System design diagrams
How systems like these fit together: what talks to what, and why it was decided that way.
News feed, like TwitterFan-out on write or on read? Timelines in a cache, the celebrity problem, and how a post reaches millions of followers.
News feed, like TwitterFan-out on write or on read? Timelines in a cache, the celebrity problem, and how a post reaches millions of followers.
URL shortener, like BitlyThe first system design interview question: key generation, heavy reads, a cache in front, and click analytics off the request path.
Ride hailing app, like UberMatching riders and drivers: live locations in a geo index, dispatch, surge, trips and payments. A classic system design interview.
RAG pipelineChat with your documents: chunking, embeddings, retrieval, reranking and how to know it works.
Video calls, like ZoomPeer to peer for two, an SFU for a meeting, signaling and what happens on a bad network.
Rate limiterToken bucket or sliding window? Where it runs, how nodes share counts and what happens when the limiter itself is down.
File sync, like Dropbox or Google DriveChunking, deduplication, syncing across devices and resolving conflicts.
Stays marketplace, like AirbnbSearch by place and date, reserve without double-booking, pay out to hosts after the stay.
E-commerce marketplaceSellers, catalog, cart, checkout and payouts, with event-driven inventory and search.
Notification systemPush, email and SMS from one place: templates, preferences, retries and no duplicates.
Web crawlerA frontier of URLs, politeness per domain, deduplication and storage at the scale of the web.
Payment systemIdempotency, a double-entry ledger, webhooks and reconciliation. Where money can go wrong and how not to lose any.
Food delivery, like iFood or DoorDashThree sides at once: customer, restaurant and courier, with one order moving through a state machine.
Photo sharing, like InstagramUpload once, resize many times, serve from a CDN, and build a feed that doesn't melt.
Chat app, like WhatsAppOne-to-one and group messages: long-lived connections, delivery receipts, offline inboxes and end-to-end encryption.
Search autocompleteSuggestions in under 100 ms: a trie built offline, served from memory, with trending terms on top.
Ticket booking, like TicketmasterA million people, ten thousand seats, one minute. Holding seats without double-selling, and a waiting room to survive the rush.
Analytics data pipelineFrom app events to dashboards: streaming in, warehouse, dbt models and BI.
Production on KubernetesIngress, autoscaling, observability and GitOps: a realistic cluster for a team of services.
Video streaming, like NetflixUpload, transcode into many qualities, push to a CDN and adapt to the network. Why almost all the traffic never reaches your servers.
