API Gateway vs Service Mesh — When to Use What

API Gateway vs Service Mesh — When to Use What

API gateways and service meshes solve different problems, but both sit in the request path. The gateway is for north-south traffic, while the mesh manages east-west traffic between services.

API Gateway Responsibilities

API gateways focus on client-facing concerns:

• Authentication and authorization.
• Rate limiting and quota enforcement.
• Request shaping and aggregation.
• API version routing and deprecation.

Service Mesh Responsibilities

Service meshes focus on service-to-service communication:

• Mutual TLS and workload identity.
• Traffic splitting for canaries.
• Retries, timeouts, and circuit breaking.
• Distributed tracing and telemetry.

When an API Gateway Is Enough

If you have a small number of services and only need centralized ingress control, an API gateway alone can be sufficient. It provides a single point to enforce security and routing policies.

When You Need a Service Mesh

A service mesh becomes valuable when:

• You have many services with complex communication patterns.
• You want uniform mTLS without modifying application code.
• You need advanced traffic management for deployment strategies.

Using Both Together

Mature platforms use both:

• Gateway handles client traffic and edge concerns.
• Mesh handles intra-service policy, telemetry, and routing.

Spring Boot Example with Gateway and Mesh Metadata

@RestController
@RequestMapping("/inventory")
public class InventoryController {
    @GetMapping("/{sku}")
    public InventoryView getInventory(@PathVariable String sku) {
        return new InventoryView(sku, 42);
    }
}

apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: inventory-policy
spec:
  host: inventory
  trafficPolicy:
    loadBalancer:
      simple: LEAST_REQUEST
    outlierDetection:
      consecutive5xxErrors: 5
      interval: 5s
      baseEjectionTime: 30s

Summary

Use an API gateway for external access control and a service mesh for internal service-to-service reliability. In large microservice platforms, both layers are complementary rather than competing.