Single points of failure

Guard 08
HTTP/2HTTP/2SQLSQLUsers200 req/sAnycast edge1 vCPU · 1 GBLoad balanc…1 vCPU · 1 GBNode API1 vCPU · 1 GBNode API1 vCPU · 1 GBPostgres1 vCPU · 2 GB
SQLSQLUsers200 req/sAnycast ed…1 vCPU · 1 GBLoad balan…1 vCPU · 1 GBNode API1 vCPU · 1 GBNode API1 vCPU · 1 GBPostgres1 vCPU · 2 GB

Fix the board. Turn every goal green.

Traffic
200 req/s · 70% reads
Limit
Node API answers about 1,000 req/s
Bill
  • Anycast edge4 USD/mo
  • Load balancer (L7)4 USD/mo
  • Node API × 28 USD/mo
  • Postgres8 USD/mo
24 of 48 USD/mo

Goals

Not tested yet

The share of requests that get a correct answer.99 of 100 requests finish faster than this.Saved data that was lost later.What the servers and services cost each month.

Orders failed twice in two minutes, each time for 30 s. Expect both faults again. Keep availability at 99.5 percent and p99 under 150 ms, and lose no orders. Budget: 48 USD/month. At most 5 edits.

02:40 UTC. Orders fail at 100 percent for 30 s, recover, then fail again. Graph: error rate with two flat walls.

Parts you can add: Load balancer (L7), API gateway, Node API. At most 6 parts. Budget: 48 USD/mo.

Tap a part to add it, or drag it onto a trace.Click a part to add it, or drag it onto a trace.