Istio's Ambient Mesh: The Real Cost of Sidecar-less Tracing
summary
Benchmarks compare sidecar mode to Ambient Mesh using ztunnel (Rust) + eBPF and waypoint proxies, showing lower steady CPU/memory for ambient L4 and slightly higher L7 median latency (2.4 ms vs 2.75 ms) but a startling P99.9 = 25 ms spike in sidecar mode. Tests include before-and-after cluster utilization and latency curves; production rollout works but requires waypoint autoscaling to avoid overload. Observability needs adding waypoints where L7 traces matter and correlating mesh telemetry with app context to close the blind spot; upgrades run with zero downtime.
AI-written summary · about
chapters (11)
- 0:00Introduction and speakers
- 3:20Service mesh challenges overview
- 8:17Ambient mesh technical deep dive
- 15:58Waypoint proxy scaling and latency
- 16:20Benchmarking sidecar vs ambient modes
- 20:17Cost savings and resource efficiency
- 22:46Locality-based load balancing
- 24:12Debugging workflow improvements
- 26:23Zero downtime upgrades
- 27:07Recommendations and security
- 28:12Q&A session
related talks
40mA Packet Eye View of the Istio Ambient Mesh
08mLightning Talk: A Practical Guide on How to Monitor and Compare Service Mesh...
27mA Service Mesh Benchmark You Can Trust
36mFine Grained Mesh Metrics for Better Visibility With Native Performance
26mAre You Sure About Your Mesh Performance?
39mThe evolution of service mesh with eBPF
11mEnvoy Namespaces - Operating an Envoy-based Service Mesh at a Fraction of the Cost
28mOptimizing Istio Autoscaling: From Resource-Centric To Connection-Aware