Our client has asked us to implement load balancing for several microservices, and our organization is considering Envoy. I'm new to Envoy and have only a basic understanding of cloud architecture and services. I'm familiar with NGINX as a load balancer, so I'd like to understand how Envoy compares, how it fits with Apigee, and what concepts or hands-on exercises I should start with.
3 Answers
If you’re using a managed Apigee or managed Kubernetes setup, you may not need to configure Envoy directly. In many deployments, the platform or infrastructure team manages it, while application developers mainly define service routes and make sure their services behave correctly. You’ll usually need deeper Envoy knowledge only for requirements such as custom routing, long timeouts, large request bodies, retries, or advanced traffic policies.
At a high level, Envoy is similar to NGINX, HAProxy, or Traefik: it’s a proxy that can handle reverse proxying, load balancing, routing, TLS, retries, and observability. Apigee can use Envoy-based components to route and manage API traffic, but the exact relationship depends on which Apigee product and deployment model you’re using.
A useful way to study it is to begin with the reverse-proxy model: listeners receive traffic, clusters represent upstream services, and routes decide where requests go. Unlike traditional NGINX configurations, Envoy commonly receives configuration dynamically through the xDS APIs. Its features are also organized into filter chains, which process network, TLS, or HTTP traffic. Start with a small local setup, route requests to two test services, and then experiment with load balancing, health checks, retries, and metrics.
That helps. I wasn’t sure whether Envoy was something application developers were expected to manage or just understand at a high level.