Benchmarks
This page contains performance benchmarks for raccoon.
Table of Contents
WebSocket
Setup
Raccoon benchmarking was done using a client which creates multiple parallel connections to Raccoon in one go and then sends batches of events every 10 seconds. This whole setup was deployed on a Kubernetes cluster running on GCP with one or multiple raccoon pods.
Result
Following are the benchmarking results for various versions
| Raccoon version | Duration | No. of Connections | No. of Raccoon Pods | No. of events/10s | Server Processing Latency(P95) | Server Processing Latency(Upper) | Workers Latency (mean p95) | Workers Latency (max upper) | Memory Used per pod | CPU Cores Used per pod |
|---|---|---|---|---|---|---|---|---|---|---|
| v0.1.0 | 1 hour | 10000 | 1 | 1500000 | ~ 6 - 22 ms | ~ 3 - 913 ms | ~ 1.7 - 7 ms | ~ 2 - 140 ms | ~ 711 - 870 MB | ~ 2.2 |
| v0.1.0 | 1 hour | 50000 | 3 | 25000 | ~ 20 - 30 ms | ~ 3 s | ~ 35 - 50 ms | ~ 35 - 45 ms | ~ 1.0- 1.5 GB | ~ 0.3 - 0.6 |
| v0.1.0 | 1 hour | 50000 | 3 | 100000 | ~ 3 - 5 ms | ~ 2.5 s | ~ 20- 30 ms | ~ 20 - 30 ms | 400 - 500 MB | ~ 0.7 |
| v0.1.0 | 1 hour | 100000 | 5 | 100000 | ~ 3 - 9 ms | ~ 2.5 s | ~ 20- 30 ms | ~ 20 - 30 ms | ~ 1.7 - 2GB | ~ 1 |
| v0.1.2 | 30 min | 10000 | 1 | 1500000 | ~ 1 - 7.13k ms | ~ 1 - 7.13k ms | ~ 0.8 - 9.25 ms | ~ 2 - 224 ms | ~ 960MB -1.2GB | ~ ~ 2.57 |