<?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 Arjun Kuduvalli Raveendra on Medium]]></title>
        <description><![CDATA[Stories by Arjun Kuduvalli Raveendra on Medium]]></description>
        <link>https://medium.com/@krarjun90?source=rss-545bcec3b8b4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*lLbkDSi5Q3GylI8PTg-QNg.png</url>
            <title>Stories by Arjun Kuduvalli Raveendra on Medium</title>
            <link>https://medium.com/@krarjun90?source=rss-545bcec3b8b4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 09:29:09 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@krarjun90/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[Hollow The Core — Vendor Systems]]></title>
            <link>https://medium.com/@krarjun90/hollow-the-core-vendor-systems-75e3fa435389?source=rss-545bcec3b8b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/75e3fa435389</guid>
            <category><![CDATA[breaking-monolith]]></category>
            <category><![CDATA[hollow-the-core]]></category>
            <category><![CDATA[monolithicvsmicroservice]]></category>
            <category><![CDATA[hollowthecore]]></category>
            <category><![CDATA[microservice-architecture]]></category>
            <dc:creator><![CDATA[Arjun Kuduvalli Raveendra]]></dc:creator>
            <pubDate>Wed, 26 Feb 2025 19:40:08 GMT</pubDate>
            <atom:updated>2025-02-26T20:10:57.605Z</atom:updated>
            <content:encoded><![CDATA[<h3>Hollow The Core</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9xpy4L-_JreAdT82r_nTcg.png" /></figure><p>We often develop applications by integrating with specialized systems that excel at solving specific problems, such as an order management system or a core banking system. While these solutions help avoid reinventing the wheel, they can come with limitations — such as lacking cloud compatibility, scalability, or flexibility — making it challenging to build high-performance, innovative products in today’s fast-evolving landscape.</p><p>In this blog, we will explore the differences between a typical microservice architecture and one that relies on a core product. We will also discuss how to overcome the limitations that come with integrating such systems.</p><p>In a <strong>microservice architecture</strong>, each service has its <strong>own domain logic and data ownership</strong>. It provides <strong>APIs</strong> to perform operations and access its data. Additionally, a service can <strong>publish events</strong>, which can be consumed by other microservices.</p><p>In some cases, a <strong>transaction workflow</strong> can be <strong>distributed across multiple services</strong>, where an event from one service <strong>triggers</strong> the next step in the workflow.</p><h4>Example Use Cases</h4><ul><li>An <strong>order placement event</strong> can be consumed by another service to send an <strong>email notification</strong>.</li><li>The same <strong>order placement event</strong> can trigger <strong>payment processing</strong>, and once the payment is completed, another event can initiate the <strong>order fulfillment workflow</strong>.</li></ul><p>In such scenarios, the <strong>initial event</strong> is published by the <strong>core service</strong>, but it may also expose <strong>APIs</strong> to provide <strong>additional details if needed</strong>.</p><p>For instance, an <strong>Order Placement event</strong> may contain <strong>minimal information</strong>, such as:</p><ul><li>OrderId</li><li>CustomerId</li><li>AddressId</li></ul><p>A <strong>consumer</strong> of this event may need <strong>additional data</strong>, like <strong>order items, price, or customer contact details</strong>. This would require API calls to:</p><ul><li><strong>Order Service</strong> (to fetch items, price, etc.)</li><li><strong>Customer Service</strong> (to retrieve customer contact details, email, phone number, etc.)</li></ul><h4>Challenges with Vendor Systems</h4><p>It is relatively <strong>easier</strong> to build a <strong>distributed system</strong> when you have <strong>full control</strong> over all services and data. However, <strong>challenges arise</strong> when integrating with <strong>third-party vendor systems</strong>:</p><ol><li>The <strong>APIs and events exposed by the system</strong> may not be sufficient for your evolving needs.</li><li>The <strong>pace of change</strong> may not align with your <strong>business requirements</strong>, especially if each enhancement requires a <strong>long approval process</strong>.</li></ol><h4>Real-World Issues</h4><ul><li>The <strong>API</strong> might expose a <strong>bulk “Get Orders” API</strong>, but you need features like <strong>filtering and sorting</strong>.</li><li>The <strong>system</strong> may <strong>not publish an event</strong> when <strong>customer profile data changes</strong>.</li><li>An <strong>Order Placement event</strong> may <strong>lack important details</strong>, forcing consumers to make additional API calls. A request to add these fields might take <strong>months</strong> to implement.</li><li><strong>Legacy infrastructure</strong>: If the application runs on <strong>non-scalable infrastructure</strong>, adding more logic <strong>increases load</strong>, leading to <strong>downtime and higher maintenance costs</strong>.</li></ul><h4>Key Pain Points</h4><ul><li><strong>Data Transformation </strong>— You may not get the data in the <strong>desired format</strong>.</li><li><strong>Slow Development Cycle</strong> — Dependence on the vendor makes changes <strong>time-consuming</strong>.</li><li><strong>Scalability Issues</strong> — Systems might not support <strong>efficient scaling</strong>.</li></ul><h3>Proposed Solutions</h3><h3>1. Build a New System on Top of the Source Data</h3><p>A potential solution is to build a <strong>new application</strong> that extends the system’s functionality <strong>while maintaining control over it</strong>.</p><h4>Approach: Read from a Replica Database</h4><ul><li>Build a <strong>new API layer</strong> that reads data from the <strong>secondary/replica database</strong> of the source system.</li><li>The <strong>original system</strong> continues handling <strong>writes</strong>, while the <strong>new service</strong> provides <strong>custom APIs</strong>.</li><li>This approach <strong>reduces dependency</strong> on core API limitations.</li></ul><p><strong>Challenge:</strong></p><ul><li><strong>Building APIs</strong> is straightforward, but <strong>publishing new events</strong> can be <strong>tricky</strong> without real-time data updates.</li></ul><h3>2. Leverage Change Data Capture (CDC) for Streaming Data</h3><p>A more <strong>scalable and real-time approach</strong> is to <strong>stream database changes</strong> into a <strong>messaging system</strong> and build a new service <strong>on top of these streams</strong>.</p><h4>How CDC Works</h4><ul><li>The source database <strong>emits change events</strong> (create, update, delete).</li><li>These events are <strong>captured</strong> and <strong>pushed into a messaging queue</strong> (Kafka, RabbitMQ, etc.).</li><li>A <strong>new application listens to these changes</strong> and triggers actions based on them.</li></ul><h4>Tools for CDC Implementation</h4><ul><li><strong>Kafka Connectors</strong>, <strong>GoldenGate</strong>, Debezium, etc., allow seamless <strong>CDC integration</strong>.</li><li>Some databases provide <strong>native change notifications</strong> to configured endpoints.</li></ul><h4>Use Cases</h4><ul><li><strong>New Order Placed?</strong> → Publish an <strong>Order Placed event</strong>.</li><li><strong>Order Status Updated?</strong> → Publish an <strong>Order Status Change event</strong>.</li><li><strong>Customer Profile Updated?</strong> → Publish a <strong>Customer Updated event</strong>.</li></ul><p>In this approach, the <strong>new system enriches the events</strong> by adding <strong>missing business-critical data</strong> (e.g., customer contact details, payment info, etc)</p><h3>What if we can’t connect to the source DB directly?</h3><p>Till now we were of the assumption that we would get access to the source database and we can build new system on top of it. However in many cases the data access would be restricted. Consider a bank where the financial services are provided by a vendor product and they own the database and only their core services will have access to it.</p><p>In such scenario, we can take an approach to get our own copy of data and then build the above discussed solutions on top of it.</p><blockquote><em>Migrate full data from source database to your own database and then keep the CDC enabled to keep getting the new changes.</em></blockquote><p>While migrating the data, you can choose to keep the structure as it is or perform some transformation to suit the chosen database. For ex, you can migrate data from an Oracle DB to MongoDB, Postgres, Aerospike etc and choose to model the data which suits the DB better.</p><h3>Evolving the System: “Hollowing the Core”</h3><p>By adopting these approaches, we are <strong>shifting domain logic to a new system</strong> while keeping the <strong>vendor system for core functionality only</strong>.</p><h3>Important Note</h3><p>This approach is <strong>different from breaking a monolith into microservices</strong> — which is a separate topic. You need to have control over the monolith to even think about splitting/refactoring one feature at a time into its own microservice.</p><p>For front-end applications, you can also find my article on <a href="https://medium.com/@krarjun90/breaking-a-monolith-via-micro-frontends-18e58f174360">breaking the monolith into micro frontends here</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=75e3fa435389" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Enthu Fans vs BCCI — CWC2023]]></title>
            <link>https://medium.com/@krarjun90/enthu-fans-vs-bcci-cwc2023-11676eea83fc?source=rss-545bcec3b8b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/11676eea83fc</guid>
            <category><![CDATA[bookmyshow]]></category>
            <category><![CDATA[bcci]]></category>
            <category><![CDATA[cricket-news]]></category>
            <category><![CDATA[cricket]]></category>
            <category><![CDATA[cricket-world-cup]]></category>
            <dc:creator><![CDATA[Arjun Kuduvalli Raveendra]]></dc:creator>
            <pubDate>Fri, 08 Sep 2023 12:05:19 GMT</pubDate>
            <atom:updated>2023-09-08T12:48:53.215Z</atom:updated>
            <content:encoded><![CDATA[<h3>The Enigma of Cricket World Cup 2023 Ticket Sales</h3><p>Who doesn’t like to be part of a WorldCup happening in their own yard? And when it is a Cricket WC (CWC) in India then the excitement is 100X. Matches would be held in some of the best stadiums in the country and each stadium can house from 30–35K spectators to even 100K in Ahmedabad. The organising committee (BCCI) chose BookMyShow (BMS) as their ticket booking partner. They even announced phase wise sale of tickets to avoid huge rush to the platform.</p><p>Initially non India match tickets sale went live. Couple of important matches went sold out within couple of minutes of going live. Few other less impact matches had seats left. Everyone thought the demand for the tickets is huge and tickets are getting sold out within minutes.</p><p>Then there was a pre-sale for India match tickets for MasterCard holders. A section of fans had high hopes that this is their chance to get early access to tickets beating the heavy traffic which would be there when all tickets will go on sale. At 6PM on August 29th, when fans rushed to BMS, they were in for a surprise. They were added to a queue and they started seeing waiting times starting from few minutes to few hours. People waited in the queue until they got a chance to see the seat selection page, and to their misery their were no seats left to select.</p><p>I was also one such fan who was in waiting queue. It started with 40 minutes waiting which eventually went on to 2 hours before I gave up.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/571/1*RhAJMPzeyn8pe6boEFvRCA.jpeg" /></figure><p>I started thinking about the user experience they could have provided in this waiting page. For example, even the most criticised website in India IRCTC shows the number of seats left effectively even during the high throughput Tatkal time slot.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qxpZ61BgdldB3msDsDq2zw.png" /></figure><p>IRCTC also shows how many seats are available, the different categories and price upfront even before the Tatkal slot opens, so that people are aware of what they are fighting for.</p><p><em>Even though the use cases, layouts, throughput split across skus are different there was still scope for BMS to give more meaningful information to the people in queue.</em></p><p>On the other hand they chose to not share any such information upfront while in waiting queue. It would be completely fine had BCCI announced how many tickets are coming up for sale.</p><p>While I was in the queue, the developer in me wanted to check how the page is loading the information about the queue and if I can get more information on the current position or total number of people waiting. I started inspecting the network calls and surprisingly found one public API which was fetching the complete event information. While I was searching for queue information, what I found instead was the information of different categories / layout of that venue.</p><pre>{<br>  &quot;ET00367559-10065-T312&quot;:<br>  {<br>    &quot;AreaCatCode&quot;: &quot;AC00000177&quot;,<br>    &quot;AvailableSeats&quot;: 71,<br>    &quot;CategorySalesChannel&quot;: &quot;|M5|MOBILE|OUTLET|PAYLATER|POS|TP|WEB|&quot;,<br>    &quot;CinemaIsSVG&quot;: &quot;NONSVG&quot;,<br>    &quot;DynamicPrice&quot;: &quot;&quot;,<br>    &quot;EventCode&quot;: &quot;ET00367559&quot;,<br>    &quot;IsSVG&quot;: &quot;Y&quot;,<br>    &quot;MaxTickets&quot;: 10,<br>    &quot;Price&quot;: &quot;3500.00&quot;,<br>    &quot;PriceCode&quot;: &quot;T312&quot;,<br>    &quot;PriceDesc&quot;: &quot;BLOCK B BAY 7-LOWER&quot;,<br>    &quot;PriceDescEX&quot;:<br>    [],<br>    &quot;PriceSequence&quot;: &quot;1&quot;,<br>    &quot;SeatLayout&quot;: &quot;Y&quot;,<br>    &quot;SeatsAvail&quot;: &quot;Y&quot;,<br>    &quot;SessionID&quot;: &quot;10065&quot;,<br>    &quot;ShowDateDisplay&quot;: &quot;Saturday, 14 Oct&quot;,<br>    &quot;ShowTime&quot;: &quot;02:00 PM&quot;,<br>    &quot;TotalSeats&quot;: 500,<br>    &quot;VenueAddress&quot;: &quot;Motera Stadium Road, Motera, Ahmedabad, Gujarat 380005, India&quot;,<br>    &quot;VenueCode&quot;: &quot;SPSM&quot;,<br>    &quot;VenueMax&quot;: &quot;&quot;,<br>    &quot;VenueName&quot;: &quot;Narendra Modi Stadium, Motera: Ahmedabad&quot;,<br>    &quot;cutOffTime&quot;: 202310142359,<br>    &quot;dtmTime&quot;: &quot;14-10-2023 14:00:00&quot;,<br>    &quot;location&quot;: &quot;Ahmedabad&quot;<br>  },<br>  // other categories<br>}</pre><p>Looking at the data, there is not much sensitive information present here right? But if we look closely, we can extract some useful information which are not available for individuals to see in the screen until they come out of waiting room. I can see how many categories are there, the price at which each category is being sold, total number of seats under each category and how many seats are still available.</p><p>I thought this would be handy information to know while waiting in the queue in the following days when all tickets would go for sale.</p><p>I wrote a quick application to parse the data and log useful information in the console.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/882/1*LJQsz_FWUG-S05gwxndGDw.png" /></figure><p>So as expected, the MasterCard early access sale had sold very few tickets. I was now ready to start booking again in the coming days and not be in dark when I am in waiting room.</p><p>The next couple of days when BMS sold tickets, I used this app and I was surprised to see the numbers. The total number of tickets sold were hardly &lt;10% of the stadium capacity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JnisL7r3bTVWBq4qGu0GpQ.png" /></figure><p>I even tried to cross verify this information with the web layout.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s9bTBD5XA4qD1rlESk3QTg.png" /></figure><p>The only information which were missing was how many tickets were earlier sold under MasterCard sale and how many new tickets were added during full sale.</p><p>So the coming day when highly anticipated India vs Pakistan match tickets were going for sale, I got the initial MasterCard sale data the day before so that I can compare with the new data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*449p1Tuu3o8atomIc_aiPw.png" /></figure><p>And on the day of sale after they updated the configuration, the latest data looked like below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QkLXfUpIII_rnTS_Yd6QhA.png" /></figure><p>I started sharing these information in twitter questioning the BCCI on selling only the partial seats to fans. Many people started noticing this. Few more started to check this themselves by inspecting the API calls.</p><h3>Sukhmanjit Singh on Twitter: &quot;I have looked at API response of bookmyshow for IndvPak worldcup match on October 14. Total of 8334 tickets in 15 blocks have been made available in a stadium with a capacity of 132000.@bookmyshow @ICC @BCCI @mufaddal_vohra @allaboutcric_ #scam #INDvPAK #worldcuptickets pic.twitter.com/SHkI4cofrs / Twitter&quot;</h3><p>I have looked at API response of bookmyshow for IndvPak worldcup match on October 14. Total of 8334 tickets in 15 blocks have been made available in a stadium with a capacity of 132000.@bookmyshow @ICC @BCCI @mufaddal_vohra @allaboutcric_ #scam #INDvPAK #worldcuptickets pic.twitter.com/SHkI4cofrs</p><h3>Ajitesh Tripathi on Twitter: &quot;Who says you can&#39;t get rich overnight?BookMyShow did it. You can too!!The Queue says to wait for more than 2 hours but the API response tell only 1 seat available.#BookMyShowScam #BCCI #INDvsPAK pic.twitter.com/YQpyJA5tYG / Twitter&quot;</h3><p>Who says you can&#39;t get rich overnight?BookMyShow did it. You can too!!The Queue says to wait for more than 2 hours but the API response tell only 1 seat available.#BookMyShowScam #BCCI #INDvsPAK pic.twitter.com/YQpyJA5tYG</p><h3>Yash Patel on Twitter: &quot;#BookMyShowScam#INDvPAK It&#39;s clear from their API response that available seats are ZERO in all stands!!!Most tickets are directly sold to black market Well player bcci 🫡 pic.twitter.com/GH9fw54IpO / Twitter&quot;</h3><p>BookMyShowScam#INDvPAK It&#39;s clear from their API response that available seats are ZERO in all stands!!!Most tickets are directly sold to black market Well player bcci 🫡 pic.twitter.com/GH9fw54IpO</p><h3>Gunz ⚡️ on Twitter: &quot;2699 tickets were available of Rs 2000 for #INDvPAK world cup match at 7:40pm . CHECK THE SCREENSHOT ATTACHEDTOTAL CAPACITY OF BLOCKS(Q,P,L,K) WHICH ARE OF 2000RS IS MORE THAN 14000.SOMETHING IS WRONG.#BookMyShow #BookMyShowScam pic.twitter.com/Alia6b3fz7 / Twitter&quot;</h3><p>2699 tickets were available of Rs 2000 for #INDvPAK world cup match at 7:40pm . CHECK THE SCREENSHOT ATTACHEDTOTAL CAPACITY OF BLOCKS(Q,P,L,K) WHICH ARE OF 2000RS IS MORE THAN 14000.SOMETHING IS WRONG.#BookMyShow #BookMyShowScam pic.twitter.com/Alia6b3fz7</p><h3>Ayan on Twitter: &quot;Inspected API response of Book My Show from network tab and found sum of Total Seats from all stands. And it&#39;s just 8334. LOL @bookmyshow how will you explain this?#ICCWorldCup2023 #BookMyShowScam pic.twitter.com/Icr07lAOoa / Twitter&quot;</h3><p>Inspected API response of Book My Show from network tab and found sum of Total Seats from all stands. And it&#39;s just 8334. LOL @bookmyshow how will you explain this?#ICCWorldCup2023 #BookMyShowScam pic.twitter.com/Icr07lAOoa</p><p>There was already a negative vibe around BMS as most fans were not able to book any tickets for India matches. Now along with the negative sentiments, people started questioning the allocation of just few thousand tickets. Few celebrities also raised their voices against this incident.</p><h3>Venkatesh Prasad on Twitter: &quot;I urge the @BCCI to have more transparency in the World Cup ticketing system and not take fans for granted. Definitely in a stadium like Ahmedabad, for an #IndvsPak clash more than the sold 8500 tickets need to be available when the capacity is 1 lakh + . Likewise for all other... / Twitter&quot;</h3><p>I urge the @BCCI to have more transparency in the World Cup ticketing system and not take fans for granted. Definitely in a stadium like Ahmedabad, for an #IndvsPak clash more than the sold 8500 tickets need to be available when the capacity is 1 lakh + . Likewise for all other...</p><p>Now BCCI has said that they are going to release additional 400K tickets in another phase. There was no mention of a 2nd phase ever before. So we are not sure if this was a pre-planned event or something they have come up now to manage PR.</p><p>Enough of the story.</p><p>Now if we try to analyse the BMS services,</p><ul><li>Firstly, why was this API present. Was it required by either mobile/web application to show some content on the screen.</li><li>If the API was required, then should it return all the data? Would it be better to restrict the amount of information.</li><li>If the API and data are necessary, then should it have better access control?</li></ul><p>Now when I tried to run the application again, the API which returned the event information is no longer accessible. This brings down to the question</p><ul><li>Was that API even supposed to be there in public?</li></ul><p>As I said earlier the information is not sensitive. I also am not sure if they would have restricted this API ever. They could have definitely used this API data to give better user experience in the BMS application.</p><p>But now that they have restricted this data, I want to ask in Arnab’s voice, the Nation Wants to Know</p><blockquote>Why is BCCI and BMS hiding the number of tickets actually sold?</blockquote><blockquote>All these things are my own version of the ticket fiasco. The API could be completely out of place not maintained and it might be returning wrong data. BCCI from the beginning might had plans to see all the tickets in multiple phases. But it was good couple of evenings to spend time around this thinking about APIs, data access and user experience while eagerly trying (and failing) to book a ticket for the CWC.</blockquote><p><a href="https://github.com/krarjun90/CWC2023-Ticket-Tracker/blob/main/README.md">CWC2023-Ticket-Tracker/README.md at main · krarjun90/CWC2023-Ticket-Tracker</a></p><p>These days as I am practicing to write down my experience, I try to ask ChatGPT to rewrite my blog so that I get to learn different ways of expressing my crude thoughts. Below is ChatGPT version of the same.</p><h3><strong>ChatGPT Version as a cricket critic writer</strong></h3><p>Cricket aficionados across the Indian subcontinent were abuzz with anticipation as the 2023 Cricket World Cup (CWC) descended upon the cricketing heartland. The prospect of witnessing a world-class tournament unfold in their own backyard, particularly a Cricket World Cup, sent excitement levels soaring to unprecedented heights. The chosen venues, some of the most renowned in the country, promised to host throngs of passionate cricket devotees, ranging from 30,000 to an astonishing 100,000 at the magnificent Ahmedabad stadium. To facilitate this grand spectacle, the Board of Control for Cricket in India (BCCI) entrusted BookMyShow (BMS) with the vital role of ticketing.</p><p>The logistical challenges were apparent. With such colossal demand anticipated, BCCI wisely decided to roll out ticket sales in phases, an astute move to mitigate the digital pandemonium expected at their online ticketing platform.</p><p>The initial phase saw the release of non-India match tickets. To no one’s surprise, several high-profile clashes were sold out within mere minutes of going live. Yet, a few less marquee encounters still had vacant seats, leaving fans under the impression that demand was insatiable.</p><p>The drama truly unfolded during the pre-sale of India match tickets, exclusively available to MasterCard holders. Eager fans, hoping for privileged access and relief from the impending rush, found themselves thrust into an unexpected waiting queue. As the clock struck 6 PM on August 29th, users watched their screens, their wait times oscillating from a few minutes to an agonizing few hours. The crescendo of anticipation eventually led them to the seat selection page, only to find a bitter truth — they were too late, as not a single seat remained available.</p><p>I, too, found myself among the hopefuls relegated to this digital purgatory. A frustrating 40-minute wait stretched into an interminable 2-hour ordeal before I reluctantly conceded defeat.</p><p>It was during these long hours of waiting that I pondered the user experience — or rather the lack thereof — offered by the waiting page. I couldn’t help but draw comparisons with the oft-criticized Indian Railway Catering and Tourism Corporation (IRCTC) website, which effectively displays the number of available seats even during the frenetic Tatkal booking window.</p><p>IRCTC not only showcases seat availability, but also presents comprehensive details on categories and pricing well before the Tatkal rush begins, ensuring prospective travelers are well-informed.</p><p>Though the two platforms serve disparate purposes and handle varying levels of throughput, there was ample room for BMS to provide more meaningful information to its waiting patrons. Unfortunately, such insights were conspicuously absent, leaving fans in the dark regarding the number of tickets to be released.</p><p>During my sojourn in the waiting queue, my developer instincts kicked in, prompting me to investigate how the system handled queue information. Much to my surprise, I stumbled upon a public API that was fetching comprehensive event details. While my initial quest was for queue statistics, I found myself gazing at an array of valuable data. It revealed the number of ticket categories available, their corresponding prices, total seat counts for each category, and the remaining inventory.</p><p>This newfound wealth of information struck me as indispensable while navigating the queue in the days that lay ahead when tickets for all matches would be up for grabs. Consequently, I swiftly devised an application to parse and log this valuable data, all in the pursuit of informed ticket purchases.</p><p>As anticipated, the MasterCard early access sale yielded disappointingly low sales figures. Armed with these insights, I eagerly awaited the subsequent ticket releases and was astounded by the statistics that emerged. The total number of tickets sold barely scratched the surface of the stadium’s vast capacity.</p><p>I sought to corroborate these findings with the web layout, but crucial data remained elusive. It was then that I decided to compare the MasterCard sale data with the forthcoming full sale data, particularly for the highly anticipated India vs. Pakistan match.</p><p>Upon sharing this data on Twitter, it sparked a wave of intrigue and questions about BCCI’s ticket allocation practices. A negative undercurrent had already pervaded BMS, primarily due to fans’ inability to secure India match tickets. Now, with evidence of a meager allocation of seats, discontent began to fester, culminating in celebrities voicing their disapproval.</p><p>In a surprising twist, BCCI later announced the release of an additional 400,000 tickets in a subsequent phase, a revelation that had not been previously communicated. This development left many wondering whether this was a premeditated strategy or a reactionary attempt at damage control.</p><p>In hindsight, one must evaluate the role and purpose of the public API that inadvertently revealed these ticketing statistics. Was its existence warranted, and if so, should it have exposed such extensive data? Could better access controls have been implemented? A perplexing twist to the tale emerged when the API, which had been a wellspring of information, was abruptly rendered inaccessible. This begs the question of whether the API should have been public-facing in the first place.</p><p>While the data unearthed was not sensitive in nature, its revelation raises questions about transparency and user experience. BMS could have harnessed this API to provide an enhanced user experience, yet chose to restrict it. The burning query now resonates, echoing in the words of Arnab Goswami, “The Nation Wants to Know”: Why are BCCI and BMS concealing the actual count of tickets sold?</p><p>In conclusion, this tale unfolds as a saga of digital intrigue. The API’s role in this enigma remains uncertain — a potential oversight or a deliberate channel of information. Regardless, it prompted contemplation on the significance of APIs, data access, and user experience, all amidst the fervent quest for a coveted Cricket World Cup ticket.</p><h3><a href="https://github.com/krarjun90/CWC2023-Ticket-Tracker/blob/main/README.md">For further details and analysis, you can refer to the author’s repository.</a></h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=11676eea83fc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Breaking a monolith via micro frontends]]></title>
            <link>https://medium.com/@krarjun90/breaking-a-monolith-via-micro-frontends-18e58f174360?source=rss-545bcec3b8b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/18e58f174360</guid>
            <category><![CDATA[micro-front-end]]></category>
            <category><![CDATA[breaking-monolith]]></category>
            <category><![CDATA[frontend-architecture]]></category>
            <category><![CDATA[refactoring]]></category>
            <category><![CDATA[front-end-development]]></category>
            <dc:creator><![CDATA[Arjun Kuduvalli Raveendra]]></dc:creator>
            <pubDate>Sat, 14 Jan 2023 09:18:44 GMT</pubDate>
            <atom:updated>2023-01-14T09:18:44.584Z</atom:updated>
            <content:encoded><![CDATA[<p>How many of us get to work on green field projects? A few. Rest of us get to work on either a legacy monolith system or in a rare cases a recent system which has now turned into a monolith due to continuous features added on top of it without refactoring or without making any design changes to it.</p><p>Breaking a monolith into micro applications is a very vast topic. Building a new system and not turning it into a monolith is also equally important. You will find lot of articles on micro service architectures and modular frontend. I will try to explain the same with a more practical approach.</p><p>In this blog I will concentrate on the frontend applications. I will be taking a typical E-commerce application as an example.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/954/1*PqyTE8sfQ_y40P1fYtXapg.png" /><figcaption>An application with common header/footer and dynamic body</figcaption></figure><p>The most common aspect in almost all the E-commerce applications is having a uniform page structure in all the pages — be it home page, product listing, product description or checkout. It will have a header and footer with common theme and almost similar content. What changes most is the body. The header section also changes in many cases where you can have a user profile, mini cart, hamburger menu etc but I am keeping it simple at the moment.</p><p>The body can be constructed with many elements. It can have a range of components like filters, product list/grid, product description, reviews etc. A simple example is below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/954/1*yANBIXRvRMv2EgUgxpCoZQ.png" /><figcaption>A page with multiple components</figcaption></figure><p>When the application is working well in production, one see no need to make architectural changes and continue adding features on the existing page until a point it becomes a maintenance hell.</p><ul><li>Adding a new feature becomes tricky in a unmanaged lengthy spaghetti codebase</li><li>Managing the tests become more complex</li><li>Coordination with multiple people working on same page delays the release process</li></ul><h3>How do we break this monolith?</h3><p>One can not just design an entire new architecture, migrate all the features to new micro frontends and release a new version to production. Working like that will take us back to waterfall time. Then how to be agile? We can plan migration in increments and test the increments in production to get quicker feedback.</p><p>A good starting point is to pick a feature which needs immediate attention and extract it into a new micro component — we call it micro frontend (MFE). The component can be injected in the existing page. It can be quickly tested to see if the changes have integrated seamlessly in the page.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Fox1bysbvUvW6cg69B0CJA.png" /><figcaption>Incremental integration in 2 iterations (Iter1 and Iter2)</figcaption></figure><p>Advantages of starting with one small component are</p><ul><li>it gives the confidence in the new micro frontend</li><li>it gives quick feedback in the way micro frontend is integrated</li><li>it creates a base structure for others to start working on other parts of the application</li></ul><p>Once we have few base components ready, it will be easier to create a full page outside the legacy codebase and over time old pages can be archived.</p><p><em>(In the above illustration, product details MFE is integrated initially in first iteration and any other MFE can be built easily in the following iterations — even multiple MFE built and integrated in parallel)</em></p><h4>Alternate way?</h4><p>There will be times when you need a completely new feature but still find it difficult to add it in the monolith. In such cases we can try to create a new full page hosted outside the monolith. As mentioned earlier, creating a full page can take a lot of effort. So a good incremental way is to create a toned down version of the old application theme with reduced functionality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*huvXkdUKXQ4Jg67jeL8FHQ.png" /><figcaption>New page with simplified theme</figcaption></figure><h3>Boundaries</h3><p>A small feature of the application can be separated out in multiple ways.</p><ul><li>Identify a page which is easy to separate</li><li>Identify a component.</li></ul><p>The components which constitute a page can have different behaviour.</p><ul><li>Read only / render only component which needs some data</li><li>Component with state — which needs some initial data, and any change in state can be conveyed by call back functions.</li></ul><h4>Component Library</h4><p>Within an organisation, there will be multiple teams working on different parts of the application. Even though the functionality changes, the UI usually matches a theme. For example, a product card can be used with same design in search results, recommended products, wishlist, cart etc. A component to collect address can be used in user profile and checkout. It is worthy enough to create a component library from the basic components and reduce the development effort for others.</p><h4>Widgets</h4><p>Components with state can become too complex. There will be cases where it need to perform multiple operations like server side validations, fetching data based on change in user interaction etc. For example in a checkout page, once user enters zip code the component can validate the zip code, fill the address, fetch the shipping options, fetch the estimated delivery range etc. All the API calls can be handled outside the component and pass the necessary data as a property to the component. However approach like this will be difficult to reuse the component in another page as you will have to do all the API calls again in other component.</p><p>The term “widgets” can refer to both third-party components that are integrated into an application, such as “Login with Google” or “Pay with ApplePay,” as well as reusable, self-sustaining components that contain business logic that have been developed in-house.</p><p>In such cases where the component has complex business logic, it is advisable to make the component self sufficient so that it can make all the necessary network calls on its own and only give the desired data by callback functions.</p><h3>Passing the data</h3><p>The moment we split the application into components and pages, there is additional overhead on passing the data between them.</p><ol><li>Data between a single page application (SPA) and its components can be passed via properties, callback functions, events and listeners.</li><li>Data between pages can be passed by<br>- storing information in client side in cookies or session storage. <br>- storing information in server session (fetching using session id / cookie)<br>- query parameters</li></ol><h3>Integrating the micro frontend</h3><p>The component integration can be done in multiple ways.</p><ul><li>inject it at run time</li></ul><p>The component can be served in many ways — as simple as keeping it in a storage like S3 bucket, in a CDN or even served by a web server.</p><ul><li>add as build time dependency (importing a react module)</li><li>hybrid composition — An approach involves a main application that can request one or more pages. Based on the context or route, the new application server can either serve the entire content or load a portion of it by proxying it to another micro frontend.</li></ul><h3>Advantages of micro frontend</h3><ul><li>Each individual component can be built, tested and published separately</li><li>Errors in each component can be managed differently and error in one part need not break the whole application</li><li>Reduces dependency on other teams</li></ul><h4><strong>Note</strong></h4><p>When components are added as build time dependency, there will still be a small bottle neck on one team who manages the base application to update the modules and release it to production. In run time integration, there is possibility to reduce the dependency only when there is a major change.</p><p>Breaking the frontend usually is not the only concern in a monolith. Extracting out the services also goes hand in hand to make a part of the system completely independent, reusable and easy to evolve.</p><p>Stay tuned for the next blog on “Breaking the monolith into micro services”.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=18e58f174360" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[It’s not JUST a feature]]></title>
            <link>https://medium.com/@krarjun90/its-not-just-a-feature-d995ad43315b?source=rss-545bcec3b8b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/d995ad43315b</guid>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[usability-testing]]></category>
            <category><![CDATA[user-experience-design]]></category>
            <category><![CDATA[service-design]]></category>
            <dc:creator><![CDATA[Arjun Kuduvalli Raveendra]]></dc:creator>
            <pubDate>Fri, 06 Jan 2023 07:55:09 GMT</pubDate>
            <atom:updated>2023-01-06T07:55:09.373Z</atom:updated>
            <content:encoded><![CDATA[<p>How many of you have heard this typical statement “Its simple, can you just add a new input field” and someone counters it with “Nothing is as simple as it seems. You need to add the field, add validations, update the tests, check overall UI compatibility, and then take the code to production.” ?</p><p>Just like that what happens if you want to introduce a new feature in your application? Is it just a new feature or are you building a new user experience? Let’s see how you need to think building a service around the feature.</p><p>In most of the e-commerce websites we would have seen a similar checkout page where you can enter an address to ship the product to. Some brands have provided customers with the option to collect their items from one of their stores.</p><figure><img alt="Collect In-Store options provided by different brands" src="https://cdn-images-1.medium.com/max/621/1*-mCtpkedZ7VMDhXTB2zH3w.png" /></figure><p>If you observe the image above, you can see different approaches taken by brands to provide the alternate option. One can just see it as just a change in checkout user interface. However making any change to existing behaviour or adding a new functionality should influence user behaviour, provide seamless service offering and involves integration with a lot of different systems.</p><h3>User Experience</h3><p>As much as we think about providing a new feature to the user, we should also make sure if the feature is understood well by the user, is it clear enough, and is it able to highlight the value it offers.</p><p>In the current example of providing an option to the customer to collect/pick the product from the store let’s see what are some of the key points which can influence the design.</p><ul><li>Do you want to promote the feature upfront so that the user is aware of the offering? In the picture above, two of the designs feature a binary selection to highlight the option.</li><li>How do you highlight the benefit or service offering? Is one option faster than the other? Does one option provide more flexibility?</li><li>How to make the comparison between services easier without making user to juggle between the options?</li></ul><h4>Usability Testing</h4><p>Usability Testing should be done to get very early feedback. It will take some attempts to frame the questions and conduct it right. Once we get it right, we can define a lot of improvements to make the customer life better.</p><ul><li>Is showing address of the store sufficient or a map view gives more clarity?</li><li>Does it help the customer to take better decision if they know the opening hours of the store or when they can collect it?</li><li>Who should be contacted in case of issue with the order - the one who placed the order or the one who was supposed to pick the order?</li></ul><p>The right questions and qualitative feedback can help us design the right experience and provide the necessary information in the right places.</p><h3>Design a Service</h3><p>The feature is not always restricted to a web page with few clicks. It goes even beyond after the customer collects the item in hand. So providing the right information at the right time and making sure the entire process is hassle free is important.</p><p>Is providing all the information we discussed in UI sufficient? How many real world users actually see the minute details while purchasing a product? This means providing related information at different points, such as in</p><ul><li>order confirmation email can highlight that order is placed but <strong>not ready yet</strong> to collect immediately</li><li>ready to collect email can mention the <strong>store address (coordinates in map) and opening hours</strong>, and mention steps to follow in case the person designated to collect can not come and someone else need to collect it on behalf of them.</li></ul><h3>Any other actors?</h3><p>The service need not involve just the customer placing the order. There will be other parties like warehouse manager, delivery partner, store manager and providing them necessary information in also important in this whole journey.</p><ul><li>The<strong> warehouse manager</strong> should know the difference between an order being shipped directly to customer vs order shipped to store.</li><li>The <strong>store manager</strong> should know the items are for which customer so that they can be better prepared for the customer’s visit. They can also out reach to the customer to understand if they need any other product or service at store.</li><li>The <strong>store manager</strong> should know how many orders are in transit to store so that they can better plan their operations.</li><li>In case of delivery issues at store, the <strong>delivery partner</strong> need the store contact information. Instead if they call the customer, it will create a bad experience.</li></ul><h3>Learnings</h3><ul><li>However small a feature is perceived in the beginning, we should start looking at overall user experience since the beginning. A feature is not restricted to just one system.</li><li>Usability Testing should be done to get very early feedback. It will take some attempts to frame the questions and conduct it right. Once we get it right, we can define a lot of improvements to make the customer life better.</li><li>An information really trivial to one system can act as a major factor in other system. Work closely with different systems involved to avoid making hot-fixes in production.</li></ul><p>— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —</p><p>I asked ChatGPT to analyse the above post and asked it to write a blog on similar topic and below is the response.</p><blockquote>When it comes to introducing new features to an e-commerce website, it’s important to think beyond just the technical implementation. A new feature is not just a simple addition, but rather it has the potential to significantly impact the user experience and the overall service offering. In this article, we’ll explore how to approach introducing a new feature in a holistic manner, considering the user experience, service design, and the various parties involved in the process.</blockquote><blockquote>First and foremost, it’s crucial to consider the user experience when introducing a new feature. How will the feature be understood and utilized by the user? Will it be clear and easy to use? Will it effectively highlight the value it offers? These are all important questions to consider, and usability testing can be a valuable tool in gathering feedback and identifying areas for improvement.</blockquote><blockquote>For example, let’s say that we want to introduce the option for customers to collect their orders from one of our stores rather than having them shipped to their address. In designing this feature, we might ask ourselves: Do we want to promote the feature upfront to make sure that users are aware of it? How can we highlight the benefits or service offerings of this option, such as faster delivery or greater flexibility? And how can we make it easy for users to compare the different options and make the best choice for them?</blockquote><blockquote>In addition to the user experience, it’s also important to consider the service design when introducing a new feature. A feature is not just limited to a single web page or a few clicks; it extends beyond that to the entire customer journey. This means that it’s important to not only provide the right information at the right time, but also to ensure that the entire process is seamless and hassle-free.</blockquote><blockquote>For example, in the case of our store pickup feature, we might consider sending relevant information in order confirmation and ready-to-collect emails, such as the store address and opening hours, and outlining steps to follow in case the person designated to collect the order is unable to do so. We might also consider the needs of other parties involved in the process, such as warehouse and store managers and delivery partners, to make sure that they have the information they need to ensure a smooth experience for all involved.</blockquote><blockquote>Overall, introducing a new feature to an e-commerce website requires a holistic approach that considers the user experience, service design, and the various parties involved in the process. By taking this approach, we can create a feature that not only adds value for our customers, but also enhances the overall service offering of our website.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d995ad43315b" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>