<?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 Mukund Kumar Jha on Medium]]></title>
        <description><![CDATA[Stories by Mukund Kumar Jha on Medium]]></description>
        <link>https://medium.com/@mukundkumarjh?source=rss-1265aa499033------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*mJmXuwQGAveuRmU19jerxA.jpeg</url>
            <title>Stories by Mukund Kumar Jha on Medium</title>
            <link>https://medium.com/@mukundkumarjh?source=rss-1265aa499033------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 17:16:36 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mukundkumarjh/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 Multi-AZ Architecture with Public and Private Subnets]]></title>
            <link>https://medium.com/@mukundkumarjh/building-multi-az-architecture-with-public-and-private-subnets-1b1260ef9dfd?source=rss-1265aa499033------2</link>
            <guid isPermaLink="false">https://medium.com/p/1b1260ef9dfd</guid>
            <category><![CDATA[aws-certification]]></category>
            <category><![CDATA[aws-ec2]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[aws-vpc]]></category>
            <dc:creator><![CDATA[Mukund Kumar Jha]]></dc:creator>
            <pubDate>Wed, 22 Jul 2026 21:04:39 GMT</pubDate>
            <atom:updated>2026-07-22T21:04:39.871Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*T-io_aLkrDoIll09oL_2ow.png" /><figcaption>VPC</figcaption></figure><p>Amazon Virtual Private Cloud (VPC) is the networking foundation of AWS. Every resource you launch in AWS lives inside a VPC, and understanding how to design one properly is a critical skill for any cloud engineer.</p><ul><li>VPC and CIDR planning</li><li>Public vs Private subnets</li><li>Internet Gateway and NAT Gateway</li><li>Route tables and traffic flow</li><li>How EC2 instances connect in each subnet</li></ul><p>By the end, you will not only understand the theory — you’ll see real screenshots from my AWS console so you can follow along yourself.</p><p>Our VPC uses the CIDR range 32.0.0.0/16. This gives us 65,536 IP addresses — plenty for most use cases.</p><p>A /16 CIDR means the first 16 bits are fixed (32.0), and the remaining 16 bits are available for subnetting. This is the most common size for production VPCs.</p><pre>VPC CIDR : 32.0.0.0/16<br>Available IPs : 65,536</pre><h3>Two Availability Zones</h3><p>AZ 1 — ap-south-1a (aps1-az1)</p><ul><li>Contains the Public Subnet (32.0.1.0/24)</li><li>Houses: EC2 Instance (bastion/jump box), Internet Gateway, NAT Gateway</li></ul><p>AZ 2 — ap-south-1b (aps1-az3)</p><ul><li>Contains the Private Subnet (32.0.2.0/24)</li><li>Houses: EC2 Instance (application/database server)</li></ul><h3>Public Subnet — Your Gateway to the Internet</h3><p>The Public Subnet (32.0.1.0/24) is where resources that need direct internet access live.</p><h3>3.1 Internet Gateway (IGW)</h3><p>The Internet Gateway is a horizontally scaled, redundant, and highly available VPC component that allows communication between your VPC and the internet.</p><ul><li>It’s a managed AWS service — no maintenance needed</li><li>One IGW per VPC</li><li>It serves as the target in your route table for 0.0.0.0/0 (all internet traffic)</li></ul><h3>EC2 Instance in Public Subnet</h3><p>The EC2 instance in the public subnet typically serves as a Bastion Host (Jump Box) — the only way to SSH/RDP into private instances.</p><p>Security considerations:</p><ul><li>Restrict SSH access to your IP only</li><li>Use a key pair for authentication</li><li>Optionally use AWS Systems Manager Session Manager instead.</li></ul><h3>NAT Gateway — Bridging Private Subnets to the Internet</h3><p>The NAT Gateway sits in the public subnet but serves the private subnet.</p><h3>What does a NAT Gateway do?</h3><p>A NAT Gateway allows instances in the private subnet to initiate outbound connections to the internet (for software updates, API calls, etc.) while preventing the internet from initiating inbound connections to those instances.</p><p>Private EC2 → NAT Gateway (in Public Subnet) → Internet Gateway → Internet</p><ul><li>Resides in a public subnet (needs an Elastic IP)</li><li>Translates the private IP of the source instance to its own Elastic IP</li><li>Forwards the response back to the private instance</li></ul><h3>Private Subnet — Keeping Your Workloads Secure</h3><p>The Private Subnet (32.0.2.0/24) hosts resources that should never be directly accessible from the internet.</p><h3>EC2 Instance in Private Subnet</h3><p>This EC2 instance:</p><ul><li>Has no public IP address</li><li>Cannot be reached directly from the internet</li><li>Can still download packages, call APIs, etc. via the NAT Gateway</li><li>Can only be accessed via the Bastion Host (public EC2)</li></ul><h3>Clean Up — Don’t Get Billed!</h3><p>Don’t forget to delete resources when you’re done:</p><ul><li>Terminate EC2 instances</li><li>Delete NAT Gateway (releases EIP)</li><li>Detach and delete Internet Gateway</li><li>Delete subnets</li><li>Delete VPC</li></ul><p><em>If you found this helpful, please clap 👏 and share it with your network. Drop questions in the comments — I reply to every one.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1b1260ef9dfd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure VNet Integration Access Private Resources Securely Without Exposing Them to the Internet.]]></title>
            <link>https://medium.com/@mukundkumarjh/azure-vnet-integration-access-private-resources-securely-without-exposing-them-to-the-internet-de74c9d3d36d?source=rss-1265aa499033------2</link>
            <guid isPermaLink="false">https://medium.com/p/de74c9d3d36d</guid>
            <category><![CDATA[azure-devops]]></category>
            <category><![CDATA[cloud]]></category>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-networking]]></category>
            <category><![CDATA[vnet-integration]]></category>
            <dc:creator><![CDATA[Mukund Kumar Jha]]></dc:creator>
            <pubDate>Sun, 21 Jun 2026 12:13:16 GMT</pubDate>
            <atom:updated>2026-06-21T12:13:16.925Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aFdGUcQ6ZUwt_Up9c54m-w.png" /><figcaption>Azure Vnet Integration</figcaption></figure><h3>What Is VNet Integration and Why Does It Exist?</h3><p>By default, your Azure App Service (or Function App) is a multi-tenant service sitting on shared infrastructure. When your app needs to reach a resource like a SQL Database, a Redis Cache, or a VM — if those resources are locked down inside a Virtual Network (VNet) — your app can’t reach them. They’re private. They have no public endpoint.</p><p>VNet Integration is the bridge. It lets your app make outbound calls into a VNet as if it were a resource living inside that network. Think of it like giving your app a VIP pass into a private club — it still stands outside (your app’s inbound URL remains public), but it can now talk to anyone inside.</p><p>To picture this, imagine an end user sends a request to https://mynewwebapp1.azurewebsites.net/. The App Service proxy intercepts that request, and through VNet Integration, routes it into the virtual network where the actual backend resources — an Nginx VM protected by a Network Security Group (NSG) and delegated subnets — sit. Without VNet Integration, those backend resources would either need to be exposed to the internet or the app would never reach them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JX0j1orgwDfb3I-oWQSLwQ.png" /><figcaption>Nginx</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/777/1*6QsDOVWAdrJUvgeMu4OAHQ.png" /><figcaption>Reversed Proxy Code</figcaption></figure><h3>The Benefits: Why You Should Care</h3><h3>1. Security — Keeping Things Off the Public Internet</h3><p>This is the big one. When you integrate your app with a VNet, you can lock down your backend resources so they only accept traffic from within the VNet. Your database doesn’t need a public IP. Your storage account can block all internet traffic. Your app talks to them privately, and the public internet never gets a look-in.</p><p>That Network Security Group (NSG) in the diagram acts as a “security guard” — it filters traffic before anything reaches the Nginx VM. That’s exactly the kind of control VNet Integration gives you.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IRsb8t8NplaYloFJYHeWLw.png" /><figcaption>App Services</figcaption></figure><h3>2. Compliance and Regulatory Requirements</h3><p>Many industries — finance, healthcare, government — require that data never traverses the public internet. VNet Integration is often the simplest path to meeting those requirements with PaaS services.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/867/1*C-Jgs2Ut2OGdCPV1PX5zWA.png" /><figcaption>Vnet Network Integration</figcaption></figure><h3>3. Hybrid Connectivity</h3><p>Once your app is integrated into a VNet, it can also reach on-premises resources through VPN Gateway or ExpressRoute. Your cloud app can query an on-prem database as if it were sitting in the same building.</p><h3>4. Simplified Network Architecture</h3><p>Instead of punching firewall holes or setting up complex NAT rules, you just enable VNet Integration, and your app can resolve private IPs and access resources by their internal addresses. It’s clean, predictable, and manageable.</p><h3>The Challenges: What Nobody Tells You</h3><p>Let’s be honest — VNet Integration isn’t all sunshine. Here’s where people usually hit walls.</p><h3>1. Outbound-Only (Regional Integration)</h3><p>Regional VNet Integration is outbound only. Your app can talk to resources inside the VNet, but nothing in the VNet can initiate a connection <em>to</em> your app through that integration. If you need inbound traffic from the VNet (say, an internal load balancer sending requests to your app), you need a Private Endpoint or an App Service Environment.</p><p>This trips up a lot of people. In the diagram, the proxy server handles the inbound request, and VNet Integration handles the outbound leg — two different mechanisms working together.</p><h3>2. Routing Complexity</h3><p>When you enable VNet Integration, all outbound traffic from your app can be routed through the VNet — including traffic to the internet. If you’ve configured forced tunneling (all traffic goes through a firewall), your app’s outbound internet traffic now goes through your firewall too, which can cause unexpected latency or even break things if the firewall blocks certain traffic.</p><h3>3. Subnet Requirements</h3><p>VNet Integration requires a dedicated subnet in your VNet. It must be delegated to Microsoft.Web/serverFarms. And here&#39;s the kicker — once you assign a subnet to integration, you can&#39;t resize it without downtime. Plan your IP space carefully. I&#39;ve seen teams run out of IPs because they sized their subnet at /27 and forgot they&#39;d need room for scale-out.</p><p>In the diagram, notice how subnet-2 is explicitly labeled &quot;delegated_subnets&quot; — that&#39;s the dedicated subnet where the integration magic happens.</p><h3>4. Regional vs. Gateway-Required</h3><p>There are two flavors of VNet Integration:</p><ul><li>Regional — uses a delegated subnet, supported on newer App Service plans (Premium v2/v3, Elastic Premium for Functions). It’s faster and simpler.</li><li>Gateway-required — older approach, uses a VPN gateway in the VNet. Slower setup, more moving parts.</li></ul><p>Choosing the wrong one for your scenario can cause headaches later. Regional is almost always the right choice for new deployments.</p><h3>5. DNS Resolution</h3><p>Once your app is inside the VNet, it needs to resolve private DNS names. If your VNet uses custom DNS servers or Azure Private DNS Zones, your app needs to be configured correctly to use them. Miss this step, and your app will try to resolve internal resource names via public DNS and fail silently.</p><h3>6. Debugging Is Hard</h3><p>When something breaks, tracing the issue is harder than with public endpoints. You can’t just curl a URL and see what happens. You need to check Route Tables, NSG rules, subnet delegation, and App Service logs. It’s a learning curve.</p><h3>When Should You Use VNet Integration?</h3><p>Use it when:</p><ul><li>Your app needs to access Azure resources that are locked inside a VNet</li><li>You need to meet compliance requirements for network isolation</li><li>You’re connecting to on-premises resources through VPN or ExpressRoute</li><li>You want to avoid exposing backend resources to the internet</li></ul><p><em>Thanks for reading! If you found this helpful, clap and follow for more Azure and cloud infrastructure stories.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=de74c9d3d36d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Journey Setting Up Azure Application Gateway with WAF — And Why Sequential Traffic Distribution…]]></title>
            <link>https://medium.com/@mukundkumarjh/my-journey-setting-up-azure-application-gateway-with-waf-and-why-sequential-traffic-distribution-5c40e8e6ccaf?source=rss-1265aa499033------2</link>
            <guid isPermaLink="false">https://medium.com/p/5c40e8e6ccaf</guid>
            <category><![CDATA[azure-devops]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[microsoft-azure]]></category>
            <category><![CDATA[azure-application-gateway]]></category>
            <category><![CDATA[networking]]></category>
            <dc:creator><![CDATA[Mukund Kumar Jha]]></dc:creator>
            <pubDate>Tue, 16 Jun 2026 11:54:58 GMT</pubDate>
            <atom:updated>2026-06-16T11:58:56.848Z</atom:updated>
            <content:encoded><![CDATA[<h3>My Journey Setting Up Azure Application Gateway with WAF — And Why Sequential Traffic Distribution Matters</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GwRkg8e57yWHc7TejLAgSw.png" /><figcaption>Complete Architecture</figcaption></figure><h3>The Problem I Was Trying to Solve</h3><p>A few weeks ago, I was working on a project where I needed to host a web application that could handle traffic securely and efficiently. I had two virtual machines (VMs) running Nginx in Azure, each sitting in its own subnet inside a virtual network (VNet). My goal was simple:</p><blockquote><em>Users should be able to hit a single public IP, and the traffic should be distributed across my two VMs — but in a sequential order, not round-robin or random.</em></blockquote><blockquote>On top of that, I wanted security. I didn’t want malicious requests reaching my VMs at all. I needed a Web Application Firewall (WAF) to inspect and filter incoming traffic before it even touched my application.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/554/1*RhLm5Rhq2WRonETaysfkAQ.png" /><figcaption>Web server 1</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/567/1*eVaSCfiIqtt1DPeboiau2Q.png" /><figcaption>web server 2</figcaption></figure><h3>The Public IP and Entry Point</h3><p>The user sends an HTTP request to a public IP address — in my case, 49.216.173.56. This IP is directly attached to my Application Gateway. Nobody accesses my VMs directly. Everything goes through this single entry point.</p><p>Why is this good? Because I control and secure all traffic from one central place. No one can bypass the system and hit my VMs directly (as long as the Network Security Groups are properly configured, which we’ll cover later).</p><p>This is the heart of my setup. Instead of having two separate services, Azure allows me to enable WAF (Web Application Firewall) directly on the Application Gateway itself. It’s all one service.</p><p>Think of it as a smart security guard who does two jobs at once:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vKkQz4umNUcU5OQknyym6A.png" /><figcaption>Prevention Mode</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/997/1*j1t-ha-9V6pQl_-YNPx7Cw.png" /><figcaption>XSS vulnerability</figcaption></figure><h3>Job 1 — The Bouncer (WAF)</h3><p>It checks every request coming in:</p><ul><li>Is it a SQL injection attack? → Blocked</li><li>Is it a cross-site scripting (XSS) attempt? → Blocked</li><li>Is it a normal, legitimate request? → Allowed</li><li>Is the request rate too high from one IP? → Throttled</li></ul><p>WAF filters out the bad stuff before it ever gets a chance to harm my servers.</p><h3>Job 2 — The Traffic Manager (Application Gateway)</h3><p>After the WAF check, the Application Gateway handles the routing. It’s a layer 7 load balancer — which is just a fancy way of saying it understands HTTP/HTTPS traffic, not just raw network packets. It can:</p><ul><li>Inspect URL paths and route traffic accordingly</li><li>Terminate SSL (so my backend VMs don’t have to handle encryption)</li><li>Distribute traffic across the backend pool</li><li>Run health checks to know which VMs are alive</li></ul><p>For my setup, I configured the Application Gateway to send traffic to a backend pool containing both VMs. And here’s the important part — I set it up for sequential distribution.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PovJmvpx3tPVfJr33l37Pg.png" /><figcaption>Backend pool</figcaption></figure><p>My backend pool consists of two VMs, each in its own subnet:</p><p>VMSubnetIP AddressRole vm1 Subnet 110.0.1.4 Primary — gets traffic first vm2 Subnet 210.0.2.4 Secondary — gets traffic when vm1 is busy</p><p>Each VM runs Nginx — a lightweight and powerful web server. I chose Nginx because it’s reliable, easy to configure, and handles concurrent connections beautifully.</p><p>The reason I placed them in separate subnets is for isolation and security. If one subnet has a problem (like a network misconfiguration or an attack), the other is not affected.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/933/1*-1QqIIh38HTJ7fE_uwI2nA.png" /><figcaption>NSG Rule</figcaption></figure><h3>Network Security Groups (NSGs)</h3><p>Each subnet has an NSG attached. NSGs act as a firewall at the subnet level. They control what traffic is allowed in and out.</p><p>For my setup:</p><ul><li>Inbound rules: Only allow traffic from the Application Gateway’s subnet. No direct internet access to the VMs.</li><li>Outbound rules: Allow necessary outbound traffic (like updates, logs, etc.)</li></ul><p>This ensures that even if someone somehow discovers the private IPs of my VMs, they can’t reach them directly. Traffic must flow through the Application Gateway and WAF.</p><h3>The Request Flow</h3><ol><li>User types http://49.216.173.56 in their browser</li><li>The request hits the Application Gateway (which has WAF enabled)</li><li>The WAF inspects the request — headers, body, query parameters, everything</li><li>WAF determines the request is safe → forwards it to the Application Gateway’s routing engine</li><li>The Application Gateway checks its rules and sends the request to the backend pool</li><li>Since vm1 is available (sequential distribution), the request goes to Subnet 1 → vm1 (10.0.1.4)</li><li>The NSG on Subnet 1 checks the traffic source — it’s from the Application Gateway, so it’s allowed</li><li>Nginx on vm1 processes the request and sends back the response</li><li>The response travels back through the Application Gateway to the user</li></ol><h3>Final Thoughts</h3><p>This setup — Application Gateway with WAF + backend VMs in separate subnets with NSGs — has been a game-changer for my project. Traffic is secure, distribution is predictable, and I sleep better knowing that SQL injection attempts get caught before they even reach my application.</p><p>Some architectures throw in Front Door, Traffic Manager, or multiple load balancers. But for my use case, keeping it simple with just a WAF-enabled Application Gateway and two backend VMs was the right call. Not every solution needs to be complicated.</p><p>If you’re building something similar in Azure, I highly recommend this architecture. It’s production-grade without being over-engineered.</p><p>Have you set up something similar? Or are you planning to? Let me know in the comments — I’d love to hear how your architecture looks!</p><p><em>Thanks for reading! If you found this helpful, clap and follow for more Azure and cloud infrastructure stories.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5c40e8e6ccaf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Traffic Manager: DNS-Based Global Load Balancing Explained]]></title>
            <link>https://medium.com/@mukundkumarjh/azure-traffic-manager-dns-based-global-load-balancing-explained-96c912b7acac?source=rss-1265aa499033------2</link>
            <guid isPermaLink="false">https://medium.com/p/96c912b7acac</guid>
            <dc:creator><![CDATA[Mukund Kumar Jha]]></dc:creator>
            <pubDate>Sat, 13 Jun 2026 18:21:29 GMT</pubDate>
            <atom:updated>2026-06-13T18:21:29.373Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1Bwsm3TJbpiDF96wJamaag.png" /><figcaption>Traffic Manager Load balancer</figcaption></figure><p>If you’ve worked with Azure, you’ve likely heard of Traffic Manager. But here’s the thing that catches most people off guard: Traffic Manager is not a traditional load balancer.</p><p>It doesn’t sit between the user and your application. It doesn’t inspect packets. It doesn’t forward traffic. In fact, once the DNS resolution is done, Traffic Manager steps completely out of the picture.</p><p>So what does it actually do? And more importantly, when should you use it?</p><p>Let’s break it down — starting with a simple diagram-based scenario that illustrates how Traffic Manager works in practice.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*u8lVI1ZYkfURk0PkovsEaA.png" /><figcaption>Traffic Manager</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4d3AX6dhrkJnpI1uq_4llA.png" /><figcaption>Traffic Manager Endpoints</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gSygV2mSLmlWrhdRB8slOA.png" /><figcaption>Vm1 &amp; vm2</figcaption></figure><h3>The Architecture: A Visual Walkthrough</h3><p>Imagine you’re running an NGINX web application across two Azure regions:</p><ul><li>Primary region: East US — this is your main production site</li><li>Secondary region: Japan East — this is your disaster recovery (DR) site</li></ul><p>Each region runs your application on a virtual machine. You want users to reach your app via a friendly URL like http://mytrafficmanager.trafficmanager.net, without having to remember individual VM IPs or deal with regional failover manually.</p><p>This is exactly where Traffic Manager shines.</p><p>Here’s how the flow works:</p><ol><li>A user sends a request to <a href="http://mytrafficmanager.trafficmanager.net/">http://mytrafficmanager.trafficmanager.net</a></li><li>Traffic Manager receives the DNS query</li><li>Based on the routing method configured (in this case, Priority), it resolves the domain to the appropriate endpoint IP</li><li>Traffic Manager returns the IP address of the primary VM (East US) to the user</li><li>The user’s browser then connects directly to that VM</li></ol><p>Crucially, step 5 happens without Traffic Manager in the data path. The user communicates directly with the VM. Traffic Manager only handled the DNS resolution.</p><h3>Priority Routing: The Failover Pattern</h3><p>The diagram shows exactly this scenario with endpoints configured like this:</p><p>EndpointRegionPrioritymyvm1-ip → myvm189.eastus.cloudapp.azure.comEast US1 (Primary)myvm2-ip → myvm245.japaneast.cloudapp.azure.comJapan East2 (Secondary)</p><p>With Priority routing, Traffic Manager sends all traffic to the highest-priority endpoint (Priority 1 — East US) as long as it remains healthy. If the primary endpoint goes down — whether due to a regional outage, VM failure, or application crash — Traffic Manager automatically redirects all traffic to the next available endpoint (Priority 2 — Japan East).</p><p>This gives you a clean, automated active-passive failover pattern.</p><h3>The Six Routing Methods</h3><p>Traffic Manager supports six routing methods, each suited to different scenarios:</p><ol><li>Priority — Active-passive failover (what we covered above)</li><li>Weighted — Distribute traffic across endpoints with defined weights (e.g., 80% to primary, 20% to canary)</li><li>Performance — Route users to the endpoint with the lowest latency</li><li>Geographic — Route users based on their geographic location (useful for data sovereignty)</li><li>Subnet — Route traffic from specific IP ranges to designated endpoints</li><li>Multivalue — Return multiple healthy endpoints in a single DNS response (client picks one)</li></ol><h3>When to Use Traffic Manager</h3><p>Traffic Manager is ideal for:</p><ul><li>Disaster recovery: Automatic failover between regions</li><li>Global application distribution: Route users to the closest or fastest region</li><li>Maintenance windows: Gracefully shift traffic for planned updates</li><li>Blue-green deployments: Use Weighted routing to gradually shift traffic</li><li>Multi-region high availability: Combine with Azure Load Balancer in each region for a complete HA strategy</li></ul><p>It’s not suitable for:</p><ul><li>Session-level load balancing within a single region (use Azure Load Balancer)</li><li>HTTP-specific features like SSL offloading or WAF (use Application Gateway)</li><li>Real-time traffic inspection or modification.</li></ul><blockquote>Thanks for reading! If you found this helpful, stay tuned for the next post where I’ll walk through the Azure Portal — showing you exactly how to configure Traffic Manager, add endpoints, and set up Priority routing step by step.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=96c912b7acac" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Securing Azure Storage with Private Endpoints]]></title>
            <link>https://medium.com/@mukundkumarjh/securing-azure-storage-with-private-endpoints-6c0c49804fdb?source=rss-1265aa499033------2</link>
            <guid isPermaLink="false">https://medium.com/p/6c0c49804fdb</guid>
            <dc:creator><![CDATA[Mukund Kumar Jha]]></dc:creator>
            <pubDate>Sat, 06 Jun 2026 13:36:08 GMT</pubDate>
            <atom:updated>2026-06-06T13:36:08.497Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WsjB8ptgFT2pks_kcDWyqA.png" /><figcaption>vnet</figcaption></figure><p>When your applications live inside Azure and need to access Blob Storage, the default path goes over the public internet. Even with encryption enabled, the endpoint is still publicly resolvable — a risk that compliance auditors will flag.</p><p>Enter Azure Private Endpoints.</p><p>Your architecture diagram shows exactly this pattern: a VM (MY_VM1 at 10.0.0.6) inside an Azure VNet accessing Blob Storage entirely over the Microsoft backbone network, without ever traversing the public internet.</p><h3>The Problem: Why Not Just Use the Public Endpoint?</h3><p>By default, mystorageaccount.blob.core.windows.net resolves to a public IP. Even with firewall rules, this is a potential data exfiltration vector.</p><p>The alternatives each have trade-offs:</p><ul><li>Service Endpoints: Free and easy, but still use public IPs with no exfiltration protection</li><li>Firewall + IP allowlist: Simple, but creates IP management overhead and doesn’t eliminate public exposure</li><li>Private Endpoints (winner): Traffic stays on Microsoft backbone; granular control; requires DNS config and has nominal cost</li></ul><h3>Architecture, Explained</h3><p>ComponentPrivate IPPublic IPRoleMY_VM110.0.0.6NoneApp VM making storage requestsPrivate Endpoint NIC10.0.0.5NoneConnects storage to VNetFirewall10.0.1.6823.101.70.213Outbound internet proxyBlob StorageBackboneN/ATarget: v1demoacconts</p><h4>Traffic Flow</h4><p>MY_VM1 resolves v1demoacconts.privatelink.blob.core.windows.net to 10.0.0.5. Traffic flows VM -&gt; Private Endpoint NIC -&gt; Azure Backbone -&gt; Blob Storage. Zero exposure to the public internet.</p><h3>DNS: The Make-or-Break Piece</h3><p>This is where most implementations fail.</p><p>What you need:</p><ol><li>A Private DNS Zone: privatelink.blob.core.windows.net</li><li>An A record: v1demoacconts -&gt; 10.0.0.5</li><li>The zone linked to your VNet</li></ol><p>Verification:</p><pre>bash</pre><pre>nslookup v1demoacconts.blob.core.windows.net<br># Should return:<br># Name: v1demoacconts.privatelink.blob.core.windows.net<br># Address: 10.0.0.5</pre><p>If it resolves to a public IP instead, traffic leaves your VNet.</p><h3>The Firewall: Defense in Depth</h3><p>The Firewall at 10.0.1.68 with a route table sending 0.0.0.0/0 traffic to it. This handles outbound internet (updates, packages), while private endpoint traffic completely bypasses it — because DNS resolves to a VNet-internal IP.</p><p>The two paths coexist beautifully:</p><ul><li>Storage traffic: VM -&gt; Private Endpoint -&gt; Backbone (no firewall)</li><li>Internet traffic: VM -&gt; Route Table -&gt; Firewall -&gt; Internet</li></ul><h3>Implementation Steps</h3><ol><li>Create your Storage Account with public access disabled</li><li>Create a Private Endpoint in the same region, selecting blob as the sub-resource</li><li>Configure Private DNS: Azure can auto-create the Private DNS Zone — let it</li><li>Verify DNS resolution from your VM (the nslookup test above)</li><li>Set up the Firewall (Azure Firewall or NVAs) for outbound internet traffic</li><li>Test the end-to-end path: upload test.txt (as shown in your diagram) and verify access from the VM</li></ol><h3>Key Takeaways</h3><ul><li>Private Endpoints keep storage traffic entirely off the public internet</li><li>DNS configuration is the most common failure point — always verify</li><li>A firewall alongside private endpoints provides defense-in-depth</li><li>The Azure backbone network handles the heavy lifting between the endpoint and the storage services.</li></ul><h4><em>If you found this article helpful, please clap, leave a comment, and follow me for more content like this.</em></h4><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6c0c49804fdb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How does Facebook serve billions of requests per second ?]]></title>
            <link>https://medium.com/@mukundkumarjh/how-does-facebook-serve-billions-of-requests-per-second-de2cd40c4a46?source=rss-1265aa499033------2</link>
            <guid isPermaLink="false">https://medium.com/p/de2cd40c4a46</guid>
            <dc:creator><![CDATA[Mukund Kumar Jha]]></dc:creator>
            <pubDate>Sun, 21 Dec 2025 07:49:35 GMT</pubDate>
            <atom:updated>2025-12-21T15:56:01.264Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/180/1*-FMG557Fh_fzDm4YR5xh1A.jpeg" /><figcaption>Facebook</figcaption></figure><p>According to reports, billions of users worldwide access Facebook to communicate with friends or engage in business and marketing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8BSvgc4i9i1dQ9H4CCtdYg.png" /><figcaption>Facebook Users Statics</figcaption></figure><p>From a technical standpoint, Facebook is viewed as a marvel of real-time engineering.</p><p>Two things are expected of the platform: it must never go down, and it must never be slow.</p><p>Any disruption would not only result in a direct loss of revenue but would also shatter user trust.</p><p>Because the platform’s billions of users are interconnected in real-time, a poor experience for one often ripples out to affect others.</p><p>This global giant had humble beginnings. In 2004, Mark Zuckerberg, then a student at Harvard University, built the first version of the site entirely from his dormitory in Kirkland House.</p><p>Originally, ‘TheFacebook’ was exclusive, accessible only to Harvard students before it eventually expanded to connect the world.</p><p>In the early days of 2004, the traffic Facebook handled was minuscule compared to the massive global load it supports today.</p><p>The original architecture was a simple <strong>Client-Server model</strong>. Users would send a request from their browser, the server would process it, and a response would be fetched from a single <strong>MySQL database</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/900/1*nQop75SY7RuC-ObeKKqoTw.png" /><figcaption>Client-server-Architecture</figcaption></figure><p>At that time, MySQL was the dominant relational database in the industry because it was flexible, easy to use, and allowed for seamless data querying. Facebook was built using the <strong>LAMP Stack</strong>, which served as the foundation for many of the world’s most successful websites</p><p>LAMP stands for the L -Linux Operating system A- Apache Web server M — MYSQL P- PHP programming Language .</p><p>In the beginning, the simple LAMP architecture was perfect for handling low traffic. Mark Zuckerberg reportedly leased a single server for around <strong>$85 USD per month</strong>, which was plenty for a few thousand Harvard students. However, by the end of 2004 and into 2005, Facebook exploded in popularity.</p><p>This rapid growth brought massive technical challenges. The single-server setup was constantly at risk of crashing due to being overloaded. Zuckerberg and his team realized that a single server has physical limits on the number of requests it can handle concurrently.</p><p>Even though the platform didn’t have 100 billion users yet, the engineering team had to start thinking about <strong>‘Horizontal Scaling.’</strong> They knew that to support a global audience, they had to move away from a single server and find a way to distribute the load across thousands of machines to prevent the system from breaking user trust</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jCSjoW_1BSEgkXdZN3wtIQ.png" /><figcaption>Horizontal Scaling</figcaption></figure><p>As Facebook grew beyond Harvard, Zuckerberg faced a massive technical hurdle the <strong>‘Mutual Friends’</strong> problem. To solve the scaling issue, one engineer suggested partitioning the database by college — essentially creating a separate database for each school.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/465/1*-_8ulSWH8lGmZ9eQCHjPyA.png" /><figcaption>Mutual Friends</figcaption></figure><p>While this seemed like an easy fix for storage, Zuckerberg realized it wouldn’t work for the user experience. If a student at Harvard wanted to find mutual friends with a student at Yale, the system would have to perform a ‘Cross-DB Join.’</p><p>Searching for <strong>‘Friends of Friends’</strong> (3rd or 4th-level deep connections) becomes exponentially slower in a relational database. In a standard MySQL setup, these deep queries require complex ‘JOIN’ operations that would eventually lag and crash the server as the network of people grew more interconnected.</p><p>To overcome these scaling hurdles, Zuckerberg and his engineering team eventually moved toward a <strong>Graph-based architecture</strong> and implemented <strong>Memcached</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Xuzj91O2HNQunto2qhYxZA.png" /><figcaption>Memcached</figcaption></figure><p>Today, Memcached plays a vital role at Facebook, allowing the system to handle billions of requests per second by storing frequently accessed data in high-speed memory.</p><p>To manage this massive global traffic, Facebook utilizes the <strong>Katran Load Balancer</strong>. Katran ensures that every request is routed to the healthiest available server, preventing any single machine from becoming a bottleneck.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8uwhENmvSztIjy9uGifUCQ.png" /><figcaption><strong>Regional Architecture</strong></figcaption></figure><p>By distributing servers across the globe using a <strong>Regional Architecture</strong>, Facebook ensures low latency for users, no matter where they are.</p><p>Understanding these ‘behind-the-scenes’ mechanics has changed how I view software — it isn’t just about code it’s about building systems that can scale to the size of the planet.</p><p>I will be diving deeper into these specific technologies in my upcoming articles.</p><p>Since this is the first article of my career, if you found it helpful, please give it a ‘Like’ and stay tuned for more! 👇✨🙏”</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=de2cd40c4a46" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>