Liberator benchmarks
To help with capacity planning, Caplin have tested Liberator’s performance at different message rates, batch intervals, and message sizes.
Download the full report: Liberator 7.1 Benchmarks (PDF).
Caveat
While the benchmark tests were designed to emulate real-world traffic and user scenarios, they were conducted using specific hardware running in an isolated environment, and therefore no guarantees can be made that identical results will be achieved in other environments.
Benchmark hardware
Liberator 7.1 was tested on a DELL PowerEdge R7415 server with the following specification:
Vendor |
DELL |
|---|---|
Model |
PowerEdge R7415 |
Processors |
AMD EPYC 7551P 32-Core 2GHz/2.55Ghz/3GHz |
NUMA nodes |
4 |
Memory |
64GB |
Operating system |
CentOS 7.4.1708 |
Network card |
Broadcom NetXtreme BCM5720 Gigabit Ethernet PCIe |
Results summary
A high-level summary of the benchmarks is reproduced below. For the full results, see the report Liberator 7.1 Benchmarks (PDF).
| Scenario | Message rate | Batching | Users | Mean latency |
|---|---|---|---|---|
Low |
1 msg/user/sec |
- |
100,000 |
3.7ms |
Med |
10 msg/user/sec |
- |
50,000 |
4.8ms |
High |
50 msg/user/sec |
- |
12,500 |
6.8ms |
Very High |
100 msg/user/sec |
- |
6,000 |
5ms |
| Scenario | Message rate | Batching | Users | Mean latency |
|---|---|---|---|---|
High |
50 msg/user/sec |
100ms |
28,000 |
53ms |
Very High |
100 msg/user/sec |
100ms |
18,000 |
54ms |
| Message batching makes more efficient use of Liberator and network resources. It allows Liberator to handle higher message rates at the expense of extra latency introduced by the batching interval. For more information on batching, see Liberator bursting (batching). |
See also: