jargon

Comparison

Service meshvsSidecar proxy

Service mesh

you get mutual TLS, retries and per-route traffic splitting without changing a line of application code, because a proxy sits next to each instance.

A layer of proxies deployed alongside services, handling routing, retries, mTLS and telemetry uniformly. It moves cross-cutting concerns out of each service's language and libraries. The cost is a substantial amount of operational machinery and a new class of failures that look like application bugs but live in the proxy.

Full entry →

Sidecar proxy

every connection your service makes actually goes through a proxy in the same pod, which is where the retries and the TLS happen.

The per-workload proxy that intercepts inbound and outbound traffic, giving a mesh its identity, encryption, routing and telemetry without application changes. It is the data plane of a service mesh, and it is why mesh features apply uniformly to services written in five languages. Its costs are a hop of latency, memory and CPU on every pod, and a genuinely difficult startup and shutdown ordering problem with the application beside it.

Full entry →

Related comparisons