githubEdit

Higress

AI Gateway | AI Native API Gateway

Introduction

Higress is an AI-native API gateway built on Envoy and Istio, originally developed by Alibaba and now a CNCF project. It unifies traffic gateway, microservice gateway, and security gateway into a single architecture, with first-class support for AI/LLM routing and management. Higress is fully compatible with Kubernetes Ingress and Gateway API standards.

Key Features

  • AI Gateway: Built-in support for LLM proxy, token rate limiting, prompt caching, and multi-model load balancing

  • Kubernetes Native: Full compatibility with Ingress, Gateway API, and Istio resources

  • Wasm Plugin System: Extend gateway functionality using WebAssembly plugins (Go, Rust, JS)

  • Hot Reload: Configuration changes take effect without connection drops

  • Service Discovery: Integrates with Nacos, Consul, Eureka, and Kubernetes service discovery

  • Rich Protocol Support: HTTP, gRPC, WebSocket, and Dubbo protocol proxying

Installation

Helm Installation

helm repo add higress https://higress.io/helm-charts
helm repo update

# Install in standalone mode (without Istio dependency)
helm install higress higress/higress \
  -n higress-system --create-namespace

# Install with Istio integration
helm install higress higress/higress \
  -n higress-system --create-namespace \
  --set global.istioEnabled=true

Configuration

Basic Ingress

AI Route Configuration

Route requests to different LLM providers with fallback:

Wasm Plugin Example

Apply a Wasm plugin to an Ingress route:

Reference:

Last updated