External request source
Backend & Runtime / Knowledge library
Envoy
service traffic routing load balancing and observability
Why learn Envoy?
Useful for managing communication between services or at an edge. Retry and timeout policies must not amplify failures or duplicate consequential operations.
Where it came from
Envoy is an out-of-process proxy with network and application-layer features. It supports service routing, load balancing, discovery, and traffic observability.
Build your understanding
Work through these exercises in order. Try the task before opening a reference.
- Route a request to a test backend.
- Configure a bounded upstream timeout.
- Inspect traffic metrics.
- Test behavior when a backend fails.
- Check whether a retry policy is safe for a write operation.
Documentation and sources
- Official project documentation (www.envoyproxy.io)
Catalog research notes
- Research snapshot: September 9, 2026. This entry summarizes the linked primary documentation; it does not assert a latest release version.
Connect the ideas
Alternatives
Explore next
Real World Example
Unified Traffic Routing for Polyglot Microservices
A global retail platform operating a polyglot architecture with services written in Java, Go, and Python.
The business problem
The engineering team struggles with inconsistent load balancing and observability across disparate service frameworks, leading to silent failures during inter-service communication.
How this topic solves it
Deploy Envoy as a sidecar proxy alongside every application instance to provide a uniform L7 communication bus, standardizing routing, load balancing, and telemetry collection regardless of the underlying application language.
- Deploy an Envoy sidecar process alongside each application container to intercept all inbound and outbound traffic.
- Configure the HTTP connection manager to handle L7 routing and protocol bridging between HTTP/1.1 and HTTP/2.
- Enable the outlier detection subsystem to perform passive health checking and automatically eject unhealthy upstream hosts.
- Integrate the Envoy administration port with a centralized statistics sink to aggregate observability data across the entire mesh.
Traffic flows from the client through the local Envoy proxy to the application and then to the destination service via its own Envoy proxy.
Edge proxy handling TLS and routing
Internal business logic
Service-to-service proxy
Target service
- Client → Request → Envoy Sidecar 1
- Envoy Sidecar 1 → Localhost → App Service A
- App Service A → Egress → Envoy Sidecar 2
- Envoy Sidecar 2 → Localhost → App Service B
Expected business outcome
A transparent communication mesh where developers focus on application logic while Envoy ensures consistent traffic management and visibility.
Limitations and checks
The out-of-process architecture introduces a small latency overhead per hop; verify this by monitoring the Envoy-reported request duration statistics via the administration interface.