configuration model</a>.</blockquote><h4>Create docker-compose.yml and Map the Configuration File</h4><p>Now that we have our custom deployment.toml file, let’s define the container setup in a docker-compose.yml file:</p><pre>touch docker-compose.yml</pre><pre>services:<br>  wso2apim:<br>    image: wso2/wso2am:latest<br>    container_name: wso2apim<br>    ports:<br>      - &quot;8280:8280&quot;<br>      - &quot;8243:8243&quot;<br>      - &quot;9443:9443&quot;<br>    volumes:<br>      - ./deployment.toml:/home/wso2carbon/wso2am-4.5.0/repository/conf/deployment.toml<br>    restart: unless-stopped</pre><p><strong>🔍 Understanding the Volume Mapping</strong></p><p>Remember when we retrieved the full path to the deployment.toml file inside the running container? This step is crucial for <strong>volume mapping</strong>, which allows us to override a file inside the container with a file from our local machine.</p><p>Here’s the breakdown:</p><pre>volumes:<br> - ./deployment.toml:/home/wso2carbon/wso2am-4.5.0/repository/conf/deployment.toml</pre><ul><li>./deployment.toml: This is the <strong>source file on our local machine</strong> (relative to our project directory). We’ll be running the <em>docker-compose</em> command directly from the analyticsconf folder.</li><li>/home/wso2carbon/wso2am-4.5.0/repository/conf/deployment.toml: This is the <strong>destination inside the container</strong>, which replaces the default configuration file.</li></ul><h4>Add the Choreo Analytics Configuration to Local deployment.toml</h4><p>Before starting the container, we need to update our local deployment.toml file to enable Choreo Analytics.</p><p>This configuration tells WSO2 API Manager to send API event data to Choreo’s cloud-based analytics service.</p><p>Let&#39;s add the following section to the end of our deployment.toml file (the one inside the <em>analyticsconf</em> folder:</p><pre>[choreo.analytics]<br>enabled = true<br>config_endpoint = &quot;https://analytics-event-auth.choreo.dev/auth/v1&quot;<br>auth_token = &quot;&lt;your-auth-token-here&gt;&quot;</pre><blockquote><em>🔐</em> Replace &lt;your-auth-token-here&gt; with the actual token generated from the <a href="https://console.choreo.dev/">Choreo Console</a>. Follow <a href="https://apim.docs.wso2.com/en/latest/monitoring/api-analytics/choreo-analytics/getting-started-guide/">this tutorial </a>to generate your token.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fMzkL0EE-hpbcudqY1TVWw.png" /><figcaption>Choreo Console: Generate On-premises Key</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fpQmW1qiRSu3902t4hrZxw.jpeg" /><figcaption>Choreo Console: Copy On-premises Key</figcaption></figure><blockquote><em>💡 </em>If you don’t have a paid subscription, Choreo still allows you to generate up to <strong>3 access tokens</strong>, each valid for <strong>2 weeks</strong>, at no cost.</blockquote><blockquote><em>😉</em> While WSO2 API Manager also supports other analytics solutions like <strong>ELK</strong>, <strong>Datadog</strong>, and more, we’ll focus on Choreo Analytics for now. We can explore those alternative setups in future tutorials. <strong>Let me know in the comments if that’s something you’d be interested in!</strong><em>👇</em></blockquote><h4>Start the Container</h4><p>Run:</p><pre>docker-compose up -d</pre><p>Then follow the logs to make sure everything started:</p><pre>docker logs -f wso2apim</pre><p>Check for this log message to confirm that API Manager started successfully:</p><pre>WSO2 Carbon started in XX seconds</pre><p>This indicates that the server has initialized properly and is ready to use (luckily 😅).</p><h3>✅ Test If It Works</h3><p><strong>Publish a Sample API</strong></p><p>Follow <a href="https://apim.docs.wso2.com/en/latest/get-started/api-manager-quick-start-guide/#step-1-create-deploy-and-publish-an-api">WSO2’s Quick Start Guide</a>. Or, use the built-in <strong>Sample API</strong> option in the Publisher portal.</p><blockquote>By this point, your WSO2 API Manager portals should be up and running. If you’re unsure how to access them, please refer to the “Access Portals” section in the <a href="https://lianecafarate.medium.com/deploying-wso2-api-manager-4-5-with-docker-319c86303b4b">previous tutorial</a>.</blockquote><ol><li>Go to: <a href="https://localhost:9443/publisher">https://localhost:9443/publisher</a></li><li>Log in with default credentials: admin / admin</li><li>Click <strong>Deploy Sample API</strong></li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YZ4b-AdXGbWj5d-pgSXzRA.png" /><figcaption>Publisher Portal: Sample API</figcaption></figure><h4>Invoke the API</h4><ol><li>Go to the Developer Portal: <a href="https://localhost:9443/devportal">https://localhost:9443/devportal</a></li><li>Subscribe to the sample API using the <strong>Default Application</strong></li><li>Generate an access token</li><li>Invoke the API using curl or the built-in test tool as many times as you want</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tsib623zPxCFJ7pU1qiZ7g.png" /><figcaption>Developer Portal: API Invocation</figcaption></figure><h4><strong>Go to Choreo Analytics Dashboard</strong></h4><p>Open <a href="https://console.choreo.dev/">Choreo Console</a>, go to your organization/project → <strong>API Analytics</strong>, and boom 💥 , your API calls should show up!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EMaFUktlmhhCPkdWkXfBVQ.png" /><figcaption>Choreo Analytics Portal: Usage Insights</figcaption></figure><p>The Insights dashboard URL typically looks like this:</p><pre>https://console.choreo.dev/organizations/&lt;your-org-name&gt;/analytics/usage-insights/</pre><blockquote><em>📝 </em>Remember to replace &lt;your-org-name&gt; with your actual organization name.</blockquote><p>With Choreo insights, we can:</p><ul><li><strong>Analyze API traffic</strong>: Monitor the volume of requests and responses to understand usage patterns.</li><li><strong>Track errors</strong>: Identify and analyze errors to improve API reliability.</li><li><strong>Monitor latency</strong>: Measure and optimize the response times of our APIs.</li><li><strong>Generate reports</strong>: Generate detailed reports to keep stakeholders informed and support data-driven decision-making.</li><li><strong>Configure alerts</strong>: Set up alerts for specific events or thresholds to proactively manage API performance.</li><li><strong>Obtain granular insights</strong>: Obtain detailed information on a per-application, per-API, and per-user basis for more targeted analysis.</li><li><strong>Drill down into data</strong>: Dive deeper into data to perform root cause analysis.</li></ul><blockquote><em>🤓</em> Want to learn more? Check out the <a href="https://wso2.com/choreo/docs/monitoring-and-insights/insights-overview/">official documentation</a>!</blockquote><h3>🧪 What’s Next?</h3><ul><li>Add a MySQL database to the Docker Compose setup.</li></ul><h3>💡 Final Thoughts</h3><p>Thanks for following along with this guide on deploying Choreo Analytics with WSO2 API Manager 4.5 using Docker! 🙌</p><p>With just a few simple tweaks to our deployment.toml and a clean Docker Compose setup, we’ve unlocked the power of real-time cloud analytics for our APIs. 🎯</p><p>If you found this article helpful, please <strong>give it a clap</strong> 👏. It truly motivates me to create more content like this. Got questions, feedback, or your own tips? Drop a comment below. I’d love to hear from you 👂!</p><p>And don’t forget to <strong>share</strong> this article with your network if you think others could benefit.</p><p>Here’s to smooth APIs and happy coding! ✨</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=53241881447f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>