<?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 Gaurvi Arora on Medium]]></title>
        <description><![CDATA[Stories by Gaurvi Arora on Medium]]></description>
        <link>https://medium.com/@gaurvia359?source=rss-12af959d5ad6------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*9WGlRmHQdy6BWe46mECXzw.jpeg</url>
            <title>Stories by Gaurvi Arora on Medium</title>
            <link>https://medium.com/@gaurvia359?source=rss-12af959d5ad6------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 21:03:26 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gaurvia359/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[Day 90 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-90-of-90daysofdevops-73341885c69b?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/73341885c69b</guid>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Mon, 08 Dec 2025 10:46:54 GMT</pubDate>
            <atom:updated>2025-12-08T10:46:54.374Z</atom:updated>
            <content:encoded><![CDATA[<p>90 days ago, I restarted this challenge with zero expectations, just the intention to show up, learn consistently, and share whatever I understood along the way.<br>Today, I’m closing this chapter with more clarity, more confidence, and a far deeper understanding of what DevOps actually means in the real world.</p><p>This journey wasn’t just about tools.<br>It was about discipline, curiosity, consistency, and the ability to learn publicly.</p><p>Here’s what these 90 days gave me:</p><p>- A solid understanding of Linux, shell scripting, processes &amp; permissions<br>- Hands-on comfort with Git, CI/CD, pipelines, automation, and deployment workflows<br>- Deep dives into Docker, containerization, orchestration, and Kubernetes<br>- Clarity on Helm, GitOps, Argo CD, Flux, and declarative deployments<br>- A full IaC foundation with Terraform, modules, state, workspaces, CI automation<br>- Strong cloud fundamentals across AWS networking, VPCs, load balancers, routing, and security<br>- Real SRE mindset: SLAs, SLOs, SLIs, error budgets, incident response, postmortems</p><p>Most importantly:<br>I now understand why DevOps exists: to build systems that scale, recover, adapt, and improve continuously.</p><p>What I Learnt About Myself</p><p>Showing up every single day taught me something far beyond tech:</p><p>Consistency &gt; motivation<br>Progress compounds<br>Learning publicly accelerates growth<br>Even big topics make sense if you break them down daily<br>Confidence comes from action, not intention</p><p>And maybe the biggest one:<br>You don’t need to know everything; you just need to keep learning.</p><p>“Closing Thoughts”</p><p>This challenge might be ending today,<br>But the journey I started is definitely not.</p><p>DevOps isn’t a skillset, it’s a mindset.<br>It’s about building things that work, break, heal, scale, and evolve.<br>And I’m just getting started.</p><p>Here’s to bigger systems, cleaner pipelines, stronger architectures, and the next phase of learning.<br>Thank you to everyone who supported, commented, or even silently read these posts; it meant more than you think. 💛</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=73341885c69b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 89 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-89-of-90daysofdevops-bc733115da28?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/bc733115da28</guid>
            <category><![CDATA[target-group]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[networking]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Sat, 06 Dec 2025 06:58:07 GMT</pubDate>
            <atom:updated>2025-12-06T06:58:07.235Z</atom:updated>
            <content:encoded><![CDATA[<p>Yesterday I learned the difference between ALB and NLB.<br>Today, I went deeper into how load balancers actually decide which server/container receives traffic, through Target Groups &amp; Health Checks.</p><p>This is the part nobody talks about, but it’s exactly what keeps apps alive during high traffic, deployments, and failures.</p><p>What is a Target Group?<br>A Target Group is simply a collection of backend resources that a load balancer sends traffic to.<br>These targets can be:</p><p>EC2 instances<br>Containers (ECS/EKS)<br>IP addresses<br>Lambda functions (ALB only)</p><p>Each target group has its own:</p><p>- health checks<br>- routing rules<br>- port numbers<br>- scaling behavior</p><p>A load balancer is useless without a target group; it needs somewhere to send traffic.</p><p>Health Checks: The Heartbeat of Your Architecture</p><p>Health checks determine whether a target is healthy and allowed to receive traffic.</p><p>For ALB, you typically configure:</p><p>Health check path (e.g., /health, /status)<br>Success codes (200, 302, etc.)<br>Interval<br>Timeout<br>Healthy/unhealthy thresholds</p><p>If a target fails health checks:<br>-&gt; It is automatically removed from rotation<br>If it becomes healthy again:<br>-&gt; It’s added back without downtime</p><p>This is what allows zero-downtime deployments, rolling updates, and graceful failures.</p><p>Real-World Traffic Flow<br>Client -&gt; Load Balancer -&gt; Healthy Target Only</p><p>If one server crashes?<br>LB routes traffic to other healthy targets instantly.</p><p>If new containers spin up (ECS/EKS)?<br>LB discovers and adds them automatically.</p><p>If you deploy a bad build?<br>Health checks catch it before users do.</p><p>Common Issues You See in DevOps<br>- Wrong health check path -&gt; app looks “down.”<br>- Targets always failing -&gt; ALB/NLB shows 0 healthy hosts<br>- Auto Scaling adds new instances, but LB doesn’t include them<br>- Containers scale but fail health checks due to startup delays</p><p>Health checks are often the difference between “my app is down” and “my app healed itself.”</p><p>Understanding target groups + health checks enables:<br>- Blue/green deployments<br>- Rolling updates<br>- Zero-downtime releases<br>- Auto-scaling<br>- Resilience under failures<br>- Stable microservice communication</p><p>This is the foundation of highly available cloud applications.</p><p>A load balancer doesn’t guarantee reliability; health checks do.<br>Target groups make the architecture scalable.<br>Health checks make it survivable.</p><p>Together, they’re the silent guardians behind every “99.99% uptime” claim.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bc733115da28" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 88 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-88-of-90daysofdevops-e41fbedfa05d?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/e41fbedfa05d</guid>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[network-load-balancer]]></category>
            <category><![CDATA[application-load-balancer]]></category>
            <category><![CDATA[https]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Fri, 05 Dec 2025 09:15:40 GMT</pubDate>
            <atom:updated>2025-12-05T09:15:40.199Z</atom:updated>
            <content:encoded><![CDATA[<p>Today, I explored one of the most important components of AWS networking, Load Balancers, specifically the difference between ALB and NLB.<br>They both distribute traffic, but they solve very different problems.</p><p>Application Load Balancer (ALB)<br>The ALB works at Layer 7 (HTTP/HTTPS).<br>It understands web traffic, URLs, headers, cookies, and paths, making it perfect for:<br>Web apps<br>Microservices<br>API routing<br>Path-based routing (/login, /api/v1)<br>Host-based routing (<a href="http://service.example.com/">service.example.com</a>)<br>Container workloads (ECS/EKS)</p><p>It also handles TLS termination, sticky sessions, and advanced routing rules.</p><p>In short: ALB is the smart, application-aware load balancer.</p><p>Network Load Balancer (NLB)<br>NLB works at Layer 4 (TCP/UDP).<br>It doesn’t inspect requests, it just moves packets insanely fast.<br>Ideal for:</p><p>High-performance workloads<br>Low-latency systems<br>Real-time apps<br>Gaming servers<br>Streaming<br>When you need static IPs or millions of requests per second</p><p>In short: NLB is the ultra-fast, low-latency traffic router.</p><p>Why This Matters<br>Pick the wrong load balancer and your app may:<br>- Slow down<br>- Fail health checks<br>- Break routing<br>- Waste money</p><p>Pick the right one and you get:<br>-&gt; Speed<br>-&gt; Stability<br>-&gt; Clean routing logic<br>-&gt; Production-grade architecture</p><p>ALB = intelligent routing for web applications.<br>NLB = raw speed for network-heavy workloads.</p><p>Choosing the right LB is a core DevOps skill; it’s literally the front door to your entire architecture.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e41fbedfa05d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 87 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-87-of-90daysofdevops-3d2dd19222e6?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/3d2dd19222e6</guid>
            <category><![CDATA[vpc]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[ec2-security-groups]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Thu, 04 Dec 2025 06:10:52 GMT</pubDate>
            <atom:updated>2025-12-04T06:10:52.703Z</atom:updated>
            <content:encoded><![CDATA[<p>Today I explored two core AWS security layers that people often mix up,<br>Security Groups (SGs) and Network ACLs (NACLs).</p><p>Both control traffic in a VPC.<br>Both act as firewalls.<br>But their behavior is completely different.</p><p>Here’s the simplest &amp; most accurate breakdown.</p><p>1. Security Groups, Instance-Level Firewall<br>Security Groups operate at the ENI (network interface) level.<br>They control what traffic reaches a specific instance or service.<br>Key traits:<br>-&gt; Stateful, if inbound is allowed, outbound is automatically allowed back<br>-&gt; Applied to EC2, ALB, RDS, EKS, ECS tasks, etc.<br>-&gt; Default: deny all inbound, allow all outbound<br>-&gt; Easier to manage than NACLs</p><p>Example:<br>If you allow inbound port 443 -&gt; response traffic is automatically allowed.</p><p>Think “Security Groups = Who can talk to this instance?”</p><p>2. NACLs, Subnet-Level Firewall<br>Network ACLs apply to an entire subnet, affecting every resource inside it.<br>Key traits:<br>- Stateless, inbound and outbound must be defined separately<br>- Processed in number order (rule #100 -&gt; #101 -&gt; #102…)<br>- Good for broad network-level restrictions<br>- Rarely changed in modern cloud setups</p><p>Example:<br>If inbound 443 is allowed, you must also explicitly allow outbound 443.</p><p>Think “NACLs = What traffic can enter or leave this subnet?”</p><p>3. Real-World Usage (What DevOps Teams Actually Do)<br>Most production setups use:</p><p>NACLs -&gt; mostly default (as a coarse safety net)<br>Security Groups -&gt; primary, precise control</p><p>Because SGs are stateful, simpler, and easier to reason about, 95% of day-to-day AWS networking relies on SGs.</p><p>4. Common Mistakes Engineers Make<br>- Adding complicated rules in NACLs, accidentally blocking half the app<br>- Forgetting outbound rules in NACLs -&gt; nothing works<br>- Using SGs and NACLs heavily = double troubleshooting<br>- Allowing <a href="http://0.0.0.0/0">0.0.0.0/0</a> everywhere out of confusion</p><p>Most production outages caused by “network issues” come from misconfigured SGs or NACLs.</p><p>Why This Matters in DevOps<br>Understanding SGs vs NACLs means you can:<br>-&gt; Design secure architectures<br>-&gt; Debug access issues faster<br>-&gt; Protect databases, ALBs, and private services<br>-&gt; Build cloud environments that scale safely</p><p>Networking isn’t just routing, it’s controlled communication.</p><p>Security Groups protect instances.<br>NACLs protect subnets.</p><p>SGs decide who can talk to you.<br>NACLs decide what traffic the neighborhood even allows.</p><p>Mastering both makes AWS networking far more predictable.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3d2dd19222e6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 86 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-86-of-90daysofdevops-448a5bd78028?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/448a5bd78028</guid>
            <category><![CDATA[routing]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[route-table]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Wed, 03 Dec 2025 05:42:26 GMT</pubDate>
            <atom:updated>2025-12-03T05:42:26.420Z</atom:updated>
            <content:encoded><![CDATA[<p>After understanding NAT Gateways and Internet Gateways yesterday, today I explored the glue that holds AWS networking together,<br>Route Tables.</p><p>Every subnet in a VPC uses a route table to decide where traffic should go.<br>If route tables are wrong, nothing works, ALBs fail health checks, EC2 can’t reach the internet, EKS nodes can’t join the cluster.</p><p>What is a Route Table?<br>A route table contains rules that tell traffic:<br>What destination CIDRs exist<br>and which gateway or target to use</p><p>Every subnet is explicitly or implicitly associated with a route table.<br>This is what gives that subnet its identity: public or private.</p><p>How Route Tables Actually Work</p><p>Public Subnet Route Table<br>Contains a route to the Internet Gateway (IGW):<br><a href="http://0.0.0.0/0">0.0.0.0/0</a> -&gt; igw-xxxx<br>This single line is what makes a subnet public.<br>No IGW route = not public.</p><p>Private Subnet Route Table<br>Contains a route to the NAT Gateway:<br><a href="http://0.0.0.0/0">0.0.0.0/0</a> -&gt; nat-xxxx<br>This allows outbound internet access without exposing the instances.</p><p>Local Routing<br>Every route table includes an automatic route:<br>VPC CIDR -&gt; local<br>This allows private communication inside the VPC.</p><p>Real Production Flow<br>User -&gt; ALB (public subnet) -&gt; App Servers (private subnet) -&gt; Database (private subnet)</p><p>Behind the scenes:<br>-&gt; ALB uses IGW route<br>-&gt; App servers use NAT route<br>-&gt; DB uses only local routes (NO internet)</p><p>Everything depends on correct routing.</p><p>Common Route Table Mistakes<br>- Subnet has no route table association -&gt; it silently uses the wrong one<br>- Private subnet pointing directly to IGW -&gt; security hole<br>- NAT Gateway in a private subnet -&gt; dead architecture<br>- Wrong routes to VPC peering -&gt; cross-VPC traffic fails<br>- Missing routes for EKS/ECS -&gt; nodes fail to register</p><p>Most “networking bugs” are actually route table mistakes.</p><p>Why This Matters in DevOps<br>Route tables decide:<br>-&gt; Who can talk to whom<br>-&gt; Which services are public or private<br>-&gt; Whether instances reach the internet<br>-&gt; Whether APIs are reachable<br>-&gt; Whether load balancers can talk to targets</p><p>They’re the traffic control system of your VPC.</p><p>Subnets define where resources live.<br>Route tables define how they communicate.<br>Get routing wrong -&gt; the whole architecture collapses.</p><p>Networking works only when routing works.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=448a5bd78028" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 85 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-85-of-90daysofdevops-9d4746625f48?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/9d4746625f48</guid>
            <category><![CDATA[vpc]]></category>
            <category><![CDATA[internet-gateway]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[networking]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Tue, 02 Dec 2025 02:02:33 GMT</pubDate>
            <atom:updated>2025-12-02T02:02:33.914Z</atom:updated>
            <content:encoded><![CDATA[<p>After exploring public and private subnets yesterday, today I dug into one of the most misunderstood parts of AWS networking,<br>NAT Gateway vs Internet Gateway.</p><p>They sound similar, both connect to the internet…<br>But they serve completely different purposes.</p><p>Here’s the clean breakdown</p><p>Internet Gateway (IGW)<br>- Attached directly to a VPC<br>- Enables inbound + outbound internet traffic<br>- Required for public subnets</p><p>Used for:<br>- ALBs<br>- Bastion Hosts<br>- Public EC2<br>- Any resource that must be reachable from the Internet</p><p>Think of IGW as:<br>“The front door of your VPC for public access.”</p><p>NAT Gateway (NGW)<br>- Lives in a public subnet<br>- Allows private subnets to reach the internet<br>- Only outbound traffic (no inbound from the internet)</p><p>Used for:<br>App servers that need package updates<br>EKS/ECS worker nodes<br>Private EC2 instances<br>Backend services<br>Anything that should not be directly exposed</p><p>Think of NAT as:<br>“A secure outbound bridge for private subnets.”</p><p>How They Work Together<br>Private subnet -&gt; NAT Gateway -&gt; Internet<br>Public subnet -&gt; IGW -&gt; Internet</p><p>Services exposed publicly use IGW.<br>Services internal to your architecture use NAT.</p><p>Common Misconfigurations Engineers Make</p><p>Putting NAT in a private subnet<br>-&gt; Then nothing works.</p><p>Sending inbound traffic through NAT<br>-&gt; Impossible by design.</p><p>Not having route tables point private subnets to NAT<br>-&gt; Instances lose outbound internet access.</p><p>Putting databases in public subnets<br>-&gt; Extremely unsafe and unnecessary.</p><p>Why This Matters in DevOps<br>A correct NAT/IGW setup determines:<br>-&gt; Whether your app can pull packages<br>-&gt; Whether your nodes can join EKS<br>-&gt; Whether your ALB can reach backend targets<br>-&gt; Whether your database stays secure<br>-&gt; Whether your system is production-grade</p><p>Networking failures often come from misunderstanding this exact difference.</p><p>IGW brings traffic in.<br>NAT lets traffic go out.<br>Both are essential, but for completely different reasons.</p><p>Mastering this distinction is the first step toward building secure, scalable AWS architectures.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9d4746625f48" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 84 of 90DaysOfDevOp]]></title>
            <link>https://medium.com/@gaurvia359/day-84-of-90daysofdevop-caf084bc6a58?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/caf084bc6a58</guid>
            <category><![CDATA[subnetting]]></category>
            <category><![CDATA[ip-address]]></category>
            <category><![CDATA[private]]></category>
            <category><![CDATA[public-subnet]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Mon, 01 Dec 2025 06:23:56 GMT</pubDate>
            <atom:updated>2025-12-01T06:23:56.902Z</atom:updated>
            <content:encoded><![CDATA[<h3>Day 84 of 90DaysOfDevOps</h3><p>Yesterday I learned the foundations of AWS VPCs.<br>Today I went deeper into one of the most important architectural decisions:<br>Public vs. Private Subnets, and How Modern Cloud Apps Actually Use Them.</p><p>Most outages, access issues, and security gaps come from wrong subnet design.</p><p>1. Public Subnets, Internet-Facing<br>A public subnet has a route to the Internet Gateway (IGW).</p><p>Use it for resources that must be accessible from the internet:</p><p>Load Balancers (ALB/NLB)<br>Bastion Hosts<br>NAT Gateways<br>Public-facing EC2 instances (rare these days)</p><p>Rule: If it needs inbound traffic from outside -&gt; public subnet.</p><p>2. Private Subnets, Internal Only<br>Private subnets have no direct internet route.<br>Instead, they reach the internet via NAT Gateway (outbound only).</p><p>Use them for:<br>Application servers<br>Containers (ECS/EKS nodes)<br>Databases (RDS)<br>Caches (Redis/ElastiCache)<br>Internal services</p><p>Rule: If it should never be publicly reachable, use a private subnet.</p><p>3. How Traffic Flows (Real Prod Setup)<br>User -&gt; Internet -&gt; ALB (Public Subnet) -&gt; App Servers (Private Subnet) -&gt; DB (Private Subnet)</p><p>Outbound traffic (like package updates):<br>App Servers -&gt; NAT Gateway -&gt; Internet</p><p>This pattern is used by:<br>Amazon, Netflix, Airbnb, Spotify, and basically every secure cloud architecture.</p><p>4. Common Mistakes Engineers Make<br>-&gt; Putting databases in public subnets<br>-&gt; Allowing <a href="http://0.0.0.0/0">0.0.0.0/0</a> ingress everywhere<br>-&gt; Forgetting NAT Gateway -&gt; Private instances lose internet<br>-&gt; Mixing ALB + app servers in the same subnet</p><p>These mistakes create outages, security risks, and unpredictable behavior.</p><p>Why This Matters in DevOps<br>Infrastructure is only as good as its network design.<br>Public/private subnet segregation ensures:<br>-&gt; Security<br>-&gt; Scalability<br>-&gt; Controlled traffic flow<br>-&gt; Predictable architecture<br>-&gt; Cloud best practices</p><p>This is the backbone of every modern cloud deployment.</p><p>Public subnets face the world.<br>Private subnets run the world.<br>Good architects know exactly what belongs where and why.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=caf084bc6a58" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 83 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-83-of-90daysofdevops-edc849381575?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/edc849381575</guid>
            <category><![CDATA[vpc]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[terraform]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Sat, 29 Nov 2025 21:44:10 GMT</pubDate>
            <atom:updated>2025-11-29T21:44:10.935Z</atom:updated>
            <content:encoded><![CDATA[<p>After spending the last few days deep in Terraform concepts, today I moved into one of the most essential building blocks of cloud infrastructure, AWS VPC Networking.</p><p>Everything in AWS runs inside a VPC, so understanding how networking works is the foundation of real DevOps.</p><p>What is a VPC?<br>A Virtual Private Cloud is your isolated network inside AWS.<br>It’s like your own private data center, fully customizable, secure, and scalable.</p><p>You define:<br>- IP ranges<br>- Subnets<br>- Routing<br>- Internet access<br>- Firewalls (security groups)<br>Every application eventually runs inside a VPC.</p><p>Key VPC Components</p><p>1. CIDR Block<br>Your network’s IP range, e.g.:<br><a href="http://10.0.0.0/16">10.0.0.0/16</a><br>This decides how many IPs you have.</p><p>2. Subnets<br>You divide the VPC into smaller networks:<br>Public Subnets -&gt; internet-facing resources (ALBs, NAT, Bastion)<br>Private Subnets -&gt; internal services (EC2, RDS, EKS nodes)</p><p>This separation = security + reliability.</p><p>3. Route Tables<br>Define how traffic flows inside the VPC.<br>Example:<br>Public subnet route table -&gt; <a href="http://0.0.0.0/0">0.0.0.0/0</a> -&gt; Internet Gateway<br>Private subnet route table -&gt; <a href="http://0.0.0.0/0">0.0.0.0/0</a> -&gt; NAT Gateway</p><p>4. Internet Gateway (IGW)<br>Allows public subnets to access the internet.</p><p>5. NAT Gateway<br>Allows private subnets to access the internet without being exposed.<br>Perfect for:<br>app servers<br>EKS nodes<br>database migrations<br>patching</p><p>6. Security Groups &amp; NACLs<br>Security Groups -&gt; instance-level firewall (stateful)<br>NACLs -&gt; subnet-level firewall (stateless)</p><p>Security Groups are used 95% of the time in real setups.</p><p>Why This Matters in DevOps<br>You can’t build:<br>-&gt; Kubernetes clusters<br>-&gt; Load balancers<br>-&gt; CI/CD runners<br>-&gt; EC2 fleets<br>-&gt; Private databases<br>without a solid VPC design.</p><p>Bad networking = outages, failures, unreachable services, and security risks.</p><p>Good networking = scalable, secure, production-ready infrastructure.</p><p>VPCs aren’t “networking concepts”, <br>they’re the foundation of everything we build in the cloud.<br>If you understand VPCs well, the rest of AWS becomes 10x easier.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=edc849381575" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 82 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-82-of-90daysofdevops-adba4bb35e2c?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/adba4bb35e2c</guid>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[provisioning]]></category>
            <category><![CDATA[terraform]]></category>
            <category><![CDATA[iac]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Fri, 28 Nov 2025 02:53:18 GMT</pubDate>
            <atom:updated>2025-11-28T02:53:18.743Z</atom:updated>
            <content:encoded><![CDATA[<p>Today I explored a Terraform topic that every DevOps engineer hears about,<br>Provisioners.<br>Useful? Yes.<br>Recommended? Rarely.</p><p>Provisioners let you run scripts inside or right after infrastructure creation… but they come with trade-offs that professionals need to understand.</p><p>What Are Terraform Provisioners?<br>Provisioners allow Terraform to run actions like:<br>executing shell commands<br>copying files<br>bootstrapping servers<br>installing packages</p><p>Common examples:<br>remote-exec<br>local-exec<br>file</p><p>Example: remote-exec Provisioner<br>resource “aws_instance” “server” {<br> ami = “ami-123456”<br> instance_type = “t2.micro”</p><p>provisioner “remote-exec” {<br> inline = [<br> “sudo apt update -y”,<br> “sudo apt install nginx -y”<br> ]<br> }<br>}<br>Terraform creates the EC2 -&gt; SSHs into it -&gt; runs commands.</p><p>Why Provisioners Are Not Recommended<br>Terraform docs themselves say it:<br>Avoid provisioners unless absolutely necessary.</p><p>Here’s why:<br>They break idempotency<br>They depend on SSH/WinRM (fragile)<br>Failures leave infra in partial state<br>Harder to debug &amp; reproduce<br>Don’t fit Terraform’s declarative model</p><p>In simple words:<br>Provisioners mix “configuration management” with “infrastructure provisioning”, which is a bad combo.</p><p>The Better Alternatives</p><p>Instead of provisioners, real DevOps teams use:<br>Cloud-init / user-data (EC2 bootstrap)<br>Ansible (config management)<br>Packer (pre-baked AMI images)<br>Kubernetes manifests<br>Startup scripts (GCP/Azure)</p><p>Terraform should focus on provisioning resources,<br>not configuring them.</p><p>Why This Matters in DevOps<br>Understanding provisioners is important because you will see them in older codebases.<br>But knowing when to avoid them is the real skill.</p><p>Good IaC is clean, declarative, and predictable,<br>Provisioners often break that.</p><p>Provisioners feel powerful, but they pull Terraform away from what it does best, declarative, predictable infrastructure.<br>Use them as a last resort… and use proper configuration tools for everything else.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=adba4bb35e2c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 81 of 90DaysOfDevOps]]></title>
            <link>https://medium.com/@gaurvia359/day-81-of-90daysofdevops-2821f004b7eb?source=rss-12af959d5ad6------2</link>
            <guid isPermaLink="false">https://medium.com/p/2821f004b7eb</guid>
            <category><![CDATA[iac]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[terraform-automation]]></category>
            <dc:creator><![CDATA[Gaurvi Arora]]></dc:creator>
            <pubDate>Thu, 27 Nov 2025 02:05:02 GMT</pubDate>
            <atom:updated>2025-11-27T02:05:02.617Z</atom:updated>
            <content:encoded><![CDATA[<p>Today I explored how Terraform integrates with CI/CD pipelines, making infrastructure fully automated, just like application deployments.<br>Because real DevOps isn’t “running terraform apply locally”, it’s automating it safely in pipelines.</p><p>Why Automate Terraform?<br>Manual IaC is risky.<br>Automation ensures:<br>- Consistency<br>- Reviewability<br>- Predictability<br>- Zero human error<br>- Proper approvals before changes</p><p>Terraform + CI/CD = Infrastructure that updates itself through pull requests.</p><p>Typical Terraform Pipeline Structure</p><p>1. terraform fmt (format check)<br>Ensures clean, consistent formatting.</p><p>2. terraform init<br>Sets up providers and modules.</p><p>3. terraform validate<br>Checks if the configuration is syntactically correct.</p><p>4. terraform plan (Pull Request Stage)<br>Generates a plan showing additions, changes, and deletions.<br>This is reviewed by the team before merging.</p><p>5. terraform apply (Post-Merge Stage)<br>Runs only after approval and merge.<br>This ensures production safety.</p><p>GitHub Actions Example<br>- name: Terraform Init<br> run: terraform init</p><p>- name: Terraform Plan<br> run: terraform plan</p><p>- name: Terraform Apply<br> if: github.ref == ‘refs/heads/main’<br> run: terraform apply -auto-approve<br>A merged PR -&gt; updates real cloud infrastructure.<br>This is how modern cloud teams work.</p><p>Secure Terraform Pipelines Should Use<br>-&gt; Remote State (S3/DynamoDB, GCS, Azure Blob)<br>-&gt; OIDC or IAM role assumption<br>-&gt; No hardcoded secrets<br>-&gt; Approval gates for production<br>-&gt; Drift detection</p><p>Why This Matters in DevOps<br>CI/CD automation turns IaC from “run it when needed” to<br>“Infrastructure updates happen through code reviews.”</p><p>It brings:<br>- reliability,<br>- traceability,<br>- collaboration, and safety.</p><p>This is what separates junior IaC from production-grade DevOps.</p><p>Terraform becomes truly powerful when paired with CI/CD.<br>When infrastructure changes only through pull requests,<br>you gain the holy trinity of DevOps: consistency, safety, and speed.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2821f004b7eb" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>