Easegress — Cloud-Native Traffic Orchestration System

What it is

Easegress is a cloud-native traffic orchestration system written in Go. It functions as an API gateway, load balancer, and service mesh sidecar, processing requests through composable filter pipelines. Each pipeline chains together filters for routing, rate limiting, authentication, and request transformation.

Easegress targets platform engineers and DevOps teams who need fine-grained control over HTTP traffic. Its declarative YAML configuration and CLI tool (egctl) make it accessible to teams already familiar with Kubernetes-style resource definitions.

How it saves time or tokens

Instead of stitching together separate tools for rate limiting, load balancing, canary releases, and request rewriting, Easegress consolidates them into a single binary. The pipeline model means you define traffic rules once in YAML and apply them across environments. Its distributed architecture uses Raft consensus, so you avoid building your own HA layer.

How to use

1. Install Easegress using the official install script or download the binary directly from the GitHub releases page.
2. Start the server with easegress-server and use egctl to create pipeline resources via YAML manifests.
3. Define an HTTP server listener on your desired port and bind it to your pipeline.

Example

# Install Easegress
bash -c "$(curl -kfsSL https://raw.githubusercontent.com/easegress-io/easegress/main/scripts/install.sh)"

# Start server
easegress-server

# Create a proxy pipeline
cat <<EOF | egctl create -f -
name: demo-pipeline
kind: Pipeline
flow:
  - filter: proxy
filters:
  - name: proxy
    kind: Proxy
    pools:
      - servers:
          - url: http://httpbin.org
EOF

# Bind to port 10080
egctl create httpserver --port 10080 --rules demo-pipeline
curl http://localhost:10080/get

Related on TokRepo

• DevOps AI Tools — More infrastructure automation tools for cloud-native teams
• API and Integration Tools — API gateway and integration resources

Common pitfalls

• Easegress uses its own filter pipeline model, not standard middleware chains. Read the filter documentation before assuming behavior from other gateways.
• Raft consensus requires at least three nodes for production HA. A single-node setup works for development but has no failover.
• YAML pipeline definitions are order-sensitive. Filter execution follows the flow array sequence exactly.