HLDintermediatePremium

WhatsApp HLD: Real-Time Chat for Billions, Online or Off

A WhatsApp / chat-at-scale system design: persistent connections and a routing registry, store-and-forward for offline users, idempotent ordered delivery, and the delivery-tick receipts.

By fiveyearsdevJune 14, 20269 min read

"Design WhatsApp." Sending a message looks like one POST, but the hard parts hide behind "real-time" and "billions." The recipient might be online on a phone connected to a different server across the world, or offline in a tunnel, or have just retried the same send because their signal flickered. The system has to deliver exactly once as the user perceives it, in order, fast when both are online, and eventually when they're not — across a fleet where each user holds a long-lived connection to some one machine. Three problems carry the design: routing (a message must find the server holding th…

What’s inside

Let's start nowhere near a computer

Where this exact shape shows up

Step 1 — Functional requirements (sentences first)

Step 2 — Non-functional requirements

Step 3 — Connections and the routing registry

Step 4 — The mailbox: ordering, idempotency, store-and-forward

Step 5 — Delivery receipts: the ticks

Step 6 — The data model, and which datastore

Step 7 — Trade-offs (each one keeping an NFR)

The complete implementation

Step 8 — Only now, the boxes

Step 9 — Scaling the design, one bottleneck at a time

Step 10 — When a piece fails: designing for failure

The interview corner

Where to go from here

Read this one free

HLD