ttRPC, a CNCF project by containerd, was built for low memory environments as an alternative of gRPC. It finds many applications such as: within containerd CRI itself, in the VM runtime by kata containers where the containerd shim communicates with kata agent via this protocol, confidential containers and over 4k projects GitHub wide, including ones from Elastic search, Google, Microsoft, Grafana, etc. While gRPC is well-supported with tools for analyzing critical metrics such as memory usage, latency, and CPU profiling, ttRPC lacks such comprehensive analytics tools. To optimize deployments using ttRPC, it is essential to benchmark its performance, security, and other characteristics rigorously. In this talk, we aim to explore different products that utilize ttRPC protocol for its low latency and memory, analyze its performance, security, and other characteristics against various benchmarks and discuss its limitations with respect to gRPC.
