Feature refresh interval

Model 18
RedisSQLPartner apps4,000 req/sAPI server2 vCPU · 2 GBRedis cache1 vCPU · 2 GBPostgres2 vCPU · 4 GB
HTTP/2RedisSQLPartner apps4,000 req/sAPI server2 vCPU · 2 GBRedis cache1 vCPU · 2 GBPostgres2 vCPU · 4 GB

Tune the parts. Turn every goal green.

Traffic
4,000 req/s · 95% reads
Limit
API server answers about 6,700 req/s
Bill
  • API server8 USD/mo
  • Redis cache16 USD/mo
  • Postgres16 USD/mo
40 of 60 USD/mo

Goals

Not tested yet

99 of 100 requests finish faster than this.The share of requests that fail or get no answer.How old the data is when a reader gets it.What the servers and services cost each month.

A fraud model reads a card velocity feature 4,000 times/s. A cached value is recomputed from the event database when it expires. The model loses its gain when the value is older than 30 s. Keep p99 under 150 ms and cost under 60 USD/mo. Tune the cache TTL and the database vCPU.

Each worker adds one network cardDrag or use arrow keys to change the workers

2 workers · 1 TB moves in 1.1 h · 1,000 Mbps per card