# Apache APISIX

Apache APISIX is an open-source API gateway and AI gateway for routing, securing, observing, and governing API, microservice, and LLM traffic. It helps teams manage dynamic gateway policies through a plugin-based architecture built on NGINX and etcd.

- Findings24: <https://findings24.com/pt/tools/apache-apisix>
- Site: <https://apisix.apache.org/>
- Preços: Freemium
- Categorias: Gateways e roteamento de IA

## O que é Apache APISIX?

Apache APISIX is an open-source API gateway and AI gateway from the Apache Software Foundation. It sits between clients and upstream services to route and manage traffic for APIs, microservices, and documented LLM providers or OpenAI-compatible endpoints. Its core capabilities include dynamic routing, load balancing, authentication, rate limiting, observability, and protocol handling.

The gateway uses NGINX and etcd as core architectural components. Its plugin pipeline supports hot updates and configuration changes without gateway restarts. APISIX can be run on bare metal or Kubernetes and includes a Docker-based getting-started path for creating and testing a first route.

## O que Apache APISIX pode fazer?

### Dynamic routing and configuration

Match requests to upstream services and distribute configuration through etcd. Routes and plugins can be updated without restarting the gateway.

### Traffic management

Support load balancing, health checks, circuit breakers, traffic splitting, and rate controls such as request-count, concurrency, and token-based limits.

### Authentication and security

Use plugins for JWT, key authentication, OpenID Connect, Keycloak, Casbin, CORS, request validation, IP restrictions, and related access-control workflows.

### Observability integrations

Export metrics, traces, and logs through integrations including Prometheus, OpenTelemetry, Zipkin, Apache SkyWalking, Datadog, Kafka, ClickHouse, and Elasticsearch.

### AI gateway capabilities

Proxy traffic to documented LLM providers and OpenAI-compatible endpoints, with multi-provider routing, LLM load balancing, retries and fallbacks, token rate limiting, prompt security, and prompt transformation.

### Protocol and deployment support

Run on bare metal or Kubernetes and handle HTTP-to-gRPC transcoding, WebSocket, gRPC, Dubbo, MQTT proxying, and ARM64 environments.

## Casos de uso

### API and microservice ingress

Place a centralized gateway in front of HTTP APIs and microservices to apply routing, authentication, rate limits, and observability policies.

### Kubernetes traffic management

Deploy APISIX with Kubernetes and manage traffic policies through its gateway and plugin model while services change behind the gateway.

### Multi-provider LLM access

Give AI applications and agents one gateway path to supported LLM providers or OpenAI-compatible endpoints, with provider and model distribution, retries, and fallback behavior.

### Local or private model proxying

Route OpenAI-compatible chat traffic to a private LocalAI deployment, with separate client and upstream authentication as described in the integration documentation.

### Protocol mediation

Use gateway capabilities such as HTTP-to-gRPC transcoding, WebSocket, gRPC, Dubbo, or MQTT proxying when clients and upstream services use different protocols.

## Perguntas frequentes

### How can I try Apache APISIX?

The getting-started guide provides a Docker-based quickstart. It starts APISIX and etcd containers, verifies the gateway, creates a route to an upstream service, and demonstrates adding a rate-limiting plugin.

### Does APISIX require etcd?

The documented architecture uses etcd as the configuration store and distributes configuration across the cluster. The quickstart script starts an etcd container automatically.

### Can APISIX change routes or plugins without a restart?

Yes. APISIX supports dynamic routing and hot-loaded plugins, and its architecture describes configuration updates through etcd without gateway restarts.

### Can it be used for LLM traffic as well as APIs?

Yes. Apache APISIX includes AI gateway features for proxying documented LLM providers and OpenAI-compatible endpoints. The documented capabilities include LLM load balancing, retries and fallbacks, token rate limiting, prompt security, and prompt transformation.

## Informações rápidas

- **Product type**: Open-source API gateway and AI gateway
- **Project**: Apache Software Foundation top-level project
- **License**: Apache 2.0
- **Core architecture**: NGINX and etcd
- **Deployment options**: Bare metal and Kubernetes
- **Plugin library**: 100+ open-source plugins