💬 Chat app system design
A simplified, healthy design: 10 app servers behind a load balancer, a cache (80% hit rate), an async write queue, a CDN for static files, a primary database. At 1,400 requests per second it runs at p95 49 ms with 0.0% errors for about $610/month.
Healthy at 1,400 req/s
- p95 latency (server)
- 49 ms
- Error rate
- 0.0%
- Est. cost / month
- $610
- excl. bandwidth
App servers
400 req/s each · $40/mo
Hit rate
DB read replicas
+1,250 reads/s each · also applies every write · $120/mo
Primary DB: 1,250 queries/s (1 query per request). How the numbers work
How it works
- Sending a message is a write, and people send a lot of them, so writes are a big share of traffic.
- A queue acknowledges each message instantly; workers write it to the database in the background. It does not add database capacity.
- Photos and stickers are static files, served from a CDN close to the user.
Push traffic to 5k requests/s. Replicas will not save you here: every write lands on one primary.
Use the Build and Traffic tabs above. The overloaded part turns red and starts dropping requests.