<?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 Harikishan on Medium]]></title>
        <description><![CDATA[Stories by Harikishan on Medium]]></description>
        <link>https://medium.com/@harikishan468?source=rss-44707dfd8e0b------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*_MFJXRt-FBeNbnmz</url>
            <title>Stories by Harikishan on Medium</title>
            <link>https://medium.com/@harikishan468?source=rss-44707dfd8e0b------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 21:44:50 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@harikishan468/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[Why Enterprise VPCs Aren’t as Big as They Look]]></title>
            <link>https://medium.com/@harikishan468/why-enterprise-vpcs-arent-as-big-as-they-look-b20f734c1544?source=rss-44707dfd8e0b------2</link>
            <guid isPermaLink="false">https://medium.com/p/b20f734c1544</guid>
            <category><![CDATA[cloud-architecture]]></category>
            <category><![CDATA[platform-engineering]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Harikishan]]></dc:creator>
            <pubDate>Mon, 27 Jul 2026 05:09:42 GMT</pubDate>
            <atom:updated>2026-07-27T05:10:16.882Z</atom:updated>
            <content:encoded><![CDATA[<h3>The enterprise networking constraint hidden behind every enterprise VPC</h3><p><em>Enterprise Engineering Decisions #2–8 min read<br></em><strong>Part of the <em>Enterprise Engineering Decisions</em> series</strong>, where I share the engineering decisions, trade-offs, and architectural thinking behind enterprise cloud platforms — not implementation tutorials.</p><p>Every AWS course teaches networking the same way.</p><p>You create a VPC. Assign a /16 CIDR. Split it into public and private subnets. Deploy applications.</p><p>Simple.</p><p>That mental model works perfectly for learning AWS.</p><p>Then you join a large enterprise.</p><p>Someone tells you: <em>“You have a /16.”</em></p><p>A week later, you discover something unexpected: <strong>not every IP address inside that VPC can actually participate in the enterprise network.</strong></p><p>The limitation isn’t AWS. It’s enterprise architecture. Enterprise networking is designed to optimize long-term scalability, governance, and hybrid connectivity — not maximum flexibility for individual applications.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ySWl3BvXxHNnh02OnwtMnA.png" /></figure><h3>The AWS Mental Model</h3><p>Most engineers naturally think:</p><blockquote><em>“My VPC is 10.20.0.0/16. <br>That means I have more than 65,000 IP addresses available. <br>I’ll decide how to use them.”</em></blockquote><p>In many enterprise environments, that assumption doesn’t hold — because cloud networking isn’t designed for a single application. <strong>It’s designed for thousands.</strong></p><h3>Enterprise Networking Optimizes Different Problems</h3><p>Enterprise networking teams aren’t thinking about your application. They’re designing for an ecosystem of applications, platforms, hybrid connectivity, and future growth — solving questions like:</p><ul><li>Can this network be routed across hundreds of AWS accounts?</li><li>Will it integrate with existing data centers?</li><li>Can future acquisitions connect without address conflicts?</li><li>Does it comply with enterprise segmentation policies?</li><li>Can traffic be inspected consistently?</li><li>Will this still work five years from now?</li></ul><p>Those priorities fundamentally change how IP addresses are allocated. Rather than giving every workload unrestricted routable address space, many organizations deliberately separate routable and non-routable networks.</p><p>The result is a VPC that looks enormous on paper but behaves very differently in practice.</p><h3>There Are No Public Subnets</h3><p>One of the biggest surprises for engineers moving from AWS training into enterprise platforms is discovering that public subnets simply don’t exist.</p><p>Not because AWS doesn’t support them. <strong>Because the enterprise doesn’t want them.</strong></p><p>AWS reference architectures often look like this:</p><pre>Internet Gateway<br>       │<br>Public Subnet<br>       │<br>Private Subnet</pre><p>In many enterprise environments, application VPCs never connect directly to the internet. They typically follow patterns like:</p><ul><li>No internet-facing workloads</li><li>No public EC2 instances</li><li>No public load balancers</li><li>No direct outbound internet access</li></ul><p>Traffic leaves through centrally managed networking services, where security inspection, logging, routing policies, and governance are applied consistently. The application team never interacts directly with the internet.</p><p><strong>And that’s entirely by design.</strong></p><h3>Kubernetes Makes the Constraint More Visible</h3><p>Platform teams now have to think beyond virtual machines. Worker nodes, pod networking, ENIs, internal load balancers, and service networking all influence how enterprise address space must be planned and managed.</p><p>The question is no longer:</p><blockquote><em>“How large is my VPC?”</em></blockquote><p>It’s:</p><blockquote><em>“How much routable address space is actually available for my platform?”</em></blockquote><p><strong>That’s a different problem entirely.</strong></p><h3>The Engineering Decision</h3><p><strong>Early cloud thinking:</strong> Give every team a VPC.</p><p><strong>Enterprise thinking:</strong> Give every team a governed networking boundary.</p><p>Those are very different philosophies. The goal isn’t maximizing flexibility. <strong>The goal is ensuring that thousands of applications can coexist</strong> without creating routing conflicts, security gaps, or operational complexity.</p><h3>📋 Enterprise Engineering Decision #2</h3><p><strong>Constraint:</strong> Large enterprise platforms have finite routable address space that must scale across applications, environments, regions, hybrid networks, and future growth.</p><p><strong>Decision:</strong> Treat routable IP allocation as a centrally governed platform capability, while allowing applications to use larger isolated address spaces where appropriate.</p><p><strong>Trade-off:</strong> Application teams have less networking autonomy.</p><p><strong>Outcome:</strong> A platform that continues scaling predictably, without constant network redesign or address conflicts.</p><h3>Closing Thoughts</h3><p>One of the biggest lessons enterprise cloud teaches is this: <strong>a VPC CIDR is not simply an allocation of IP addresses. It’s an architectural boundary.</strong></p><p>The size of a VPC tells you almost nothing about the networking freedom you actually have.</p><p>In enterprise cloud platforms, the question that matters isn’t <em>“How large is my CIDR?”</em> — it’s <em>“How much of that CIDR can actually participate in the enterprise network?”</em></p><p>That’s the difference between learning AWS, and designing enterprise cloud platforms.</p><h3>Continue the series</h3><p><strong>Enterprise Engineering Decisions</strong></p><p>✅ #1 — <em>Why Enterprise Cloud Networking Looks Nothing Like AWS Reference Architectures</em></p><p>✅ #2 — <em>Why Enterprise VPCs Aren’t as Big as They Look</em></p><p>Coming next:</p><p><strong>#3 — Why Platform Teams Standardize Everything</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b20f734c1544" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Enterprise Cloud Networking Looks Nothing Like AWS Reference Architectures]]></title>
            <link>https://medium.com/@harikishan468/why-enterprise-cloud-networking-looks-nothing-like-aws-reference-architectures-9a78317deb5d?source=rss-44707dfd8e0b------2</link>
            <guid isPermaLink="false">https://medium.com/p/9a78317deb5d</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[cloud-architecture]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <dc:creator><![CDATA[Harikishan]]></dc:creator>
            <pubDate>Sun, 19 Jul 2026 13:21:08 GMT</pubDate>
            <atom:updated>2026-07-19T13:21:08.138Z</atom:updated>
            <content:encoded><![CDATA[<h3>Lessons learned building enterprise cloud platforms that no certification or documentation prepared me for</h3><p><em>Enterprise Engineering Decisions #1–10 min read</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OUpvAF331WP8YYDRPVPYug.png" /></figure><p>Every AWS networking course starts with the same architecture: a VPC, public and private subnets, an Internet Gateway, a NAT Gateway, maybe a load balancer.</p><p>It’s clean. It’s elegant. It’s easy to understand.</p><p>If you’re learning AWS, it’s exactly how networking should be taught.</p><p>Then you join a large enterprise — and suddenly none of those diagrams resemble the environment you’re working in.</p><ul><li>Instead of a handful of VPCs, there are <strong>hundreds of AWS accounts</strong>.</li><li>Instead of applications talking directly to the internet, <strong>every outbound connection is inspected</strong>.</li><li>Instead of teams building their own networking components, <strong>most services are shared across the organization</strong>.</li><li>Every connectivity request needs justification. Every route is governed. Every IP allocation is planned.</li></ul><p>At first, I assumed this was unnecessary complexity — that enterprise architects had simply over-engineered the cloud.</p><p>I couldn’t have been more wrong.</p><p>Over the next few years, I learned that almost every additional networking component existed because someone had already hit a problem that a standard AWS reference architecture couldn’t solve.</p><p>This article isn’t about <em>how</em> to build enterprise networking. It’s about <strong>why enterprise networking evolves the way it does.</strong></p><h3>The Cloud Doesn’t Remove Complexity — It Moves It</h3><p>One of the biggest misconceptions about cloud adoption is that infrastructure becomes simpler. Provisioning it certainly does. <em>Managing it at enterprise scale does not.</em></p><p>Launching an EC2 instance takes minutes. Designing a network that securely supports thousands of workloads across multiple business units, regulatory regimes, hybrid connectivity, and dozens of AWS accounts is an entirely different challenge.</p><blockquote><em>Cloud platforms remove operational overhead. They don’t eliminate architectural decisions — they often make those decisions </em>more<em> important, because mistakes now scale.</em></blockquote><h3>Lesson 1: Networking Is No Longer Owned by One Team</h3><p>In smaller environments, application teams manage their own networking end-to-end — VPCs, security groups, load balancers, ports. Everything stays self-contained.</p><p>Enterprise platforms work differently. Networking becomes shared infrastructure, and every decision touches multiple stakeholders:</p><p>Team Responsibility Security Defines governance Platform Defines reusable patterns Cloud Networking Manages connectivity Compliance Sets regulatory requirements Identity Shapes access patterns</p><p>A routing change made for one application can unintentionally impact dozens of others.</p><p>This changes the objective entirely. <strong>The goal is no longer to connect one application — it’s to create a networking model that hundreds of teams can safely use without constantly redesigning it.</strong></p><h3>Standard AWS Architectures Solve Technical Problems. Enterprise Architectures Solve Organizational Problems.</h3><p>This was the biggest mindset shift for me.</p><p>AWS reference architectures answer questions like:</p><ul><li>How do I expose an application to the internet?</li><li>How do I connect two VPCs?</li><li>How do I build highly available workloads?</li></ul><p>Enterprise architecture asks different questions:</p><ul><li>How do we ensure every application follows the same security standard?</li><li>How do we stop teams from building hundreds of different networking patterns?</li><li>How do we reduce operational risk and simplify audits?</li><li>How do we support future growth without redesigning the platform every year?</li></ul><p>These aren’t networking problems. <strong>They’re governance problems</strong> — and networking is simply one of the tools used to solve them.</p><h3>Every Additional Component Exists for a Reason</h3><p>My first look at an enterprise networking diagram was pure confusion: Transit Gateways, shared services, centralized DNS, private connectivity, inspection layers, dedicated networking accounts. It looked nothing like the AWS docs.</p><p>Over time, I realized something important: <strong>nobody wakes up and decides to add complexity for its own sake.</strong></p><ul><li>A centralized networking service exists because hundreds of applications needed the same capability.</li><li>A shared connectivity layer exists because duplicate infrastructure became unsustainable.</li><li>Centralized egress exists because security needed consistent inspection and logging.</li><li>Private connectivity exists because exposing services publicly created unacceptable risk.</li></ul><p><strong>Complexity is usually the accumulated history of solved problems.</strong></p><p>That realization changed how I read architecture diagrams. Instead of asking <em>“Why is this so complicated?”</em> — I started asking:</p><blockquote><strong><em>“Which problem was this component introduced to solve?”</em></strong></blockquote><p>That one question almost always leads to a better understanding of enterprise design.</p><h3>Kubernetes Changes the Conversation</h3><p>Kubernetes adds another dimension entirely. Infrastructure stops being static — pods appear and disappear, deployments roll forward, clusters grow and shrink continuously.</p><p>Design decisions that worked well for virtual machines often don’t hold up for highly elastic platforms. The conversation shifts from <em>provisioning infrastructure</em> to <strong>designing systems that keep working as the infrastructure underneath them constantly changes</strong> — all without sacrificing governance, security, or operational consistency. That balance is far harder than simply standing up a cluster.</p><h3>Enterprise Architecture Is About Standardization</h3><p>One lesson became clear across every platform I worked on:</p><blockquote><strong><em>Good enterprise architecture isn’t measured by how many services it uses — it’s measured by how few decisions application teams need to make.</em></strong></blockquote><p>If every team designs networking differently, operational complexity grows exponentially. If every team follows well-defined patterns, the platform becomes easier to operate, secure, troubleshoot, and evolve.</p><p>Standardization isn’t about restricting engineers — it’s about letting hundreds of teams move faster without each one reinventing the architecture. Ironically, adding <em>more</em> platform capability often <em>reduces</em> complexity for the teams building on top of it.</p><h3>The Biggest Lesson</h3><p>When I started with AWS, I believed cloud architecture was mostly about <strong>choosing the right services</strong>.</p><p>Today, I think it’s about <strong>choosing the right boundaries</strong>:</p><ul><li>What should be shared, and what should stay isolated?</li><li>Who owns which responsibility?</li><li>Which decisions belong to individual teams, and which should the platform make once, for everyone?</li></ul><p>Those questions matter far more than whether a workload runs on EC2, ECS, or Kubernetes. Technology changes fast. <strong>Good architectural boundaries last much longer.</strong></p><h3>Final Thoughts</h3><p>AWS reference architectures are excellent learning tools — they teach the core concepts every cloud engineer should understand. Enterprise platforms build on those foundations, adding layers not because the cloud requires them, but because <strong>organizations operating at scale do</strong>.</p><p>The next time you see an enterprise networking diagram that looks far more complicated than anything in the docs, resist the urge to dismiss it as over-engineering. Ask instead:</p><p><strong>What engineering problem was this designed to solve?</strong></p><p>That’s where the real learning begins.</p><h3>📋 Enterprise Engineering Decision #1</h3><p><strong>Constraint:</strong> Enterprise platforms must support hundreds of applications, multiple engineering teams, regulatory requirements, hybrid connectivity, and continuous growth.</p><p><strong>Decision:</strong> Move networking responsibility from individual application teams into standardized platform capabilities with shared governance and reusable patterns.</p><p><strong>Trade-off:</strong> The architecture becomes more sophisticated, requiring dedicated platform and networking expertise.</p><p><strong>Outcome:</strong> Application teams spend less time solving infrastructure problems, while the organization gains consistency, stronger security, operational scalability, and a platform that evolves without every team reinventing networking on its own.</p><p><em>This is the first post in</em> Enterprise Engineering Decisions*, a series where I share lessons from designing and operating enterprise cloud platforms — the engineering decisions, trade-offs, and architectural thinking behind platforms that operate at scale, rather than service docs or implementation tutorials.*</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9a78317deb5d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[“Unlocking Scalability and Efficiency: Deploying AWS EKS with API Endpoints”]]></title>
            <link>https://medium.com/@harikishan468/unlocking-scalability-and-efficiency-deploying-aws-eks-with-api-endpoints-97489e22a61d?source=rss-44707dfd8e0b------2</link>
            <guid isPermaLink="false">https://medium.com/p/97489e22a61d</guid>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Harikishan]]></dc:creator>
            <pubDate>Sun, 17 Sep 2023 19:47:57 GMT</pubDate>
            <atom:updated>2023-09-17T19:47:57.164Z</atom:updated>
            <content:encoded><![CDATA[<p>Introduction:</p><p>In the rapidly evolving landscape of cloud computing, Amazon Web Services (AWS) has emerged as a frontrunner, providing businesses with a suite of powerful tools and services. One of the most noteworthy offerings is Amazon Elastic Kubernetes Service (EKS), a managed Kubernetes service that simplifies the deployment and management of containerized applications. In this article, we’ll explore how AWS EKS can be harnessed to deploy projects with API endpoints, delivering scalability and efficiency to your applications.</p><p><strong>The Power of AWS EKS</strong></p><p>Amazon EKS is designed to make it easier for developers to manage containerized applications using Kubernetes, an open-source container orchestration platform. Here’s why EKS stands out:</p><ol><li><strong>Managed Kubernetes</strong>: AWS takes care of the underlying infrastructure, ensuring that your Kubernetes clusters are highly available, scalable, and secure.</li></ol><p>2. <strong>Seamless Scaling</strong>: EKS allows you to scale your applications up or down effortlessly, adapting to varying workloads and traffic patterns.</p><p>3. <strong>Compatibility</strong>: EKS is compatible with existing Kubernetes applications, meaning you can migrate your containerized projects without significant modifications.</p><p>4. <strong>Security</strong>: AWS provides robust security features, including encryption, identity and access management (IAM), and Virtual Private Cloud (VPC) integration, to protect your applications.</p><p><strong>Deploying AWS EKS for Your Project</strong></p><p><strong>Step 1: Setting Up EKS Cluster</strong></p><p>To begin, you need to create an EKS cluster. Use the AWS Management Console or the AWS Command Line Interface (CLI) to configure your cluster. Define the cluster’s capacity, instance types, and networking options.</p><p><strong>Step 2: Containerization</strong></p><p>Containerize your application using Docker or another containerization tool. Create a Docker image for your project, including all dependencies.</p><p><strong>Step 3: Kubernetes Configuration</strong></p><p>Write Kubernetes configuration files (YAML) to define your application’s deployment, services, and pods. These files specify how your application should run within the EKS cluster.</p><p><strong>Step 4: Deployment</strong></p><p>Deploy your application by applying the Kubernetes configuration files. EKS will ensure that your application runs across multiple Availability Zones for high availability.</p><p><strong>Step 5: Expose Endpoints</strong></p><p>Now, let’s expose your project to the outside world. This is where API endpoints come into play. You can deploy an Ingress Controller to manage incoming traffic and route it to the appropriate services within your EKS cluster.</p><p><strong>Implementing API Endpoints</strong></p><p>AWS offers several options for implementing API endpoints in conjunction with EKS:</p><ol><li><strong>AWS API Gateway</strong>: This managed service makes it easy to create, publish, and secure APIs for your applications. Integrate it with your EKS cluster to handle API traffic.</li></ol><p>2. <strong>AWS App Runner</strong>: For simpler, serverless-like deployments, AWS App Runner can automatically build, deploy, and scale your containerized applications, including APIs.</p><p>3. <strong>Nginx Ingress Controller</strong>: If you prefer more control over your API endpoints, you can use the Nginx Ingress Controller to manage routing and load balancing for your services within EKS.</p><p><strong>Benefits of EKS and API Endpoints</strong></p><ul><li><strong>Scalability</strong>: EKS can automatically scale your containerized applications based on demand, while API endpoints ensure efficient handling of incoming requests.</li><li><strong>High Availability:</strong> With multi-Availability Zone deployments, EKS ensures that your applications are resilient to failures, and API Gateway or Ingress Controllers can distribute traffic effectively.</li><li><strong>Security</strong>: AWS offers robust security features to protect your APIs and containerized applications, including encryption, IAM, and VPC isolation.</li><li><strong>Developer Productivity</strong>: By simplifying infrastructure management and offering managed services, EKS allows developers to focus on coding, enhancing productivity.</li></ul><p><strong>Conclusion</strong></p><p>Deploying AWS EKS with API endpoints can transform the way you manage and scale your containerized projects. It brings together the power of Kubernetes with the flexibility of API management, delivering a seamless and efficient experience for both developers and users. As businesses continue to embrace cloud-native solutions, AWS EKS proves to be a game-changer in the world of container orchestration and application deployment.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=97489e22a61d" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>