Cybersecurity glossary

What is a Service Mesh?

Learn what a service mesh is, how sidecars or ambient proxies handle service-to-service traffic, which problems it solves, and the security and operational trade-offs to expect.

Application securityUpdated July 22, 2026
Also known asMesh networking for servicesService-to-service mesh

Definition

A service mesh is an infrastructure layer that manages service-to-service communication—typically via sidecar or node proxies—providing features such as mTLS, traffic routing, retries, observability, and policy enforcement without changing each application’s business code.

Why service meshes exist

Microservices multiply connections. Each team rewriting TLS, retries, timeouts, circuit breakers, and metrics collection produces inconsistent security and brittle failure modes.

A service mesh standardizes those cross-cutting network concerns in proxy infrastructure. Applications keep business logic; the mesh handles how bytes move securely and observably between services.

How a mesh handles a call

1

Service A sends a normal request

The app talks to localhost or a cluster DNS name as usual.

2

Outbound traffic is intercepted

iptables/eBPF or explicit proxy config steers packets to the mesh proxy.

3

mTLS and identity are applied

Certificates identify workloads; policies decide which identities may connect.

4

Traffic rules execute

Retries, timeouts, canaries, fault injection, and load balancing run uniformly.

5

Peer proxy delivers to Service B

The destination sidecar terminates the mesh hop and forwards to the app port.

6

Telemetry is emitted

Golden signals and distributed traces appear without custom instrumentation for every hop.

Mesh vs API gateway vs library approach

ApproachPrimary scopeTrade-off
Service meshEast-west service trafficStrong consistency; higher platform complexity
API gatewayNorth-south external entryGreat edge control; less internal coverage alone
Shared client librariesIn-process featuresNo sidecars; language sprawl and version drift

Capabilities teams adopt meshes for

Universal mTLS

Encrypt and authenticate service calls even on flat cluster networks.

Identity-aware policy

Allow only payment-service to call ledger-service—regardless of pod IP.

Safe traffic shifting

Canary and weighted routes reduce release risk across many languages.

Default observability

Consistent metrics and traces for every hop without rewriting apps.

Security caveats

Mesh is not app authz

Workload identity ≠ user permission on a specific document or tenant row.

Control plane is critical

Compromise of mesh control components can weaken or redefine trust.

Certificate lifecycle

Short-lived workload certs need reliable rotation and breakage monitoring.

Escape paths

Pods that bypass the proxy undo guarantees—enforce with network policy.

Adoption checklist

  • Start with clear goals (mTLS, telemetry, canaries)—not mesh for its own sake.
  • Inventory protocols (HTTP/gRPC) and ensure proxy support matches your stack.
  • Enable mTLS gradually with reporting modes before strict enforcement.
  • Define who owns mesh upgrades, policies, and break-glass procedures.
  • Pair mesh identity with application authorization for user/tenant actions.
  • Budget CPU/memory for proxies and watch tail latencies after rollout.
  • Lock down control-plane access and audit policy changes.
  • Test failure modes: proxy crash, cert outage, and partial namespace onboarding.

The practical takeaway

A service mesh moves service-to-service networking concerns—mTLS, routing, retries, telemetry—into shared proxies so microservices behave consistently. It is powerful platform infrastructure, not a substitute for application authorization.

Adopt a mesh when uniformity and identity-based east-west security justify the operational cost; keep the control plane hardened and applications responsible for business access control.

Related security terms

Frequently asked questions

What is a service mesh in simple terms?

It is a shared network helper for microservices. Instead of every service coding its own retries, encryption, and metrics, a mesh proxy next to each service handles those jobs consistently.

Do I need a service mesh?

Not always. Small systems may be fine with libraries and a gateway. Meshes help when you have many services and need uniform mTLS, traffic policy, and telemetry.

What is a sidecar?

A proxy container co-located with each service instance that intercepts inbound and outbound traffic for that workload.

How is a mesh different from an API gateway?

Gateways usually manage north-south (external) entry. Meshes focus on east-west (service-to-service) traffic inside the cluster—though products increasingly overlap.

Does a mesh replace application authorization?

No. mTLS proves service identity. Object-level and business authorization still belong in application logic or a dedicated policy engine with rich context.

What are the downsides?

Added latency, resource cost, operational complexity, and a new critical dependency to upgrade safely.

Is ambient mesh the same as sidecars?

Ambient designs move some proxy functions to node or shared components to reduce per-pod sidecars, with different trade-offs.

References

Explore authoritative guidance and frameworks related to service mesh.

Explore every security definition

Return to the glossary to search by term, alias, starting letter, or security category.

Browse glossary