<?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 Azizul maqsud on Medium]]></title>
        <description><![CDATA[Stories by Azizul maqsud on Medium]]></description>
        <link>https://medium.com/@azizulmaqsud?source=rss-4df114cb0773------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*WUb9uBOiyMUybG4cHaaGLA.png</url>
            <title>Stories by Azizul maqsud on Medium</title>
            <link>https://medium.com/@azizulmaqsud?source=rss-4df114cb0773------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 17:39:53 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@azizulmaqsud/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[Building an AI DevOps Agent on AWS with Strands and Amazon Bedrock AgentCore]]></title>
            <link>https://medium.com/@azizulmaqsud/building-an-ai-devops-agent-on-aws-with-strands-and-amazon-bedrock-agentcore-d447a8dcf4b0?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/d447a8dcf4b0</guid>
            <category><![CDATA[strands-agents]]></category>
            <category><![CDATA[amazon-bedrock-agentcore]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[cloud-engineering]]></category>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Tue, 08 Sep 2026 02:26:52 GMT</pubDate>
            <atom:updated>2026-09-08T02:44:32.467Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Written by:</strong> Azizul Maqsud<br><strong>Date:</strong> September 07, 2026</p><p><strong>From a simple Python agent to a containerized, IAM-controlled AWS workload</strong></p><p><strong>A hands-on project building a read-only AWS DevOps assistant with Strands Agents, Amazon Bedrock, Docker, ECR, CodeBuild, IAM, and AgentCore Runtime.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/551/1*XL7sVG8kAA2-Ngeti_ANZQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yu8lXIwsLQm4sC6ZAWunlg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*huLTp_c6a2D4MtNYzZlhig.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/777/1*V2NGx3Xdg3EwMtAFZSirfg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/861/1*3jcHQbKMQ0pUtzyE6vQaAg.png" /><figcaption>How an AgentCore Works in the Realtime Architecture</figcaption></figure><p>AI agents are easy to demonstrate.</p><p>But running one reliably in a real cloud environment is a different challenge.</p><p>Recently, I explored how to build an AI agent using <strong>Strands Agents and Amazon Bedrock AgentCore</strong>, with the goal of creating something more practical than a chatbot.</p><p>The idea was simple:</p><p><strong>What if I could ask an AI agent about my AWS environment instead of manually jumping between AWS Console, CLI, CloudWatch, and Kubernetes tools?</strong></p><p>For example:</p><p>“List my EKS clusters.”</p><p>“Show my EC2 instances.”</p><p>“List my S3 buckets.”</p><p>“What’s the CPU utilization of this instance?”</p><p>The agent would reason about the request, select the appropriate tool, call the AWS API, and return the result.</p><p>But getting there exposed several interesting DevOps problems</p><p>The final architecture looks like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/659/1*NzadQvv8zCjVHZby0Q-dVQ.png" /></figure><p>Here, the LLM doesn’t directly “know” my AWS environment.</p><p>Instead, I give the agent tools.</p><p><strong>Building the Agent</strong></p><p>The first layer is a normal Python application using Strands.</p><p>The agent has tools such as:</p><p>@tool</p><p>def list_eks_clusters():</p><p>response = eks.list_clusters()</p><p>return response.get(“clusters”, [])</p><p>Other tools handle:</p><ul><li>Amazon S3</li><li>Amazon EKS</li><li>Amazon EC2</li><li>CloudWatch</li><li>Weather API</li></ul><p>This creates a useful separation:</p><p>LLM</p><p>↓</p><p>Reasoning</p><p>↓</p><p>Tool selection</p><p>↓</p><p>AWS API</p><p>↓</p><p>Real infrastructure data</p><p>That’s much safer than asking the model to generate infrastructure information from its own knowledge.</p><p><strong>But The First Real Problem Arises: When The Agent Needs a Runtime Contract</strong></p><p>A Python script that works on my laptop isn’t automatically an AgentCore application.</p><p>The runtime expects the containerized application to expose the appropriate HTTP interface.</p><p>That’s why I added FastAPI:</p><p>GET /ping</p><p>POST /invocations</p><p>The health endpoint is intentionally simple:</p><p>@app.get(“/ping”)</p><p>def ping():</p><p>return {“status”: “Healthy”}</p><p>And the agent request goes through:</p><p>POST /invocations &gt; Strands Agent &gt; Bedrock &gt; Tool</p><p>This small change turns the agent from a Python script into a deployable service.</p><p><strong>The ARM64 Problem</strong></p><p>This was one of the most important deployment considerations.</p><p>My development machine may be running:</p><p>x86_64</p><p>while the AgentCore container environment expects:</p><p>ARM64</p><p>A normal Docker build can therefore produce the wrong architecture.</p><p>Instead, the project explicitly builds:</p><p>docker buildx build \</p><p>— platform linux/arm64 \</p><p>-t $ECR_URI:latest \</p><p>— push .</p><p>This is one of those problems that doesn’t show up while writing the Python code.</p><p>Everything can work perfectly locally and still fail when deployed.</p><p>That’s why I now treat <strong>container architecture as part of application configuration</strong>, not just infrastructure configuration.</p><p><strong>From Docker to ECR</strong></p><p>Once the application works locally, the next step is Amazon ECR.</p><p>The pipeline becomes:</p><p>Python &gt; FastAPI &gt; Docker &gt; ARM64 Image &gt; Amazon ECR &gt; AgentCore Runtime</p><p><strong>The Docker container listens on port 8080:</strong></p><p>EXPOSE 8080</p><p>CMD [“uvicorn”,</p><p>“app.agent.server:app”,</p><p>“ — host”, “0.0.0.0”,</p><p>“ — port”, “8080”]</p><p>Before deploying, I test the container locally:</p><p>curl <a href="http://localhost:8080/ping">http://localhost:8080/ping</a></p><p>Then:</p><p>curl -X POST <a href="http://localhost:8080/invocations">http://localhost:8080/invocations</a> \</p><p>-H “Content-Type: application/json” \</p><p>-d ‘{“prompt”:”List my EKS clusters”}’</p><p>This gives me a useful troubleshooting boundary.</p><p>If the container fails locally, AgentCore isn’t the problem.</p><p><strong>IAM Becomes Part of the AI Architecture</strong></p><p>This is probably my favorite part of the project.</p><p>An AI agent shouldn’t have administrator access just because it needs to inspect AWS.</p><p>The runtime execution role is intentionally limited.</p><p>For example:</p><p>eks:ListClusters</p><p>ec2:DescribeInstances</p><p>s3:ListAllMyBuckets</p><p>cloudwatch:GetMetricStatistics</p><p>bedrock:InvokeModel</p><p>The design principle is:</p><p><strong>The model decides which tool to use. IAM decides what the tool is allowed to do.</strong></p><p>That’s an important distinction.</p><p>If the model makes a bad decision, restrictive IAM can still prevent destructive operations.</p><p>For the first version, I intentionally kept the tools read-only.</p><p>The agent can:</p><p>Observe, Analyze, Explain</p><p>But it cannot:</p><p>Delete, Terminate, Modify</p><p><strong>Then Comes CodeBuild</strong></p><p>The next step was turning the deployment into a DevOps workflow.</p><p>Instead of manually building images:</p><p>Developer</p><p>↓</p><p>docker build</p><p>↓</p><p>docker push</p><p>Here I use:</p><p>GitHub</p><p>↓</p><p>AWS CodeBuild</p><p>↓</p><p>ARM64 Docker Build</p><p>↓</p><p>Amazon ECR</p><p>↓</p><p>AgentCore</p><p>The buildspec.yml handles the image build and push.</p><p>This gives me a repeatable deployment process and removes the classic:</p><p>“It works on my machine.”</p><p>problem.</p><p><strong>Troubleshooting: Where Things Get Interesting</strong></p><p>The most valuable part of this project wasn’t actually writing the agent.</p><p>It was understanding what could go wrong.</p><p><strong>Container won’t start</strong></p><p>Check:</p><p>Architecture, Docker command, Dependencies, Environment variables</p><p><strong>/ping fails</strong></p><p>The problem is likely the application/container contract.</p><p>Don’t start debugging Bedrock yet.</p><p><strong>AccessDeniedException</strong></p><p>Check the AgentCore execution role.</p><p>Don’t immediately attach:</p><p>AdministratorAccess</p><p>Find the missing API permission instead.</p><p><strong>504 Gateway Timeout</strong></p><p>Work from the outside in:</p><p>AgentCore</p><p>↓</p><p>Container</p><p>↓</p><p>/ping</p><p>↓</p><p>/invocations</p><p>↓</p><p>Bedrock</p><p>↓</p><p>Tool</p><p>↓</p><p>AWS API</p><p><strong>Agent gives an incorrect AWS answer</strong></p><p>The solution isn’t a better prompt alone.</p><p>The agent should use the AWS tool to retrieve current information rather than inventing it.</p><p><strong>Observability Matters</strong></p><p>Once an agent is running in AWS, another question appears:</p><p><strong>Why was this request slow?</strong></p><p>Consider:</p><p>User</p><p>↓</p><p>AgentCore</p><p>↓</p><p>Strands</p><p>↓</p><p>Bedrock</p><p>↓</p><p>Tool selection</p><p>↓</p><p>CloudWatch API</p><p>↓</p><p>Response</p><p>If the request takes 10 seconds, which component caused it?</p><p>Without observability, it’s difficult to know.</p><p>AgentCore integrates with AWS observability capabilities including CloudWatch logs, metrics, and tracing.</p><p>That makes the agent much closer to a real production workload.</p><p><strong>What I Learned</strong></p><p>This project reinforced a few things for me.</p><p><strong>1. AI doesn’t eliminate DevOps</strong></p><p>It adds another layer to it.</p><p>You still need:</p><p>Docker, IAM, CI/CD, Networking, Monitoring, Testing, Security</p><p>Now you also have:</p><p>Models, Agents, Tools, Prompt behavior, Inference latency</p><p><strong>2. Tool permissions are security boundaries</strong></p><p>The model may decide what it wants to do.</p><p>IAM decides what the workload can actually do.</p><p><strong>3. Read-only is a great starting point</strong></p><p>A useful AI DevOps assistant doesn’t need to modify production infrastructure on day one.</p><p>Start with:</p><p>Observe → Analyze → Recommend</p><p>Then eventually introduce:</p><p>Approve → Remediate</p><p>with an explicit policy and human approval layer.</p><p><strong>Final Thoughts</strong></p><p>Here the interesting lesson from this project isn’t really about building a chatbot. It’s about treating an AI agent as <strong>another production workload</strong>. Once an agent has access to AWS APIs, it needs the same engineering discipline as any other cloud application:</p><p><strong>least-privilege IAM, container security, CI/CD, health checks, observability, testing, and controlled access to infrastructure.</strong></p><p>The LLM provides the reasoning layer. The cloud platform provides the execution environment. The DevOps engineer provides the guardrails.</p><p>And that’s where I think <strong>AI + DevOps</strong> becomes particularly interesting.</p><p><strong>GitHub Project</strong></p><p><strong>Repository:</strong> aws-agentcore-devops-agent</p><p><strong>Link: </strong><a href="https://github.com/azizulmaqsud/aws-agentcore-devops-agent.git"><strong>https://github.com/azizulmaqsud/aws-agentcore-devops-agent.git</strong></a></p><p>In this project, I have included the Strands agent, FastAPI server, AWS tools, Dockerfile, ARM64 build, ECR deployment, CodeBuild, IAM policies, tests, deployment scripts, troubleshooting documentation, and cleanup scripts.</p><p>You can try it.</p><p>Thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d447a8dcf4b0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[FinOps Project: An AI-Powered Cost Optimization Platform on AWS]]></title>
            <link>https://medium.com/@azizulmaqsud/aws-finops-project-an-ai-powered-cost-optimization-cloud-platform-in-the-realtime-01f356f7f15d?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/01f356f7f15d</guid>
            <category><![CDATA[aws-cost-optimization]]></category>
            <category><![CDATA[cloud-finops]]></category>
            <category><![CDATA[ai-finops]]></category>
            <category><![CDATA[aws-finops]]></category>
            <category><![CDATA[finops]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Sat, 08 Aug 2026 19:56:27 GMT</pubDate>
            <atom:updated>2026-09-08T02:38:12.031Z</atom:updated>
            <content:encoded><![CDATA[<h3>FinOps Project: An AI-Powered Cost-Optimization Platform on AWS</h3><p><strong>Written by:</strong> Azizul Maqsud<br><strong>Date:</strong> August 08, 2026</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/660/1*aIYSnIPe-vrtowyHjhTMCQ@2x.jpeg" /></figure><p><strong>The Idea Behind The Project</strong></p><p>On a Monday morning, the FinOps engineering team receives the AWS invoice for last month. The number is higher than expected. Not dramatically higher. Just enough to cause concern. The previous month was around $18,000. This month: <strong>$26,700.</strong> Someone asks the obvious question:</p><p>“What changed?”</p><p>But, Nobody has a complete answer.</p><p>The problem isn’t necessarily that AWS is expensive.</p><p>The problem is that the organization cannot clearly answer:</p><ul><li>Which team generated the additional cost?</li><li>Which application caused it?</li><li>Was the increase expected?</li><li>Was it caused by traffic growth?</li><li>Did someone deploy an oversized instance?</li><li>Are there idle resources?</li><li>Did a development environment run all weekend?</li><li>Did storage unexpectedly increase?</li><li>Did data transfer costs spike?</li><li>Are we paying for resources nobody owns?</li><li>How much of the increase can actually be eliminated?</li></ul><p>This is where FinOps becomes an engineering problem.</p><p>Instead of treating the AWS bill as something the finance team reviews at the end of the month, we treat cloud cost as another operational signal.</p><p>Similar to CPU utilization, latency, error rate, and availability, <strong>cloud cost becomes something engineers can measure, monitor, investigate, and optimize.</strong></p><p>That is the idea behind this project.</p><p><strong>1. Project Objective</strong></p><p>Build a realistic FinOps engineering demo that treats AWS cost as an operational signal. The platform explains cost changes, analyzes multiple accounts, detects anomalies, surfaces rightsizing opportunities, calculates unit economics, and supports governed remediation through natural language.</p><p>• AWS Organizations multi-account foundation</p><p>• CUR 2.0 → S3 → Glue/Athena analytics</p><p>• QuickSight FinOps dashboard</p><p>• Cost Explorer and Budgets</p><p>• AWS Cost Anomaly Detection + notifications</p><p>• AWS Compute Optimizer rightsizing</p><p>• Unit economics</p><p>• Amazon Bedrock FinOps agent</p><p>• Optional AgentCore Runtime/Gateway/Memory</p><p>• EventBridge + Lambda automation</p><p>• IAM/SCP/permission-boundary governance</p><p>• Optional Cognito + Amplify UI</p><p><strong>2. Target Architecture</strong></p><p>Organizations / Management Account</p><p>├─ Dev ─ Staging ─ Production ─ Security</p><p>└─ Consolidated Billing</p><p>↓</p><p>CUR 2.0 → S3 → Glue/Athena → QuickSight</p><p>↓</p><p>Cost Explorer + Budgets + Compute Optimizer + Cost Anomaly Detection</p><p>↓</p><p>Amazon Bedrock FinOps Agent</p><p>├─ Cost Explorer Tool</p><p>├─ CUR/Athena Tool</p><p>├─ Compute Optimizer Tool</p><p>├─ Budgets Tool</p><p>├─ Anomaly Tool</p><p>├─ Unit Economics Tool</p><p>└─ Remediation Tool</p><p>↓</p><p>EventBridge → Lambda → Approval/Guardrails → Remediation</p><p>↓</p><p>CloudTrail + Savings Measurement</p><p><strong>3. Prerequisites</strong></p><p>• AWS Organizations with All features if using multi-account governance</p><p>• Management-account billing access</p><p>• CUR 2.0 configured with resource IDs and Athena integration</p><p>• Amazon Bedrock model access enabled in the chosen Region</p><p>• AWS CLI configured</p><p>• Python 3.10+ for Lambda/tools</p><p>• Node.js 18+ for optional frontend</p><p>• Terraform/CDK optional</p><p>• AgentCore CLI/SDK optional for advanced deployment</p><p>aws sts get-caller-identity</p><p><strong>NOTE: </strong>The uploaded source estimates 4–6 hours and about $250/month depending on usage. These are planning estimates only; actual cost depends on CUR, QuickSight, Bedrock, logs, and automation usage.</p><p><strong>4. Phase 1 — Multi-Account Billing Foundation</strong></p><p><strong>4.1 Console: AWS Organizations</strong></p><p>Organizations → Create organization → All features</p><p>2. Create OUs: Dev, Staging, Production, Security</p><p>3. Create/invite member accounts</p><p>4. Move accounts into the correct OUs</p><p>5. Confirm consolidated billing in the management account</p><p><strong>4.2 CLI</strong></p><p>aws organizations create-organization — feature-set ALL</p><p>aws organizations list-accounts</p><p>aws organizations list-roots</p><p>aws organizations list-organizational-units-for-parent — parent-id &lt;ROOT_ID&gt;</p><p><strong>5. Phase 2 — CUR Data Lake</strong></p><p><strong>5.1 Console</strong></p><p>6. Billing and Cost Management → Cost &amp; Usage Reports → Create report</p><p>7. Name: finops-cur</p><p>8. Enable resource IDs and split cost allocation data where needed</p><p>9. Create/use a dedicated S3 bucket and prefix cur/</p><p>10. Choose hourly granularity for the detailed demo</p><p>11. Enable Athena integration</p><p>12. Wait for billing data to populate before validating queries</p><p><strong>5.2 S3/Glue/Athena</strong></p><p>13. Create dedicated CUR S3 bucket with Block Public Access and encryption</p><p>14. Glue → Crawlers → Create crawler</p><p>15. Point crawler to the CUR prefix</p><p>16. Create database finops_db</p><p>17. Run crawler and verify the CUR table in Athena</p><p><strong>5.3 Athena SQL</strong></p><p>— Service cost</p><p>SELECT product_product_name,</p><p>SUM(line_item_unblended_cost) AS total_cost</p><p>FROM finops_db.&lt;cur_table&gt;</p><p>GROUP BY product_product_name</p><p>ORDER BY total_cost DESC;</p><p>— Daily trend</p><p>SELECT DATE(line_item_usage_start_date) AS usage_day,</p><p>SUM(line_item_unblended_cost) AS total_cost</p><p>FROM finops_db.&lt;cur_table&gt;</p><p>GROUP BY DATE(line_item_usage_start_date)</p><p>ORDER BY usage_day;</p><p>— Account cost</p><p>SELECT line_item_usage_account_id AS account_id,</p><p>SUM(line_item_unblended_cost) AS total_cost</p><p>FROM finops_db.&lt;cur_table&gt;</p><p>GROUP BY line_item_usage_account_id</p><p>ORDER BY total_cost DESC;</p><p><strong>5.4 CLI</strong></p><p>aws athena start-query-execution — query-string “SELECT product_product_name, SUM(line_item_unblended_cost) FROM finops_db.&lt;cur_table&gt; GROUP BY product_product_name” — query-execution-context Database=finops_db — result-configuration OutputLocation=s3://&lt;athena-results-bucket&gt;/</p><p><strong>6. Phase 3 — QuickSight FinOps Dashboard</strong></p><p>18. QuickSight → Create analysis → Athena</p><p>19. Connect to finops_db/CUR table</p><p>20. KPI: Total AWS Spend</p><p>21. Line chart: Daily Cost Trend</p><p>22. Bar chart: Service-Level Spending</p><p>23. Account-Level Cost Breakdown</p><p>24. Top Cost Contributors</p><p>25. Optimization Opportunities</p><p>26. Unit Economics such as cost/customer or cost/transaction</p><p>27. Publish/share the dashboard</p><p><strong>7. Phase 4 — Cost Explorer and Budgets</strong></p><p>28. Enable Cost Explorer in the management account if needed</p><p>29. Save views for daily service cost, monthly account cost, and top drivers</p><p>30. Create organization/account/workload budgets</p><p>31. Configure actual and/or forecast thresholds</p><p>32. Send alerts to approved recipients or SNS</p><p><strong>7.1 Cost Explorer Lambda Tool</strong></p><p>import boto3</p><p>ce = boto3.client(“ce”)</p><p>def lambda_handler(event, context):</p><p>return ce.get_cost_and_usage(</p><p>TimePeriod={“Start”: event[“start”], “End”: event[“end”]},</p><p>Granularity=”DAILY”,</p><p>Metrics=[“UnblendedCost”],</p><p>GroupBy=[{“Type”:”DIMENSION”,”Key”:”SERVICE”}]</p><p>)</p><p><strong>8. Phase 5 — Automated Cost Anomaly Detection</strong></p><p><strong>8.1 Native AWS Console</strong></p><p>33. Billing and Cost Management → Cost Anomaly Detection</p><p>34. Create an AWS-services monitor</p><p>35. For Organizations, configure appropriate linked-account monitoring from the management account</p><p>36. Create an alert subscription</p><p>37. Use SNS for individual alerts when needed</p><p>38. Optionally route SNS notifications to supported chat integrations</p><p><strong>8.2 Lambda Investigation</strong></p><p>import boto3</p><p>ce = boto3.client(“ce”)</p><p>def lambda_handler(event, context):</p><p>monitors = ce.get_anomaly_monitors().get(“AnomalyMonitors”, [])</p><p>anomalies = []</p><p>for m in monitors:</p><p>r = ce.get_anomalies(</p><p>MonitorArn=m[“MonitorArn”],</p><p>DateInterval={“StartDate”:event[“start_date”], “EndDate”:event[“end_date”]}</p><p>)</p><p>anomalies.extend(r.get(“Anomalies”, []))</p><p>return {“anomalies”: anomalies}</p><p><strong>8.3 EventBridge Schedule</strong></p><p>aws events put-rule — name daily-finops-anomaly-check — schedule-expression “cron(0 8 * * ? *)”</p><p><strong>NOTE: </strong>AWS Cost Anomaly Detection itself uses machine learning to detect unusual spend. The Lambda layer is an investigation/orchestration layer, not a replacement for the native detector.</p><p><strong>9. Phase 6 — AI-Driven Rightsizing</strong></p><p>39. Compute Optimizer → Get started/opt in</p><p>40. Enable supported resource types such as EC2, Auto Scaling, EBS, Lambda and ECS/Fargate as applicable</p><p>41. Wait for sufficient utilization data</p><p>42. Review finding classification, performance risk, recommendation and estimated savings</p><p>43. Export recommendations if you want historical analysis</p><p><strong>9.1 Lambda Tool</strong></p><p>import boto3</p><p>co = boto3.client(“compute-optimizer”)</p><p>def get_recommendations():</p><p>return {</p><p>“ec2”: co.get_ec2_instance_recommendations().get(“instanceRecommendations”, []),</p><p>“lambda”: co.get_lambda_function_recommendations().get(“lambdaFunctionRecommendations”, [])</p><p>}</p><p><strong>NOTE: </strong>Validate performance risk and workload behavior before applying a rightsizing change.</p><p><strong>10. Phase 7 — Unit Economics</strong></p><p>Create a small workload-metrics dataset that links infrastructure spend to business activity.</p><p>CREATE TABLE finops_unit_economics (</p><p>metric_date date,</p><p>account_id string,</p><p>environment string,</p><p>workload string,</p><p>customers bigint,</p><p>transactions bigint,</p><p>api_requests bigint</p><p>);</p><p>— Derived metrics:</p><p>— cost_per_customer = aws_cost / customers</p><p>— cost_per_transaction = aws_cost / transactions</p><p>— cost_per_request = aws_cost / api_requests</p><p>• Join business metrics to cost by date/account/workload</p><p>• Use tags, cost categories, accounts or workload identifiers for ownership</p><p>• Trend both absolute spend and cost-per-unit in QuickSight</p><p><strong>11. Phase 8 — Amazon Bedrock FinOps Agent</strong></p><p><strong>11.1 Agent Responsibilities</strong></p><p>• Interpret FinOps questions</p><p>• Select the right data/tool source</p><p>• Query cost data</p><p>• Investigate anomalies</p><p>• Retrieve rightsizing recommendations</p><p>• Compare accounts/environments</p><p>• Calculate unit economics</p><p>• Explain evidence and uncertainty</p><p>• Generate recommendations</p><p>• Request approval before sensitive remediation</p><p><strong>11.2 Tool Catalog</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/499/1*IpUqfBnAI1wbSChOLo-zew@2x.jpeg" /></figure><p><strong>11.3 Manual Bedrock Agent Path</strong></p><p>44. Bedrock → Agents → Create Agent</p><p>45. Name: FinOpsAgent</p><p>46. Select a model available in your Region</p><p>47. Add instructions requiring evidence-backed analysis and approval for disruptive actions</p><p>48. Create action groups backed by Lambda for Cost Explorer, Athena, Compute Optimizer, anomaly analysis and remediation</p><p>49. Test each action group independently</p><p>50. Deploy a version/alias before connecting the UI</p><p>You are a FinOps engineering assistant.</p><p>Use tools to retrieve current AWS cost, usage, anomaly, budget and optimization data.</p><p>Separate observed facts from recommendations.</p><p>Prioritize cost impact, business context and performance risk.</p><p>Never execute a disruptive action without required approval.</p><p>For every recommendation, explain reason, expected impact, risk and validation.</p><p><strong>11.4 Advanced AgentCore Path</strong></p><p>For the advanced architecture, use Amazon Bedrock AgentCore Runtime for the agent runtime, Gateway for centralized tool access, and Memory for conversational continuity.</p><p>pip install bedrock-agentcore</p><p>npm install -g @aws/agentcore</p><p>agentcore create — name finops-agent — no-agent</p><p>cd finops-agent</p><p>python -m venv .venv</p><p>source .venv/bin/activate</p><p>agentcore add memory — name FinOpsMemory — strategies SEMANTIC</p><p>agentcore deploy</p><p><strong>NOTE: </strong>Use the current AgentCore CLI/AWS documentation for exact scaffolding and deployment options in your Region. Runtime hosts your agent code; AgentCore primitives such as Memory and Gateway are accessed through the SDK/CLI.</p><p><strong>12. Phase 9 — Event-Driven Automation</strong></p><p>Detect → Investigate → Recommend → Approve → Remediate → Measure</p><p>EventBridge</p><p>↓</p><p>Lambda validation/orchestration</p><p>↓</p><p>Approval/guardrail boundary</p><p>↓</p><p>Controlled remediation</p><p>↓</p><p>CloudTrail + metrics</p><p>↓</p><p>Savings measurement</p><p>• Start with read-only recommendations</p><p>• Require approval for resource-changing actions</p><p>• Use tags to scope eligible workloads</p><p>• Use deny-by-default controls for sensitive operations</p><p>• Record actor, reason, resource, old state and new state</p><p>• Measure actual savings after remediation</p><p><strong>13. Security and Governance</strong></p><p>• Dedicated least-privilege IAM role per tool Lambda</p><p>• Separate read-only analysis roles from write/remediation roles</p><p>• Permission boundaries on automation roles</p><p>• SCPs for prohibited services/regions/actions where appropriate</p><p>• Tags to define automation eligibility</p><p>• CloudTrail for auditability</p><p>• Version-control prompts, policies and infrastructure</p><p>• No unrestricted AI access to AWS</p><p><strong>14. Optional User Interface</strong></p><p>• Cognito for authentication/MFA</p><p>• Amplify-hosted FinOps web interface</p><p>• Dashboard page for QuickSight/analytics</p><p>• Chat page for the Bedrock agent</p><p>• Optimization page for recommendations and approval</p><p>• Audit page for remediation history</p><p><strong>15. Functional Test Plan</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/371/1*ilz4bjrCG4wgzJ1DstOH1A@2x.jpeg" /></figure><p><strong>16. Demo Questions</strong></p><p>• Why did production AWS spend increase this month?</p><p>• Show the top five cost drivers across all accounts.</p><p>• Which account has the largest anomaly?</p><p>• Which EC2 resources are over-provisioned?</p><p>• What is the estimated monthly savings from the top recommendations?</p><p>• What is our cost per transaction?</p><p>• Which recommendations have low performance risk?</p><p>• Which opportunities are safe for automated remediation?</p><p><strong>17. End-to-End Demo Flow</strong></p><p>1. User asks a FinOps question</p><p>2. Bedrock agent interprets intent</p><p>3. Agent selects the appropriate tool</p><p>4. Tool queries Cost Explorer/CUR/Compute Optimizer/Budgets/anomaly data</p><p>5. Agent correlates account, service, workload and unit economics</p><p>6. Agent explains evidence</p><p>7. Agent recommends optimization</p><p>8. Sensitive actions enter approval/guardrail path</p><p>9. EventBridge/Lambda performs approved action</p><p>10. CloudTrail records activity</p><p>11. Cost/workload metrics are measured again</p><p>12. QuickSight reports trend and savings</p><p><strong>18. Repository Structure</strong></p><p>ai-finops-platform/</p><p>├── README.md</p><p>├── docs/architecture.md</p><p>├── docs/console-runbook.md</p><p>├── docs/demo-script.md</p><p>├── docs/security-model.md</p><p>├── lambda/</p><p>│ ├── cost_explorer/</p><p>│ ├── cur_athena/</p><p>│ ├── compute_optimizer/</p><p>│ ├── anomaly/</p><p>│ ├── budgets/</p><p>│ ├── unit_economics/</p><p>│ └── remediation/</p><p>├── agent/</p><p>│ ├── prompts/</p><p>│ ├── tools/</p><p>│ └── policies/</p><p>├── sql/</p><p>├── dashboard/</p><p>├── infrastructure/terraform/</p><p>├── infrastructure/cdk/</p><p>└── tests/</p><p><strong>19. Cleanup</strong></p><p>51. Delete EventBridge rules and demo Lambdas</p><p>52. Remove temporary IAM roles/permissions</p><p>53. Delete Bedrock/AgentCore resources if no longer needed</p><p>54. Delete QuickSight analysis/datasets if lab-only</p><p>55. Remove Athena query-result data</p><p>56. Disable CUR before deleting its S3 data</p><p>57. Remove anomaly monitors/subscriptions</p><p>58. Remove test budgets/SNS subscriptions</p><p>59. Review retained logs and sensitive billing data</p><p><strong>20. Portfolio Deliverable</strong></p><p>The finished demo demonstrates a complete FinOps engineering loop:</p><p>Visibility → Analytics → Anomaly Detection → AI Reasoning → Rightsizing → Unit Economics → Recommendation → Governance → Automation → Measurement</p><p>It is designed to show practical AWS Cloud Engineering, FinOps, AI-agent architecture, DevOps automation and cloud-security skills in one coherent project.</p><p><strong>21. Conclusion</strong></p><p>This project changed the way I look at cloud cost management. Instead of treating the AWS bill as a monthly report, I designed the platform to treat <strong>cost as an engineering signal</strong> that can be collected, analyzed, explained, optimized, and continuously measured.</p><p>By combining multi-account visibility, CUR analytics, anomaly detection, unit economics, Compute Optimizer recommendations, and an AI-driven Amazon Bedrock agent, the workflow moves from simply asking “How much did we spend?” to asking “Why did we spend it, what value did it create, and what should we do next?”</p><p>The most important part is the balance between <strong>AI and control</strong>. The agent can investigate data, connect the dots, and recommend actions, while IAM, governance policies, approval workflows, and controlled automation define what it is actually allowed to do.</p><p>Here this FinOps platform brings <strong>AI, cloud architecture, DevOps, automation, and security</strong> into one practical engineering workflow.</p><p><strong>Collect the data. Understand the cost. Let AI find the opportunity. Govern the action. Measure the result. Then continuously improve.</strong></p><p>Thank You!</p><p>Stay Connected.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=01f356f7f15d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Zero Trust Architecture: Why Modern Security Requires “Never Trust, Always Verify”]]></title>
            <link>https://medium.com/@azizulmaqsud/zero-trust-architecture-why-modern-security-requires-never-trust-always-verify-40dfe61cefe5?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/40dfe61cefe5</guid>
            <category><![CDATA[cyber-security-awareness]]></category>
            <category><![CDATA[comptia-security-plus]]></category>
            <category><![CDATA[comptia-certifications]]></category>
            <category><![CDATA[zero-trust]]></category>
            <category><![CDATA[comptia-security-sy0-601]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Tue, 24 Mar 2026 02:51:01 GMT</pubDate>
            <atom:updated>2026-09-08T02:35:31.007Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Written by:</strong> Azizul Maqsud<br><strong>Date:</strong> March 23, 2026</p><p><strong>Introduction</strong></p><p>For many years, enterprise security relied on a simple philosophy: <em>trust but verify</em>. Organizations protected their networks with perimeter defenses like firewalls and VPNs, assuming that anything inside the network boundary could generally be trusted.</p><p>That assumption no longer holds.</p><p>Cloud computing, remote work, mobile devices, and increasingly complex supply chain attacks have reshaped the threat landscape. Attackers frequently bypass perimeter defenses and move laterally inside networks once they gain initial access.</p><p>Because of this shift, many organizations are adopting <strong>Zero Trust Architecture (ZTA)</strong>. Instead of trusting users or systems based on their location in the network, Zero Trust assumes that <strong>every request must be verified before access is granted</strong>.</p><p>For professionals studying or maintaining the <strong>CompTIA Security+ certification</strong>, Zero Trust concepts appear across several exam domains, particularly those covering security implementation and governance.</p><p><strong>What Zero Trust Really Means</strong></p><p>Zero Trust is a security model built on the assumption that <strong>no user, device, or application should be automatically trusted</strong>.</p><p>Even if a system is inside the corporate network, it must still authenticate and meet security requirements before accessing resources.</p><p>At its core, Zero Trust focuses on <strong>continuous verification and least-privilege access</strong>. Instead of broad network trust, organizations validate identity, device health, and context before allowing access to data or systems.</p><p>Most Zero Trust frameworks focus on several core pillars:</p><p><strong>Identity Security</strong><br> User identities are verified continuously through methods such as multi-factor authentication, adaptive policies, and risk-based authentication.</p><p><strong>Device Security</strong><br> Only devices that meet security standards — such as updated patches and endpoint protection — are allowed to access systems.</p><p><strong>Network Segmentation</strong><br> Networks are divided into smaller segments to limit lateral movement if an attacker compromises one area.</p><p><strong>Application Protection</strong><br> Applications are treated as individual resources with their own authentication and security controls.</p><p><strong>Data Security</strong><br> Sensitive data is protected through encryption, classification, and strict access governance.</p><p><strong>Visibility and Monitoring</strong><br> Security teams continuously monitor activity and analyze behavior to detect anomalies.</p><p><strong>Automation and Response</strong><br> Security policies can automatically respond to suspicious activity, reducing response times.</p><p><strong>How Zero Trust Connects to Security+ Concepts</strong></p><p>Zero Trust principles closely align with several key areas covered in the Security+ certification.</p><p><strong>Secure Network Architecture</strong></p><p>Zero Trust environments rely heavily on segmentation and least-privilege access. These design principles help prevent attackers from moving across networks once a system is compromised.</p><p><strong>Identity and Access Management</strong></p><p>Identity verification plays a central role in Zero Trust. Multi-factor authentication, conditional access policies, and identity monitoring are all essential components.</p><p><strong>Security Governance and Policy</strong></p><p>Zero Trust is not only a technical framework. It also requires strong policies that define how access decisions are made, how incidents are handled, and how compliance requirements are enforced.</p><p><strong>Common Misunderstandings About Zero Trust</strong></p><p>Some organizations believe they have implemented Zero Trust simply because they added multi-factor authentication or deployed a single security tool.</p><p>In reality, Zero Trust is <strong>not a product or single technology</strong>. It is a comprehensive security strategy that combines identity management, network design, monitoring, and policy enforcement.</p><p>Key capabilities typically include:</p><ul><li>Advanced identity verification beyond passwords</li><li>Endpoint posture validation before granting access</li><li>Network micro-segmentation to contain breaches</li><li>Continuous monitoring and behavioral analytics</li></ul><p>Adopting only one of these elements does not create a full Zero Trust environment.</p><p><strong>Practical Steps Toward Implementing Zero Trust</strong></p><p>Organizations often begin their Zero Trust journey with a phased approach.</p><p><strong>Identify Critical Assets</strong></p><p>Instead of trying to secure everything at once, organizations should first identify critical data, applications, and systems that require the highest protection.</p><p><strong>Understand Communication Flows</strong></p><p>Mapping how users, applications, and services interact helps security teams design proper segmentation and access controls.</p><p><strong>Design Access Policies</strong></p><p>Access decisions should be based on identity, device status, location, and other contextual factors.</p><p><strong>Implement Enforcement Points</strong></p><p>Policy enforcement technologies, such as identity proxies or access gateways, ensure that every request is evaluated before access is granted.</p><p><strong>Continuously Monitor and Improve</strong></p><p>Security monitoring and analytics help organizations identify unusual behavior and refine their security policies over time.</p><p><strong>Conclusion</strong></p><p>Zero Trust is becoming a foundational model for modern cybersecurity. As organizations move toward cloud platforms, distributed workforces, and complex digital ecosystems, traditional perimeter defenses are no longer enough. By shifting security toward identity verification, device validation, and continuous monitoring, Zero Trust helps reduce risk and limit the impact of breaches. For Security+ professionals, understanding this model bridges the gap between certification knowledge and real-world security architecture. The traditional perimeter is fading. Modern security requires a new mindset: <strong>never assume trust — always verify.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40dfe61cefe5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ultimate AWS Three-Tier Enterprise Architecture]]></title>
            <link>https://medium.com/@azizulmaqsud/ultimate-aws-three-tier-enterprise-architecture-de822299919c?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/de822299919c</guid>
            <category><![CDATA[aws-community]]></category>
            <category><![CDATA[awscommunitybuilder]]></category>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Thu, 01 Jan 2026 04:49:58 GMT</pubDate>
            <atom:updated>2026-01-01T04:49:58.127Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/584/1*ZnNgykNmNhiE-YHXXp93_Q.png" /></figure><h3>1️⃣ Architecture Diagram</h3><p>· User → Global Accelerator → Route 53 → AWS Shield → WAF → CloudFront → ALB → EC2 (Auto Scaling) → RDS (Private Subnet)</p><p>· Security layers: IAM, KMS, Secrets Manager, Parameter Store</p><p>· DR &amp; Backups: Elastic DR, Multi-AZ RDS, Cross-region snapshots, S3 replication</p><p>· Monitoring: CloudWatch, GuardDuty, CloudTrail, Config</p><h3>2️⃣ VPC &amp; Networking</h3><p><strong>Console:</strong> 1. Create VPC 10.0.0.0/16 2. Subnets: Public 10.0.1.0/24, Private 10.0.2.0/24 3. IGW → Attach 4. NAT Gateway → Public subnet 5. Route Tables → Public → IGW, Private → NAT</p><p><strong>CLI:</strong></p><p>aws ec2 create-vpc — cidr-block 10.0.0.0/16 — tag-specifications ‘ResourceType=vpc,Tags=[{Key=Name,Value=ThreeTierVPC}]’<br> aws ec2 create-subnet — vpc-id &lt;VPC_ID&gt; — cidr-block 10.0.1.0/24<br> aws ec2 create-subnet — vpc-id &lt;VPC_ID&gt; — cidr-block 10.0.2.0/24<br> aws ec2 create-internet-gateway<br> aws ec2 attach-internet-gateway — vpc-id &lt;VPC_ID&gt; — internet-gateway-id &lt;IGW_ID&gt;<br> aws ec2 allocate-address<br> aws ec2 create-nat-gateway — subnet-id &lt;PUBLIC_SUBNET_ID&gt; — allocation-id &lt;EIP_ALLOC_ID&gt;</p><h3>3️⃣ IAM &amp; Security</h3><p><strong>Console:</strong> — IAM Roles: EC2 → Policies: S3ReadOnly, CloudWatchAgent — KMS CMK → Use for EBS, RDS, S3 — Secrets Manager → DB credentials → Rotation — Parameter Store → Encrypted configs</p><p><strong>CLI:</strong></p><p>aws iam create-role — role-name EC2AppRole — assume-role-policy-document file://trust-policy.json<br> aws iam attach-role-policy — role-name EC2AppRole — policy-arn arn:aws:iam::aws:policy/AmazonS3ReadOnlyAccess<br> aws kms create-key — description “ThreeTierCMK”<br> aws secretsmanager create-secret — name DBSecret — secret-string ‘{“username”:”admin”,”password”:”Passw0rd!”}’<br> aws ssm put-parameter — name “/ThreeTierApp/DBEndpoint” — value “mydb.123456789.us-east-1.rds.amazonaws.com” — type SecureString</p><h3>4️⃣ EC2/EKS + Auto Scaling + ALB</h3><p><strong>Console:</strong> — EC2 Launch Template → Attach IAM Role → EBS encrypted — EKS Cluster → Node Groups → Attach IAM Role, enable Auto Scaling — Auto Scaling Group → Multi-AZ → Attach ALB — ALB → HTTPS → ACM certificate</p><p><strong>CLI:</strong></p><p># EC2<br> aws ec2 run-instances — image-id ami-0abcdef1234567890 — count 1 — instance-type t3.micro — key-name MyKeyPair — subnet-id &lt;PUBLIC_SUBNET_ID&gt; — iam-instance-profile Name=EC2EKSRole — block-device-mappings ‘[{“DeviceName”:”/dev/sda1&quot;,”Ebs”:{“VolumeSize”:20,”Encrypted”:true}}]’<br> <br> # EKS<br> aws eks create-cluster — name ThreeTierEKS — role-arn &lt;IAM_ROLE_ARN&gt; — resources-vpc-config subnetIds=&lt;PUBLIC_SUBNET_ID&gt;,securityGroupIds=&lt;SG_ID&gt;<br> aws eks create-nodegroup — cluster-name ThreeTierEKS — nodegroup-name NodeGroup1 — node-role &lt;IAM_ROLE_ARN&gt; — subnets &lt;PUBLIC_SUBNET_ID&gt; — scaling-config minSize=1,maxSize=3,desiredSize=2</p><h3>5️⃣ CloudFront + WAF + Shield + Global Accelerator</h3><p><strong>Console:</strong> — CloudFront → Origin: ALB → Attach WAF &amp; Shield — Global Accelerator → Add ALB endpoint → Enable health checks</p><p><strong>CLI:</strong></p><p>aws cloudfront create-distribution — distribution-config file://cf-distribution.json<br> aws wafv2 create-web-acl — name ThreeTierWAF — scope CLOUDFRONT — default-action Allow — rules file://waf-rules.json — visibility-config SampledRequestsEnabled=true,CloudWatchMetricsEnabled=true,MetricName=ThreeTierWAF<br> aws globalaccelerator create-accelerator — name ThreeTierGA — ip-address-type IPV4 — enabled</p><h3>6️⃣ RDS Multi-AZ + Backups + DR</h3><p><strong>Console:</strong> — RDS → Multi-AZ → KMS encryption → Automated backups → Cross-region snapshots — Elastic DR → Enable EC2 replication to DR region</p><p><strong>CLI:</strong></p><p>aws rds create-db-instance — db-instance-identifier ThreeTierDB — db-instance-class db.t3.micro — engine mysql — master-username admin — master-user-password Passw0rd! — allocated-storage 20 — vpc-security-group-ids &lt;SG_ID&gt; — multi-az — storage-encrypted — kms-key-id &lt;KMS_KEY_ID&gt; — backup-retention-period 7<br> aws rds copy-db-snapshot — source-db-snapshot-identifier &lt;SNAPSHOT_ID&gt; — target-db-snapshot-identifier CrossRegionSnapshot — source-region us-east-1 — kms-key-id &lt;KMS_KEY_ID&gt;</p><h3>7️⃣ S3 + Backup + Replication</h3><p><strong>Console:</strong> — S3 → Versioning + Cross-Region Replication + KMS — AWS Backup → Daily schedule → EC2 + RDS + S3</p><p><strong>CLI:</strong></p><p>aws s3api put-bucket-versioning — bucket my-three-tier-bucket — versioning-configuration Status=Enabled<br> aws s3api put-bucket-replication — bucket my-three-tier-bucket — replication-configuration file://replication.json<br> aws backup create-backup-plan — backup-plan file://backup-plan.json</p><h3>8️⃣ Monitoring &amp; Logging</h3><p><strong>Console:</strong> — CloudWatch → Alarms, metrics — CloudTrail → Enable logging → S3 bucket (encrypted) — GuardDuty → Enable — Config → Track resource changes</p><p><strong>CLI:</strong></p><p>aws cloudwatch put-metric-alarm — alarm-name EC2HighCPU — metric-name CPUUtilization — namespace AWS/EC2 — statistic Average — period 300 — threshold 80 — comparison-operator GreaterThanThreshold — dimensions Name=InstanceId,Value=&lt;INSTANCE_ID&gt; — evaluation-periods 2 — alarm-actions &lt;SNS_TOPIC_ARN&gt;<br> aws cloudtrail create-trail — name ThreeTierTrail — s3-bucket-name my-cloudtrail-bucket — is-multi-region-trail<br> aws cloudtrail start-logging — name ThreeTierTrail<br> aws guardduty create-detector — enable</p><h3>9️⃣ Interview-Ready Talking Points</h3><p>· Multi-layer security: IAM roles, KMS encryption, Secrets Manager, WAF, Shield</p><p>· DR &amp; Backup: Elastic DR, Multi-AZ RDS, S3 replication, AWS Backup</p><p>· Monitoring: CloudWatch, GuardDuty, CloudTrail, Config</p><p>· Auto-healing: ASG, ALB health checks, RDS Multi-AZ failover</p><p>· Global performance: CloudFront + Global Accelerator</p><p><strong>This is a comprehensive end-to-end guide for a new DevOps engineer to deploy, secure, monitor, back up, and demonstrate a fully resilient AWS three-tier architecture.</strong></p><h3>Compiled by: Azizul Maqsud</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=de822299919c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ Realtime Project: AWS Infrastructure Automation]]></title>
            <link>https://medium.com/@azizulmaqsud/realtime-project-aws-infrastructure-automation-6b5c9585ac38?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/6b5c9585ac38</guid>
            <category><![CDATA[aws-lambda]]></category>
            <category><![CDATA[iac]]></category>
            <category><![CDATA[aws-ec2]]></category>
            <category><![CDATA[terraform]]></category>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Fri, 05 Sep 2025 15:09:00 GMT</pubDate>
            <atom:updated>2025-09-05T15:24:19.954Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Industry Standard (us-east-1)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/709/1*DRCxKpMpVf4hKqR9LKHx-g@2x.jpeg" /></figure><p>🏗️ <strong>Project Overview</strong></p><p>This project simulates how modern infrastructure teams design, provision, secure, and operate AWS workloads using Infrastructure as Code, automation, observability, and cost optimization.</p><p>It is divided into 5 modules:</p><p>1.	Networking &amp; Compute → VPC, Subnets, EC2 (Terraform)</p><p>2.	Storage &amp; IAM Security → S3 buckets, IAM policies, least-privilege</p><p>3.	Configuration &amp; Containers → Linux setup, Docker, MySQL container</p><p>4.	Serverless &amp; Automation → AWS Lambda, EventBridge jobs, Python/Bash</p><p>5.	Observability &amp; Cost Optimization → CloudWatch alarms, logging, budgets</p><p>🔑 <strong>Technologies Used</strong></p><p>•	AWS → EC2, VPC, IAM, S3, Lambda, CloudWatch</p><p>•	Terraform → Infrastructure as Code</p><p>•	Linux (Amazon Linux 2) → Server environment</p><p>•	Docker → Containers</p><p>•	MySQL → Database (containerized)</p><p>•	Bash + Python → Automation scripts</p><p>•	Puppet → Optional config management</p><p><strong>📂 Project Structure</strong></p><p>project-name/</p><p>│── modules/</p><p>│. ├── vpc-ec2/</p><p>│. ├── s3-iam/</p><p>│. ├── docker-mysql/</p><p>│. ├── lambda-automation/</p><p>│. └── monitoring-cost/</p><p>│── scripts/</p><p>│. ├── setup_nginx.sh</p><p>│. ├── run_docker_mysql.sh</p><p>│. └── cleanup.sh</p><p>│── terraform/</p><p>│. ├── main.tf</p><p>│. ├── variables.tf</p><p>│. └── outputs.tf</p><p>│── README.md</p><p>🚀 Step-by-Step Implementation</p><p>1️⃣ Networking &amp; Compute (VPC + EC2)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/343/1*3QfmqVRQvtkdjw4Shq8yCw@2x.jpeg" /></figure><p>Terraform example (run in terraform/):</p><p>terraform init</p><p>terraform plan</p><p>terraform apply -auto-approve</p><p>SSH into EC2:</p><p>ssh -i mykey.pem ec2-user@&lt;PUBLIC-IP&gt;</p><p>Install &amp; run Nginx:</p><p>sudo yum update -y</p><p>sudo amazon-linux-extras enable nginx1</p><p>sudo yum install nginx -y</p><p>sudo systemctl start nginx</p><p>sudo systemctl enable nginx</p><p>Validate → http://&lt;PUBLIC-IP&gt;</p><p>2️⃣ Storage &amp; IAM Security</p><p>Create an S3 bucket:</p><p>aws s3 mb s3://cloudops-student-bucket — region us-east-1</p><p>Upload a file:</p><p>echo “Hello CloudOps!” &gt; index.html</p><p>aws s3 cp index.html s3://cloudops-student-bucket/</p><p>Attach IAM Role with least privilege (example policy JSON):</p><p>aws iam create-role — role-name S3ReadOnlyRole \</p><p>. — assume-role-policy-document file://trust-policy.json</p><p>aws iam attach-role-policy — role-name S3ReadOnlyRole \</p><p>. — policy-arn arn:aws:iam::aws:policy/AmazonS3ReadOnlyAccess</p><p>3️⃣ <strong>Configuration &amp; Containers (Docker + MySQL)</strong></p><p>Install Docker on EC2:</p><p>sudo yum install docker -y</p><p>sudo systemctl start docker</p><p>sudo usermod -aG docker ec2-use</p><p>Run MySQL container:</p><p>docker run — name mysql-db -e MYSQL_ROOT_PASSWORD=cloudops123 -d -p 3306:3306 mysql:5.7</p><p>Check container:</p><p>docker ps</p><p>docker logs mysql-db</p><p>4️⃣ Serverless &amp; Automation (Lambda + Bash/Python)</p><p>Example: Python Lambda function to log S3 events.</p><p>Create deployment package:</p><p>mkdir lambda-func &amp;&amp; cd lambda-func</p><p>echo ‘def lambda_handler(event, context): print(“Event:”, event)’ &gt; lambda_function.py</p><p>zip function.zip lambda_function.py</p><p>Deploy Lambda:</p><p>aws lambda create-function \</p><p>. — function-name S3EventLogger \</p><p>. — runtime python3.9 \</p><p>. — role arn:aws:iam::&lt;ACCOUNT-ID&gt;:role/S3ReadOnlyRole \</p><p>. — handler lambda_function.lambda_handler \</p><p>. — zip-file fileb://function.zip</p><p>Create EventBridge rule for daily trigger:</p><p>aws events put-rule — schedule-expression “rate(1 day)” — name DailyTrigger</p><p>aws lambda add-permission \</p><p>. — function-name S3EventLogger \</p><p>. — statement-id EventBridgeInvoke \</p><p>. — action “lambda:InvokeFunction” \</p><p>. — principal events.amazonaws.com \</p><p>. — source-arn arn:aws:events:us-east-1:&lt;ACCOUNT-ID&gt;:rule/DailyTrigger</p><p>5️⃣ Observability &amp; Cost Optimization</p><p>Enable CloudWatch Agent:</p><p>sudo yum install amazon-cloudwatch-agent -y</p><p>sudo /opt/aws/amazon-cloudwatch-agent/bin/amazon-cloudwatch-agent-config-wizard</p><p>sudo systemctl start amazon-cloudwatch-agent</p><p>Set an alarm:</p><p>aws cloudwatch put-metric-alarm \</p><p>. — alarm-name HighCPU \</p><p>. — metric-name CPUUtilization \</p><p>. — namespace AWS/EC2 \</p><p>. — statistic Average \</p><p>. — period 300 \</p><p>. — threshold 70 \</p><p>. — comparison-operator GreaterThanThreshold \</p><p>. — dimensions Name=InstanceId,Value=&lt;EC2-ID&gt; \</p><p>. — evaluation-periods 2 \</p><p>. — alarm-actions &lt;SNS-TOPIC-ARN&gt;</p><p>Cost Budget:</p><p>aws budgets create-budget \</p><p>. — account-id &lt;ACCOUNT-ID&gt; \</p><p>. — budget file://budget.json</p><p>🧹 <strong>Cleanup</strong></p><p>terraform destroy -auto-approve</p><p>aws s3 rb s3://cloudops-student-bucket — force</p><p>aws lambda delete-function — function-name S3EventLogger</p><p>👉 <strong>This project, once fully implemented, mirrors real industry DevOps/CloudOps workflows:</strong></p><p>•	IaC (Terraform)</p><p>•	Security (IAM least privilege, patching, TLS readiness)</p><p>•	Containers (Docker + MySQL)</p><p>•	Serverless (Lambda + Event-driven automation)</p><p>•	Observability &amp; Cost control</p><p><strong>Compiled by: Azizul Maqsud</strong></p><p><strong><em>Let’s connect @ YouTube</em></strong></p><p><a href="https://youtube.com/@Tech_with_Azizul-Maqsud">Tech With Azizul</a></p><p>@ LinkedIn</p><p><a href="https://www.linkedin.com/in/azizul-maqsud?utm_source=share&amp;utm_campaign=share_via&amp;utm_content=profile&amp;utm_medium=ios_app">https://www.linkedin.com/in/azizul-maqsud?utm_source=share&amp;utm_campaign=share_via&amp;utm_content=profile&amp;utm_medium=ios_app</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6b5c9585ac38" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Istio Service Mesh: Complete Guide with Realtime Implementation]]></title>
            <link>https://medium.com/@azizulmaqsud/istio-service-mesh-complete-guide-with-realtime-implementation-cd7fc192048d?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd7fc192048d</guid>
            <category><![CDATA[istio]]></category>
            <category><![CDATA[istio-service-mesh]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[continuous-deployment]]></category>
            <category><![CDATA[canary-deployments]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Fri, 05 Sep 2025 13:00:17 GMT</pubDate>
            <atom:updated>2025-09-05T13:00:17.263Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*LOIMq9rkg70_nLm24Gfa6A@2x.jpeg" /></figure><ol><li><strong>What is Istio?</strong></li></ol><p>Istio is an open-source service mesh that provides a unified way to secure, connect, and monitor microservices. As a CNCF-graduated project, it solves critical challenges in distributed systems by adding a dedicated infrastructure layer without requiring application code changes.</p><p><strong>Core Architecture:</strong></p><p>· Data Plane: Envoy proxies deployed as sidecars alongside each service</p><p>· Control Plane: istiod manages configuration, security, and discovery</p><p>· Key Components: Gateways, VirtualServices, DestinationRules</p><p>2. <strong>Key Features</strong></p><p>Traffic Management</p><p>· Advanced routing (canary releases, A/B testing, blue-green deployments)</p><p>· Load balancing with multiple algorithms</p><p>· Fault injection for resilience testing</p><p>· Circuit breaking and retry policies</p><p><strong>Security</strong></p><p>· Automatic mTLS encryption between services</p><p>· Fine-grained access control with authorization policies</p><p>· JWT-based authentication</p><p>· Automatic certificate management</p><p><strong>Observability</strong></p><p>· Detailed metrics (latency, traffic volume, errors)</p><p>· Distributed tracing across service boundaries</p><p>· Comprehensive access logging</p><p>· Integration with Prometheus, Grafana, Jaeger</p><p>3. <strong>Realtime Project: Bookinfo Application</strong></p><p><strong>Application Components</strong></p><p>· productpage (Python): Main page that calls details and reviews</p><p>· details (Ruby): Book information service</p><p>· reviews (Java): Reviews service with three versions</p><p>· ratings (Node.js): Book ranking service</p><p><strong>Deployment Steps</strong></p><ol><li><strong>Enable sidecar injection:</strong></li></ol><p>kubectl label namespace default istio-injection=enabled</p><p>2. Deploy application:</p><p>kubectl apply -f samples/bookinfo/platform/kube/bookinfo.yaml</p><p>3. Configure gateway for external access:</p><p>kubectl apply -f samples/bookinfo/networking/bookinfo-gateway.yaml</p><p>4. Apply destination rules:</p><p>kubectl apply -f samples/bookinfo/networking/destination-rule-all.yaml</p><p><strong>Verification</strong></p><p>Check all pods have sidecar proxies (2/2 containers ready):</p><p>kubectl get pods</p><p><strong>NAME — READY — STATUS</strong></p><p>details-v1–558b8b4b76–2llld. 2/2. Running</p><p>productpage-v1–6987489c74-lpkgl. 2/2. Running</p><p>ratings-v1–7dc98c7588-vzftc. 2/2. Running</p><p>reviews-v1–7f99cc4496-gdxfn. 2/2. Running</p><p>reviews-v2–7d79d5bd5d-8zzqd. 2/2. Running</p><p>reviews-v3–7dbcdcbc56-m8dph. 2/2. Running</p><p>4. <strong>Practical Use Cases</strong></p><p><strong>Canary Release Implementation</strong></p><p>apiVersion: networking.istio.io/v1alpha3</p><p>kind: VirtualService</p><p>metadata:</p><p>name: reviews</p><p>spec:</p><p>hosts:</p><p>— reviews</p><p>http:</p><p>— route:</p><p>— destination:</p><p>host: reviews</p><p>subset: v1</p><p>weight: 90</p><p>destination:</p><p>host: reviews</p><p>subset: v2</p><p>weight: 10</p><p>```</p><p><strong>Fault Injection for Testing</strong></p><p>apiVersion: networking.istio.io/v1alpha3</p><p>kind: VirtualService</p><p>metadata:</p><p>. name: ratings</p><p>spec:</p><p>. hosts:</p><p>. — ratings</p><p>. http:</p><p>. — fault:</p><p>. delay:</p><p>. percentage:</p><p>. value: 10.0</p><p>. fixedDelay: 5s</p><p>. route:</p><p>. — destination:</p><p>. host: ratings</p><p>. subset: v1</p><p>```</p><p><strong>Security Policy</strong></p><p>apiVersion: security.istio.io/v1beta1</p><p>kind: AuthorizationPolicy</p><p>metadata:</p><p>. name: require-productpage-to-reviews</p><p>spec:</p><p>. selector:</p><p>. matchLabels:</p><p>. app: reviews</p><p>. action: ALLOW</p><p>. rules:</p><p>. — from:</p><p>. — source:</p><p>. principals: [“cluster.local/ns/default/sa/productpage-service-account”]</p><p>5. <strong>Performance Considerations</strong></p><p>· Latency overhead: 2–10ms per request</p><p>· Resource consumption: Each sidecar requires CPU/memory</p><p>· Optimization strategies:</p><p>. · Tune proxy resources</p><p>. · Use selective deployment</p><p>. · Simplify configuration</p><p>. · Implement HTTP/2</p><p>6. <strong>Alternatives and Future Trends</strong></p><p>Competing Solutions</p><p>· Linkerd: Lighter-weight alternative</p><p>· Consul Connect: Integrated with HashiCorp ecosystem</p><p>· AWS App Mesh: Managed AWS service</p><p><strong>Evolution</strong></p><p>· Istio ambient mesh: Simplified architecture</p><p>· Platform engineering integration</p><p>· Enhanced API security capabilities</p><p>7. <strong>Conclusion</strong></p><p>Istio provides enterprise-grade service mesh capabilities with advanced traffic management, zero-trust security, and comprehensive observability. While it introduces some complexity and overhead, the benefits for microservices environments typically justify the investment.</p><p>Here, the Bookinfo project demonstrates practical implementation of canary deployments, fault injection, and security policies. For organizations operating complex microservices architectures, Istio offers critical capabilities for maintaining reliability, security, and visibility at scale.</p><p><strong>Best Practices:</strong></p><p>· Start with non-critical applications</p><p>· Implement features gradually</p><p>· Monitor performance overhead</p><p>· Invest in team training</p><p>· Use phased adoption strategy</p><p><strong>Documented by: Azizul Maqsud</strong></p><p><strong><em>Let’s connect @ YouTube</em></strong></p><p>https://youtube.com/@Tech_with_Azizul-Maqsud</p><p><strong><em>@Linkedin</em></strong></p><p>https://www.linkedin.com/feed/update/urn:li:activity:7369695977022107651?utm_source=share&amp;utm_medium=member_ios&amp;rcm=ACoAADFImp4Bxx-CWmP7MsSJSi9VdZjZJBPkTlY</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd7fc192048d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building a Three-Tier Serverless Enterprise App on AWS]]></title>
            <link>https://medium.com/@azizulmaqsud/building-a-three-tier-serverless-enterprise-app-on-aws-ccd9e55e264a?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/ccd9e55e264a</guid>
            <category><![CDATA[serverless-architecture]]></category>
            <category><![CDATA[aws-lambda]]></category>
            <category><![CDATA[serverless]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[dynamodb]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Thu, 04 Sep 2025 13:56:15 GMT</pubDate>
            <atom:updated>2025-09-04T14:16:27.554Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>End-to-End Project</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*b1Q7JEilxMOlQAxVCU_q7Q@2x.jpeg" /></figure><p>Cloud adoption is growing fast, and serverless architecture is one of the most in-demand skills for cloud engineers, DevOps practitioners, and software developers. In this article, let’s walk through an end-to-end project using AWS services to build a three-tier enterprise-ready serverless application.</p><p>🎯 <strong>Why Three-Tier Architecture?</strong></p><p>Three-tier architecture separates an app into:</p><p>1.	Presentation Tier — Frontend/UI (Web or Mobile).</p><p>2.	Application Tier — Business logic (APIs, processing).</p><p>3.	Data Tier — Persistent storage.</p><p>On AWS, we can build this fully serverless, so we pay only for what we use.</p><p>🏗️ <strong>Architecture Overview</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*B9-le_1TmIkI0BoABg1haw@2x.jpeg" /></figure><p>Here’s what we’ll use:</p><p>•	Presentation Tier → Amazon Cognito (authentication), S3 static hosting, CloudFront (CDN).</p><p>•	Application Tier → API Gateway, AWS Lambda, Amazon SQS, Step Functions.</p><p>•	Data Tier → DynamoDB (NoSQL, on-demand).</p><p>📌 All components are event-driven, fully managed, and scale automatically.</p><p>🛠️ Step-by-Step Implementation</p><p>1️⃣ Data Tier — DynamoDB</p><p>•	Create a table: serverless-3tier-items</p><p>•	Keys: pk (Partition Key), sk (Sort Key)</p><p>•	Billing: On-Demand</p><p>•	Enable point-in-time recovery for backups.</p><p>2️⃣ Presentation Tier — Cognito &amp; Static Web</p><p>•	Cognito User Pool for user authentication.</p><p>•	Create a User Pool client (no secret, OAuth code flow).</p><p>•	Configure callback URLs for local testing or your domain.</p><p>•	S3 Bucket to host index.html (your web app).</p><p>•	CloudFront Distribution to serve globally with HTTPS.</p><p>3️⃣ Application Tier — Lambda + API Gateway + SQS</p><p>•	Create Lambdas:</p><p>o	GetItems: Reads from DynamoDB.</p><p>o	PutItem: Writes item &amp; sends message to SQS.</p><p>o	AsyncWorker: Processes messages from SQS.</p><p>•	Set Environment Variables:</p><p>o	TABLE_NAME = serverless-3tier-items</p><p>o	QUEUE_URL = &lt;SQS URL&gt;</p><p>•	Permissions:</p><p>o	DynamoDB read/write</p><p>o	SQS send/consume</p><p>•	API Gateway:</p><p>o	Routes: GET /items, PUT /items</p><p>o	Integrations: Map to Lambdas</p><p>o	JWT Authorizer: Protect APIs with Cognito</p><p>4️⃣ Async Processing — SQS + Worker Lambda</p><p>•	Create SQS Standard Queue serverless-3tier-work.</p><p>•	Trigger AsyncWorker Lambda from the queue.</p><p>•	This ensures background jobs don’t block API calls.</p><p>5️⃣ Frontend Integration</p><p>•	Upload your static web app (React, Angular, or simple HTML/JS) to S3.</p><p>•	Use Amplify SDK or amazon-cognito-identity-js to:</p><p>o	Log in users.</p><p>o	Get JWT tokens.</p><p>o	Call API Gateway with Authorization: Bearer &lt;token&gt;.</p><p>🔍 Testing</p><p>•	Sign up a test user in Cognito.</p><p>•	Log in → Get JWT.</p><p>•	Test APIs with Postman or curl:</p><p>curl -X PUT “$API/items” \</p><p>. -H “Authorization: Bearer $TOKEN” \</p><p>. -d ‘{“pk”:”PUBLIC”,”data”:{“hello”:”world”}}’</p><p>•	Confirm items are saved in DynamoDB &amp; processed by AsyncWorker.</p><p>🧩 Why This Is Enterprise-Ready</p><p>✔️ Authentication &amp; Authorization (Cognito)</p><p>✔️ Secure APIs (JWT via API Gateway)</p><p>✔️ Decoupled async processing (SQS + Lambda)</p><p>✔️ Reliable persistence (DynamoDB with PITR)</p><p>✔️ Scalable frontend delivery (CloudFront + S3)</p><p>✔️ Cost-efficient pay-per-use model</p><p>🎓 <strong>Final Notes for Students</strong></p><p>This project is resume-ready and a realistic cloud use case. By building it, you’ll gain hands-on experience with:</p><p>•	Identity &amp; Access (Cognito)</p><p>•	Event-driven compute (Lambda, API Gateway, SQS)</p><p>•	NoSQL data modeling (DynamoDB)</p><p>•	Global app delivery (CloudFront)</p><p>📢 Pro-tip: Extend this with CI/CD pipelines, custom domains, or monitoring dashboards to showcase advanced skills.</p><p>👉 If you’re a student looking to get into Cloud &amp; DevOps roles, mastering serverless projects like this will help you stand out in job interviews. Best wishes!</p><p>Thank you 🙏</p><p><strong>Compiled by: Azizul Maqsud</strong></p><p>Let’s connect @ YouTube</p><p><a href="https://youtube.com/@Tech_with_Azizul-Maqsud">Tech With Azizul</a></p><p>@linkedin</p><p>https://www.linkedin.com/feed/update/urn:li:activity:7369364557829468160?utm_source=share&amp;utm_medium=member_ios&amp;rcm=ACoAADFImp4Bxx-CWmP7MsSJSi9VdZjZJBPkTlY</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ccd9e55e264a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Deploying Amazon FSx for NetApp ONTAP with EKS & EC2]]></title>
            <link>https://medium.com/@azizulmaqsud/deploying-amazon-fsx-for-netapp-ontap-with-eks-ec2-7c2374ae3c14?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/7c2374ae3c14</guid>
            <category><![CDATA[aws-efs]]></category>
            <category><![CDATA[amazon-fsx]]></category>
            <category><![CDATA[netapp]]></category>
            <category><![CDATA[file-storage]]></category>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Wed, 03 Sep 2025 13:23:22 GMT</pubDate>
            <atom:updated>2025-09-04T14:18:07.130Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*BnufPWVXAA_ibTVvfXKFXA@2x.jpeg" /></figure><p>In 2025, enterprise workloads demand flexible, reliable, and high-performance storage. That’s exactly what Amazon FSx for NetApp ONTAP (FSx ONTAP) delivers — the familiar power of NetApp’s ONTAP file system, now fully managed on AWS.</p><p>🚀 <strong>Hands-on Guide:</strong></p><p>In this step-by-step guide, we’ll walk through how to:</p><p>✅ Deploy FSx ONTAP in AWS (console-based setup)</p><p>✅ Connect it with Amazon EKS for container workloads</p><p>✅ Mount the same storage from EC2 for operations</p><p>✅ Use snapshots &amp; clones for dev/test agility</p><p>🔹 <strong>Why FSx for NetApp ONTAP?</strong></p><p>FSx ONTAP brings the best of on-prem enterprise storage to AWS:</p><p>•	Multi-protocol support (NFS, SMB, iSCSI)</p><p>•	Automatic tiering to Amazon S3 for cost optimization</p><p>•	Snapshots &amp; cloning for near-instant test/dev environments</p><p>•	High availability with Multi-AZ deployment</p><p>•	Seamless integration with EKS, ECS, EC2, and RDS</p><p>🔹 <strong>Step 1 — Create FSx ONTAP File System</strong></p><p>1.	Open AWS Console → FSx → Create File System</p><p>2.	Choose FSx for NetApp ONTAP → Quick Create</p><p>3.	Configure basics:</p><p>o	Deployment: Multi-AZ</p><p>o	Storage: 1 TiB</p><p>o	Throughput: 512 MBps</p><p>o	Admin password (fsxadmin)</p><p>4.	Choose your VPC &amp; subnets (same as EKS/EC2).</p><p>5.	Adjust security groups to allow NFS (2049), SMB (445), RPC (111,635).</p><p>🔹 <strong>Step 2 — Set up Amazon EKS Cluster</strong></p><p>1.	Go to EKS Console → Create Cluster.</p><p>2.	Name: fsxn-eks-demo</p><p>3.	Select your VPC &amp; subnets (same as FSx).</p><p>4.	Add a Node Group (e.g., t3.medium × 2 — 3 nodes).</p><p>5.	Verify the nodes are READY and can reach FSx SG.</p><p>🔹 <strong>Step 3 — Install NetApp Trident CSI</strong></p><p>1.	In EKS Console → Add-ons, search NetApp Trident CSI.</p><p>2.	Install the operator.</p><p>3.	Verify with:</p><p>kubectl get pods -n trident</p><p>🔹 <strong>Step 4 — Connect Trident to FSx</strong></p><p>1.	In Secrets Manager, create a secret with FSx SVM credentials:</p><p>{ “username”: “vsadmin”, “password”: “YourPassword” }</p><p>2.	Apply a TridentBackendConfig to connect to FSx:</p><p>apiVersion: trident.netapp.io/v1</p><p>kind: TridentBackendConfig</p><p>metadata:</p><p>. name: fsxn-nfs</p><p>spec:</p><p>. version: 1</p><p>. storageDriverName: ontap-nas</p><p>. backendName: fsxn-nfs</p><p>. svm: svm01</p><p>. managementLIF: &lt;SVM-MGMT-DNS&gt;</p><p>. dataLIF: &lt;SVM-NFS-DNS&gt;</p><p>. autoExportPolicy: true</p><p>. credentials:</p><p>. name: “arn:aws:secretsmanager:region:account:secret:fsxn-credentials”</p><p>. type: awsarn</p><p>🔹 Step 5 — Create StorageClass &amp; PVC</p><p>StorageClass</p><p>apiVersion: storage.k8s.io/v1</p><p>kind: StorageClass</p><p>metadata:</p><p>. name: fsxn-sc</p><p>provisioner: csi.trident.netapp.io</p><p>allowVolumeExpansion: true</p><p>reclaimPolicy: Delete</p><p>PersistentVolumeClaim</p><p>apiVersion: v1</p><p>kind: PersistentVolumeClaim</p><p>metadata:</p><p>. name: webdata</p><p>spec:</p><p>. accessModes: [ “ReadWriteMany” ]</p><p>. resources:</p><p>. requests:</p><p>. storage: 20Gi</p><p>. storageClassName: fsxn-sc</p><p>🔹 <strong>Step 6 — Deploy Nginx App with FSx Storage</strong></p><p>apiVersion: apps/v1</p><p>kind: Deployment</p><p>metadata:</p><p>. name: nginx-fsxn</p><p>spec:</p><p>. replicas: 2</p><p>. selector:</p><p>. matchLabels:</p><p>. app: nginx-fsxn</p><p>. template:</p><p>. metadata:</p><p>. labels:</p><p>. app: nginx-fsxn</p><p>. spec:</p><p>. containers:</p><p>. — name: nginx</p><p>. image: nginx:stable</p><p>. ports:</p><p>. — containerPort: 80</p><p>. volumeMounts:</p><p>. — name: content</p><p>. mountPath: /usr/share/nginx/html</p><p>. volumes:</p><p>. — name: content</p><p>. persistentVolumeClaim:</p><p>. claimName: webdata</p><p><strong>Verify app:</strong></p><p>kubectl exec -it deploy/nginx-fsxn — sh -c ‘echo “Hello FSx ONTAP” &gt; /usr/share/nginx/html/index.html’</p><p>kubectl port-forward deploy/nginx-fsxn 8080:80</p><p>curl localhost:8080</p><p>🔹 <strong>Step 7 — Mount from EC2</strong></p><p>1.	<strong>Launch an EC2 instance in the same VPC.</strong></p><p>2.	<strong>Install NFS client:</strong></p><p>sudo yum install -y nfs-utils</p><p>3.	<strong>Mount FSx:</strong></p><p>sudo mkdir -p /mnt/apps</p><p>sudo mount -t nfs4 &lt;SVM-NFS-DNS&gt;:/apps /mnt/apps</p><p>ls -l /mnt/apps</p><p>✅ You’ll see the same files written by nginx pods.</p><p>🔹 Step 8 — Snapshots &amp; Clones</p><p>•	Go to FSx Console → Volumes → Create Snapshot.</p><p>•	Create snapshot webdata-snap.</p><p>•	Clone snapshot into a new volume → mount to another PVC for test/dev.</p><p>🎯 <strong>Conclusion</strong></p><p>With this setup you get:</p><p>•	FSx ONTAP as managed enterprise storage</p><p>•	Dynamic volumes for Kubernetes (EKS)</p><p>•	Shared access from EC2 + containers</p><p>•	Instant snapshots/clones for agile DevOps</p><p>This is cloud-native enterprise storage that accelerates both traditional and modern workloads.</p><p>🔗 Try it out, and let me know how FSx ONTAP fits into your architecture!</p><p>PDF Link:</p><p>https://www.linkedin.com/feed/update/urn:li:activity:7368994453556031490?utm_source=share&amp;utm_medium=member_ios&amp;rcm=ACoAADFImp4Bxx-CWmP7MsSJSi9VdZjZJBPkTlY</p><p>#AWS #NetApp #FSxONTAP #EKS #DevOps #CloudStorage #Architecture</p><p><strong>Compiled by: Azizul Maqsud</strong></p><p><em>Let’s connect @</em>YouTube:</p><p><a href="https://m.youtube.com/@Tech_with_Azizul-Maqsud">Tech With Azizul</a></p><p>LinkedIn:</p><p><a href="https://www.linkedin.com/in/azizul-maqsud">https://www.linkedin.com/in/azizul-maqsud</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7c2374ae3c14" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Kubernetes 7 Common Errors: The Essential Troubleshooting Guide!]]></title>
            <link>https://medium.com/@azizulmaqsud/kubernetes-7-common-errors-the-essential-troubleshooting-guide-e81a0e2c61b7?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/e81a0e2c61b7</guid>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[kubernetes-operator]]></category>
            <category><![CDATA[kubernetes-cluster]]></category>
            <category><![CDATA[kubernetes-security]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Fri, 29 Aug 2025 12:45:35 GMT</pubDate>
            <atom:updated>2025-08-29T12:45:35.530Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*shO77lhLEpave-slh0OS4Q@2x.jpeg" /></figure><p>Navigating a Kubernetes cluster can feel like commanding a fleet. When everything works, it’s powerful. When it doesn’t, the errors can be cryptic. This guide demystifies the most common Kubernetes errors, providing a clear action plan for each.</p><ol><li><strong>CrashLoopBackOff</strong></li></ol><p>The Problem: Your pod starts, crashes, and Kubernetes restarts it in an endless loop. The Fix:</p><p>· First Command: kubectl logs &lt;pod-name&gt; — previous</p><p>· Diagnose: kubectl describe pod &lt;pod-name&gt; (Check Events)</p><p>· Likely Culprits: Application bugs, misconfigured environment variables, missing files, or failing liveness probes.</p><p>2. <strong>ImagePullBackOff / ErrImagePull</strong></p><p>The Problem: Kubernetes can’t retrieve your container image. The Fix:</p><p>· Check: Image name, tag, and registry for typos.</p><p>· Verify: Access to private registries using imagePullSecrets.</p><p>· Test: Manually pull the image on a node: docker pull &lt;image:tag&gt;.</p><p>3. <strong>Pending</strong></p><p>The Problem: The scheduler can’t find a suitable node for your pod. The Fix:</p><p>· Investigate: kubectl describe pod &lt;pod-name&gt; — the Events will tell you why.</p><p>· Check: Resource quotas (kubectl top nodes) and Node Selectors / Taints &amp; Tolerations.</p><p>4. <strong>RunContainerError / </strong>CreateContainerConfigError</p><p>The Problem: The container fails at the very moment of creation. The Fix:</p><p>· For `RunContainerError`: Check your container’s command and args; often a typo or wrong path.</p><p>· For `CreateContainerConfigError`: A missing Secret or ConfigMap. Use kubectl describe pod to find the exact missing resource.</p><p>5. <strong>OOMKilled (Out of Memory)</strong></p><p>The Problem: Your container exceeded its memory limit and was killed. The Fix:</p><p>· Short-term: Increase the memory limit in your pod spec.</p><p>· Long-term: Optimize your application’s memory usage and set appropriate requests.</p><p>6. <strong>LivenessProbe Failed</strong></p><p>The Problem: Kubernetes keeps restarting your container because it thinks it’s unhealthy. The Fix:</p><p>· Check: Is the health check endpoint (path, port) correct and responsive?</p><p>· Tune: Increase initialDelaySeconds and adjust timeoutSeconds to match your app’s startup time.</p><p>7. <strong>NodeNotReady</strong></p><p>The Problem: A worker node is down, causing pod evictions. The Fix:</p><p>· Diagnose: kubectl describe node &lt;node-name&gt; for issues like MemoryPressure or DiskPressure.</p><p>· SSH into the node and check:</p><p>. · systemctl status kubelet (Is it running?)</p><p>. · journalctl -u kubelet (For logs)</p><p>. · df -h (For disk space)</p><p><strong>When in doubt, run through these steps:</strong></p><ol><li>Describe it: kubectl describe pod &lt;pod-name&gt; (Look at Events)</li><li>2. Log it: kubectl logs &lt;pod-name&gt;</li><li>3. Check Dependencies: Verify all Secrets, ConfigMaps, and PVCs exist.</li><li>4. Validate YAML: kubectl apply — dry-run=client -f your-file.yaml</li></ol><p>Mastery isn’t about preventing errors — it’s about solving them efficiently.</p><p><strong>Compiled by: Azizul Maqsud.</strong></p><p><strong>Connect for more insights like this.</strong></p><p>#devops #kubernetes #cloudnative #interviewprep</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e81a0e2c61b7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Golang : A Comprehensive Tutorial]]></title>
            <link>https://medium.com/@azizulmaqsud/golang-a-comprehensive-tutorial-2f1afab30598?source=rss-4df114cb0773------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f1afab30598</guid>
            <category><![CDATA[golang-tools]]></category>
            <category><![CDATA[golang-tutorial]]></category>
            <category><![CDATA[go]]></category>
            <category><![CDATA[golang]]></category>
            <category><![CDATA[golang-development]]></category>
            <dc:creator><![CDATA[Azizul maqsud]]></dc:creator>
            <pubDate>Wed, 04 Oct 2023 15:26:52 GMT</pubDate>
            <atom:updated>2023-10-06T12:19:32.334Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/946/1*15_NIbrRTDyCcTQOtzn-sw@2x.jpeg" /></figure><p>Go, often referred to as Golang, is a modern and versatile programming language created by Google engineers Robert Griesemer, Rob Pike, and Ken Thompson in 2007. Designed with simplicity and efficiency in mind, Go has gained popularity for its clean syntax, strong concurrency support, and robust standard library.</p><p>In this introduction, we’ll explore the key features and advantages of Go, as well as its suitability for various application domains, making it a language of choice for both seasoned developers and newcomers to the programming world. Whether you’re interested in web development, systems programming, or building scalable and concurrent applications, Go has something to offer. Let’s dive into the world of Go and discover why it has become such a compelling language in the world of software development.</p><h4>A “Hello, World!” program in Golang</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*bwJ7MJFFFTwX5Gyb-EFaNQ@2x.jpeg" /></figure><h4>Basic Go concepts</h4><p>In basic if Golang we find variables, constants, data types, operators, and control flow. Here is an example of if else:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*giHDD3wjokqXb8Jb4UD9nQ@2x.jpeg" /></figure><h4>Functions and Packages</h4><p>An example to define a function, create a custom package, and importing and using functions from the package.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*pmOglnZmXQGxghr2xniznQ@2x.jpeg" /></figure><h4>Object-Oriented Programming in Go</h4><p>Here is a simple struct representing a person, define methods for that struct, and demonstrated interface implementation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*6eYxtqhYpTV0ML8_7ZEEQg@2x.jpeg" /></figure><h4>Concurrency in Go</h4><p>Here is a simple concurrent program that calculates the sum of numbers using Goroutines and channels.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*RBdulTby_rGsBAEancchYw@2x.jpeg" /></figure><h4>Working with Data</h4><p>Here is an example of data structures in Go, including error handling:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*dazCZFi9PhoU4uBhZvIi4Q@2x.jpeg" /></figure><h4>Building Web Applications in Go</h4><p>An example of web development in Go using the package, routing, handling HTTP requests and responses, and creating a simple web server hat responds with “Hello, World!” to incoming requests.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*sweG4tuTCLkxLnMlyMF0OQ@2x.jpeg" /></figure><h4>Testing and Debugging</h4><p>Here is a simple unit test for a function with basic debugging using `fmt.Println()`.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*siBmVLJBqOJidZ_uyspe6g@2x.jpeg" /></figure><h4>Advanced Go Topics</h4><p>Here is a reflection to inspect struct fields and use Go’s context package to handle timeouts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*uUCNbUZSAuVVPU5LHJTWHQ@2x.jpeg" /></figure><h3>Future Predictions for Go</h3><p>The future of Golang is immense! It’s relevance in emerging technologies, and its adoption in various industries is important. Here is some predictions.</p><h4>Anticipated Language Features:</h4><p>1. Generics: Go is actively working on adding generics, which would allow developers to write more flexible and reusable code. For example:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/645/1*j85ATr-03cckLL-o1JtEDw@2x.jpeg" /></figure><p>2. Error Handling Enhancements: Go may introduce improvements in error handling, making it even more robust and developer-friendly.</p><h4>Role in Emerging Technologies:</h4><p>1. WebAssembly (Wasm) Integration: Go has made strides in supporting WebAssembly, enabling developers to write Go code that runs in web browsers. This opens up possibilities for building frontend applications in Go.</p><p>2. Edge Computing: As edge computing gains prominence, Go’s efficiency and low resource consumption make it an excellent choice for building applications that run on edge devices.</p><p>3. Machine Learning and AI: While not a primary focus, Go’s simplicity and performance may lead to increased adoption in machine learning and AI applications.</p><h4>Industry Adoption:</h4><p>1. Cloud-Native Development: Go has a strong presence in the cloud-native ecosystem, and this is expected to continue. Kubernetes and Docker, two prominent cloud-native technologies, are written in Go.</p><p>2. Finance and Fintech: Go’s speed and efficiency make it well-suited for the financial industry, where low-latency systems are crucial.</p><p>3. Cybersecurity: With its focus on performance and concurrency, Go is increasingly used in the development of security tools and applications.</p><p>4. Healthcare: Go’s reliability and ease of deployment make it a viable choice for healthcare applications, particularly in telemedicine and health data analytics.</p><h3>Conclusion</h3><p>In this chapter, we’ve explored some of the potential directions and scenarios for the future of Go. As the Go community and development team continue to evolve the language, we can anticipate exciting developments and increased adoption across various domains.</p><p>While these are predictions and potential scenarios, the actual future of Go may vary depending on community contributions and the evolving technology landscape.</p><p>Remember that the future of a programming language is shaped by the needs of its users and the challenges of the technology landscape. Stay engaged with the Go community, follow updates from the Go team, and be prepared to adapt and leverage Go’s capabilities in emerging areas of technology.</p><h4>References</h4><p>1. “The Go Programming Language” by Alan A. A. Donovan and Brian W. Kernighan — This book offers an in-depth exploration of Go, covering all aspects of the language with clear explanations and practical examples.</p><p>2. “Concurrency in Go” by Katherine Cox-Buday — A valuable resource for understanding Go’s concurrency features and patterns.</p><h4>Stay Connected for more..</h4><p>https://www.youtube.com/channel/UCNwP7KEElaJ7cdDTLP-KbBg</p><p>https://www.linkedin.com/in/azizul-maqsud/</p><p>https://azizulmaqsud-1684501031000.hashnode.dev/</p><p>https://medium.com/@azizulmaqsud</p><p>https://twitter.com/Sohail2me</p><p>https://github.com/azizulmaqsud</p><p></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f1afab30598" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>