INFRASTRUCTURE PLANNING

Auto-size

Find the cheapest cluster that meets your SLA at peak load. For each machine type, the twin searches for the fewest nodes that pass, then prices them.

CURRENT SERVICEExample checkout API

Edit setup

Measured baseline: 2 × 4 vCPU at 40 req/s · CPU 62% · p95 420 ms
SLA thresholds: p95 ≤ 800 ms · errors ≤ 1% · CPU ≤ 80%

Requirement

Peak load to carry

req/s

Fewest nodes

2 keeps one node of redundancy.

Most nodes

Pool sizing per node

Worker threads per vCPU

DB connections per node

Request timeout

Machine types and prices

These prices are illustrative, not any provider's. Replace them with your own for real numbers.

Name	vCPU	RAM GB	$/hour	Remove

Estimated time: about 9 s.

Play with your cluster

p95: —
CPU: —
Threads: —
Errors: —

Each change re-runs the SimPy simulation for this exact setup. Deltas compare with where you started.

Set the peak load and the machine types you can buy, then search.