The AI architect's design
Primary DB overloaded
- p95 latency (server)
- 92 ms
- target < 250 ms
- Error rate
- 2.7%
- target < 1%
- Est. cost / month
- $650
- budget $950
Challenge · intermediateFix the AI
An AI assistant designed this for a ticket-booking site at 3,600 requests per second and called it "production ready". It throws errors at peak. Find the bottleneck and fix it without blowing the budget.
Handle 3,600 req/s with p95 under 250 ms, errors under 1%, every part under 90% busy, for at most $950/month.
Need a hint?
The cache helps reads, but every miss and every write still lands on one primary database. Where else could reads go?