We performed a performance test in relation to alterante id. We registered 5100 CmHandles with the batch size of 300.
We executed 3 test scenarios: one with an older version of CPS then we tested the latest alternate id feature (with database look-ups instead of the caches) where we registered CmHandles with an alternate id and in the last scenario without an alternate id.
Note that the watchdog timeout was set to 5 second.
Test case | Time taken to POST 17 requests | Time taken for all to be READY |
3.4.1 release (no alternate-id) | 43 seconds (avg. 2,5 sec) | 51 seconds |
latest without alternate-id | 30 seconds (avg. 1,9 sec) | 55 seconds |
latest with alternate-id | 47 seconds (avg. 3,2 sec) | 55 seconds |
Test Environment
# | Environment/Workload | Description |
|---|---|---|
| 1 | Teed on | laptop : |
| 2 | Number of CPS Instance | 1 |
| 4 | Total number of cm handles | 5,100 |
| 5 | Batch-size | 300 cm handles per request |
| 6 | Response payload size | ~2 KB per cm handles |
| 7 | Concurrent request | 1 |
| 8 | Number of DMI PLugins | 1 |
| 9 | Commit ID | I0adcac52a7a49e26301758eafc4684152ddfcaf5 |
| 10 | Commit ID link | Use DB for checking Alternate IDs |