Architecture lab

Challenge 2: An order service that survives failuresTarget 5,000 RPS

0% busy
Users
1x inst5,000 RPS
100% busy
Load balancer
1x inst5,000 RPS
132% busy
App servers
4x inst5,000 RPS
152% busy
Database
1x inst3,793 RPS

Design summary

Components

  • Users: 1 × Users
  • Load balancer: 1 × Load balancer, 5,000 requests a second, 100% busy
  • App servers: 4 × App server, 5,000 requests a second, 132% busy
  • Database: 1 × Relational database, 3,793 requests a second, 152% busy

Connections

  • Users sends requests to Load balancer
  • Load balancer sends requests to App servers
  • App servers sends requests to Database

2,500 of 5,000 requests a second served, average 231 ms, p99 601 ms, 425 USD a month. Challenge not yet passed: throughput, availability, losing one instance.