sydepsystem design atlas

Service Mesh

A dedicated infrastructure layer for service-to-service traffic.

AdvancedPlatform
Mesh Control PlaneNetworkingService AComputeSidecarNetworkingSidecarNetworkingService BCompute

Every service gets a sidecar proxy handling mTLS, retries, load balancing and telemetry for all its traffic, all controlled centrally by a control plane -- moving networking concerns out of application code entirely.

When to use it

  • Dozens of services need consistent mTLS, retries and traffic policy without per-service code
  • You need fine-grained traffic shaping (canary, mirroring) at the infrastructure layer

Trade-offs

  • Meaningful operational complexity and resource overhead for smaller service counts
  • Debugging traffic issues now spans app code, sidecar and control plane

Components used

Service MeshContainer Service

How it works

  • Every service instance gets a proxy that intercepts all inbound and outbound traffic, forming the data plane.
  • A control plane distributes policy — routing rules, mTLS certificates, retry and timeout configuration — to every proxy.
  • Traffic management, encryption and telemetry become infrastructure concerns rather than application code.

Used in the wild

  • Large microservice estates needing consistent mTLS without touching every service.
  • Fine-grained traffic shifting for canaries and A/B tests at the network layer.
  • Uniform golden-signal telemetry across services nobody wants to re-instrument.

Good to know

  • It is essentially the sidecar pattern applied uniformly plus a control plane to configure it centrally.
  • The operational complexity is significant and frequently underestimated. Below roughly a few dozen services, a mesh usually costs more than it returns.