Skip to main contentBased on internal benchmarks, a single instance of the Sidecar service—connected to a Redis cluster as the cache layer—can handle:
- Throughput: 1200 requests per second (RPS) steady.
- Latency: p95 under 10ms.
- Deployment model: Sidecar container next to your app or standalone internal service.
- CPU and memory: 1 vCPU, 512MB RAM per instance is sufficient.
- CPU usage: 70% average, commonly between 40%–75% under constant load.
- Memory footprint: ~250MB per instance.
- Scalability: Stateless and horizontally scalable.