The sizing depends on the distribution of the transaction volume in time. You would need to have an estimate of the peak transaction volumes and their types (as some are faster than others) in order tocalculate the production HW sizing. Or in another words you need requirements (X transactions per minute in the peak time) and work from there backwards.
Reference test server (same for both DOT Core Server and DOT OCR Server)
- 2 vCPU (Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz)
- 4GB RAM
Server Reference Performance Results
- Face Capture <200ms
- Verification <10ms
- Liveness check < 350ms
- Document OCR ~ 5 seconds (two sides of one document)