# Serverless architecture ### **Whats is Serverless?** 🏗️ Traditional servers (old way) * You rent or buy servers (physical or virtual). * You install operating systems, updates, and patches. * You must decide how much CPU/RAM you need before you know the traffic. * If too many users come at once, your server may crash. ⚡ Serverless (new way) * You don’t manage servers — the cloud provider (AWS, Azure, GCP) runs them for you. * You just write small pieces of code (called functions) or run apps. * The cloud automatically gives you as much power as you need when people use your app. * When no one is using it, you pay nothing. 📖 Simple analogy Think of it like Uber vs. owning a car: * Owning a car (traditional server): You pay for it even if it sits unused. You must maintain it. * Uber (serverless): You only pay when you need a ride. You don’t worry about fuel, repairs, or parking. ### **Components of serverless** 1\. Compute (Brains) * This is where your code runs without servers. * Example: * AWS Lambda * Azure Functions * Google Cloud Functions 👉 Runs only when triggered, then stops. 2\. Event Sources (Triggers) * Something has to wake up your function. * Example triggers: * A user clicks a button on a website * A file is uploaded to storage (like S3 bucket) * A new row is added to a database * A message arrives in a queue 3\. Storage * Since compute is stateless & ephemeral, you need permanent storage. * Example: * Object storage: AWS S3, Azure Blob * Database: DynamoDB, Cosmos DB, Firebase 4\. API Gateway * Lets users/apps talk to your serverless functions over the internet. * Example: AWS API Gateway, Azure API Management, Google API Gateway. 👉 Think of it like a receptionist that takes requests and forwards them to your functions. 5\. Messaging & Queues * Helps functions communicate and handle events in order. * Example: * AWS SQS (queue) * AWS SNS (notifications) * Kafka, Pub/Sub 6\. Orchestration / Workflow * Sometimes functions need to run in a sequence or flow. * Example: * AWS Step Functions (workflow) * Azure Durable Functions 7\. Monitoring & Logging * You still need to see what’s happening. * Example: * AWS CloudWatch * Azure Monitor * Google Cloud Logging Imagine a restaurant kitchen: * Compute (Chef): Prepares the dish (runs code). * Event source (Waiter): Brings the order (trigger). * Storage (Fridge): Keeps ingredients safe (data). * API Gateway (Receptionist): Takes customer orders. * Queue (Order board): Manages multiple requests. * Orchestration (Head Chef): Coordinates the cooking flow. * Monitoring (CCTV/Manager): Ensures quality and reports issues.
### **Stateless** * Meaning: No memory of the past. * Each request is treated as brand new, with no knowledge of what happened before. 👉 Example: * When you go to a fast-food counter, each order is fresh. The cashier doesn’t remember your last order unless you tell them again. * In cloud/serverless, this means: if your app gets a request, it doesn’t “remember” the previous request. ⚡ Why? * Makes scaling easy (any server/function can handle the next request). ### **Ephemeral** * Meaning: Short-lived, temporary, not permanent. * Resources or sessions exist only for a little while, then disappear. 👉 Example: * Snapchat messages = ephemeral (they vanish after you read them). * In cloud, a container, virtual machine, or function may run just for a few seconds/minutes, then shut down. ⚡ Why? * Saves cost and resources. * Perfect for tasks that don’t need to live forever. **Architecture**
### 🔐 **Amazon Cognito** * A user identity and access management service from AWS. * It helps you add login, signup, and authentication to your apps without building it from scratch. 🧩 Main parts of Cognito 1\. User Pools 👥 * This is like a user directory. * Stores usernames, passwords, emails, phone numbers. * Handles: * Signup & Sign-in * Forgot password * Multi-Factor Authentication (MFA) * Social logins (Google, Facebook, Apple) 👉 Example: If you build an app, Cognito can be the “Login with Email or Google” system. 2\. Identity Pools (Federated Identities) 🎭 * Lets users get temporary AWS credentials to access AWS services (like S3 or DynamoDB). * Works with User Pools OR external identity providers. * Example: * Your app’s users log in → Cognito gives them a temporary token → They can upload files to S3 without needing permanent AWS keys. 3\. Cognito Sync (deprecated, replaced by AppSync / Amplify) 📦 * Was used for syncing app data across devices (like preferences). ⚡ Simple Analogy Think of Cognito like a bouncer + guest list at a club: * User Pool = The guest list (who is allowed in, with what password/ID). * Identity Pool = The wristband that lets you access certain areas inside (like AWS services). ### 🎬 **Transcoding** * Definition: Converting a media file (video or audio) from one format, size, or quality into another. * It’s like changing the “clothes” of a video so it can fit different devices and networks. 🧩 Example in real life 1. You record a 4K video on your phone. 2. Your friend has slow internet and an old laptop. 3. You transcode the video into 720p smaller file size, so they can watch it smoothly. 🛠️ In the cloud (like AWS) * AWS Elemental MediaConvert is a serverless service for transcoding. * Example: * Upload video to S3 → MediaConvert transcodes it → outputs multiple versions (4K, 1080p, 720p). * Streaming services (like Netflix, YouTube) do this automatically, so every user gets the best quality for their device. ✅ Difference from related terms * Encode: Convert raw video/audio into a compressed format (e.g., camera video → MP4). * Transcode: Convert from one encoded format to another (e.g., MP4 → WebM, or 4K → 1080p). ⚡ Analogy Transcoding is like translating a book into different languages and versions: * Same story (video content). * Different format/size so everyone can enjoy it on their own “device.”