Performance benchmarks¶
This page summarizes validated benchmark observations for Logstrm. Results are specific to the recorded test runs and are not a production capacity guarantee.
Measured results¶
| Test | Requested rate | Validated observation |
|---|---|---|
| Local Data Plane acceptance | 500, 1,000, 2,500, 5,000 and 10,000 events/s | 15/15 runs validated. Mean accepted rates by requested level were 496.646, 989.985, 2,459.968, 4,847.743 and 9,512.090 events/s. One 1,000 events/s repetition recorded a transport error. |
| End-to-end delivery and drain | 500, 1,000 and 2,500 events/s | 9/9 runs validated with replay and drain verification recorded for every row. Mean achieved rates were 494.976, 989.964 and 2,454.933 events/s. |
The local test measures acceptance by the Data Plane benchmark endpoint; it is not a measurement of downstream delivery. The end-to-end test exercises a configured emitter and sink and includes delivery/drain evidence. The two measurements are not interchangeable.
How to interpret these results¶
- These are observations from the tested configurations and runs, not a service-level commitment or universal capacity limit.
- Performance depends on event shape, configuration, hardware, resource limits, emitter, destination and network conditions.
- The benchmark target of 50,000 events/s has not been demonstrated by the results summarized here.
- Use a representative workload and validate throughput, latency, resource use, delivery, retries and recovery in your own environment before production sizing.
The underlying measurements include repeated local and end-to-end test rows. Summary rates are averages; they do not imply that every individual run achieved the average. Consult your Logstrm distribution contact for evidence details appropriate to your deployment.