The latest articles on DEV Community by NOEMA (@noema). https://dev.to/noema https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3874528%2F8f17af8b-458c-4bb8-8310-bd42afb81eb1.jpg https://dev.to/noema en NOEMA Mon, 13 Apr 2026 11:40:18 +0000 https://dev.to/noema/most-people-learn-system-design-the-wrong-way-i04 https://dev.to/noema/most-people-learn-system-design-the-wrong-way-i04 <p>Most People Learn System Design the Wrong Way</p> <p>I did too.</p> <p>I spent weeks reading theory, watching videos, and still couldn’t design even a basic system.</p> <p>Here’s what I was doing wrong<br> <a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fg3tymk7tqttl4svrwlon.jpeg" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fg3tymk7tqttl4svrwlon.jpeg" alt="A split-screen illustration showing the difference between poor and effective system design learning. The left side depicts a messy, chaotic diagram with scattered components and unclear flow, representing confusion. The right side shows a clean, structured system architecture with clearly connected components, representing clarity and proper understanding of system design." width="800" height="800"></a> and what actually works.</p> <h2> 1. Starting with theory first </h2> <p>I began with concepts like CAP theorem and consistency models.</p> <p>It felt productive, but nothing clicked.</p> <p>System design is learned by building systems, not memorizing concepts.</p> <h2> 2. Jumping into massive architectures </h2> <p>I tried understanding systems like Netflix and Uber too early.</p> <p>They’re too complex when you’re just starting out.</p> <p>Start small. Focus on features, not entire platforms.</p> <h2> 3. Memorizing components instead of understanding flow </h2> <p>I knew terms like load balancer, cache, and database.</p> <p>But I didn’t understand how data actually moves through a system.</p> <p>Flow &gt; components.</p> <h2> 4. Ignoring trade-offs </h2> <p>There is no perfect design.</p> <p>Every decision is a trade-off between latency, consistency, and cost.</p> <p>This is what system design is really about.</p> <h2> 5. Not drawing systems </h2> <p>I was just reading and watching.</p> <p>The moment I started drawing architectures, everything became clearer.</p> <p>If you can’t draw it, you don’t understand it.</p> <h2> 6. Treating it like a subject, not a skill </h2> <p>System design isn’t something you “finish”.</p> <p>It’s something you build over time.</p> <h2> What actually works </h2> <p>If you’re learning system design right now:</p> <ul> <li>Start simple</li> <li>Focus on flow</li> <li>Think in trade-offs</li> </ul> <p>Everything else follows.</p> <h2> Final thought </h2> <p>System design isn’t about memorizing concepts or copying big architectures.</p> <p>It’s about understanding how systems behave, making trade-offs, and applying that thinking to real problems.</p> <p>The earlier you start applying these ideas in your projects, the faster you improve.</p> <p>What’s one mistake you made while learning system design?</p> systemdesign programming beginners architecture NOEMA Sun, 12 Apr 2026 07:29:09 +0000 https://dev.to/noema/how-whatsapp-handles-11-messaging-at-scale-system-design-explained-5bcp https://dev.to/noema/how-whatsapp-handles-11-messaging-at-scale-system-design-explained-5bcp <p>At first glance, sending a message on WhatsApp feels instant and simple.</p> <p>But behind that single action lies a highly optimized distributed system designed for low latency, high reliability, and massive scale.</p> <p>In this article, we break down how WhatsApp handles 1:1 messaging step by step.</p> <h2> 1. High Concurrency with Erlang </h2> <p>WhatsApp uses Erlang as its backend language.</p> <p>Erlang is designed for:</p> <ul> <li>Handling millions of concurrent processes</li> <li>Fault tolerance</li> <li>Distributed systems</li> </ul> <p>This makes it ideal for real-time messaging at scale.</p> <h2> 2. Persistent Connections via Gateway Servers </h2> <p>Each user maintains a long-lived connection with WhatsApp servers using TCP.</p> <p>These are called gateway or socket servers.</p> <p>Why this matters:</p> <ul> <li>No need to reconnect for every message</li> <li>Lower latency</li> <li>Instant delivery</li> </ul> <h2> 3. Session Service (User-to-Server Mapping) </h2> <p>A session service tracks which user is connected to which server.</p> <p>So when User A sends a message to User B:</p> <ul> <li>The system routes it directly to the correct server</li> <li>No unnecessary broadcasting</li> </ul> <p>This improves efficiency and scalability.</p> <h2> 4. Message Flow (End-to-End Path) </h2> <p>User A → Gateway → Chat Server → Session Service → Receiver Server → Gateway → User B</p> <p>Each component ensures fast and reliable delivery.</p> <h2> 5. Messaging Protocol </h2> <p>WhatsApp uses a modified version of XMPP with persistent communication.</p> <p>This enables:</p> <ul> <li>Lightweight communication</li> <li>Low overhead</li> <li>Real-time delivery</li> </ul> <h2> 6. Message Queue </h2> <p>Messages are placed into a queue before delivery.</p> <p>Benefits:</p> <ul> <li>Handles traffic spikes</li> <li>Prevents message loss</li> <li>Decouples sender and receiver</li> </ul> <h2> 7. Delivery Guarantees </h2> <p>WhatsApp uses acknowledgements:</p> <ul> <li>Sent → server received</li> <li>Delivered → reached recipient device</li> <li>Read → opened by user</li> </ul> <p>This ensures reliability and transparency.</p> <h2> 8. End-to-End Encryption </h2> <p>Messages are encrypted on the sender’s device and decrypted only on the receiver’s device.</p> <p>This ensures privacy and security.</p> <h2> 9. Offline Handling </h2> <p>If the receiver is offline:</p> <ul> <li>Message is stored temporarily</li> <li>Push notification is triggered</li> <li>Delivered when user reconnects</li> </ul> <h2> 10. Local Storage </h2> <p>Messages are stored locally using SQLite.</p> <p>Advantages:</p> <ul> <li>Fast access</li> <li>Offline availability</li> <li>Smooth user experience</li> </ul> <h2> 11. Scalability </h2> <p>Most backend services are stateless.</p> <p>This allows:</p> <ul> <li>Horizontal scaling</li> <li>Load distribution</li> <li>Handling billions of messages</li> </ul> <h2> Conclusion </h2> <p>What seems simple is actually a system involving gateways, queues, session mapping, and distributed services.</p> <p>Understanding this is key to mastering system design.</p> <p>If you found this helpful, follow for more real-world system design breakdowns.</p> architecture backend distributedsystems systemdesign