Skip to main content

Distributed instances

Distributed Instances​

Goma Gateway supports Redis-based distributed rate limiting and caching, enabling scalable deployments across multiple instances or nodes.

By connecting to a shared Redis backend, the Gateway synchronizes request throttling and cache states, ensuring consistent behavior in high-availability and load-balanced environments.

This makes Goma Gateway well-suited for modern cloud-native, containerized, or multi-instance deployments.

Redis Integration​

To enable distributed capabilities, configure the redis section in your gateway configuration. This is optional, but highly recommended when running multiple Gateway instances.

Example Configuration​

version: 2
gateway:
  # Redis connection for distributed rate limiting and caching
  redis:
    addr: redis:6379         # Redis server address (host:port)
    password: password       # Optional password for Redis authentication
    db: 0                    # Redis database index (default is 0)
    flushOnStartup: false  # Whether to flush Redis DB on startup (use with caution, default: false)
    tls:                     # Optional: TLS is used only when all three are set
      clientCa: /path/to/ca.crt
      clientCert: /path/to/client.crt
      clientKey: /path/to/client.key

addr and password accept ${VAR} environment references. The environment variables GOMA_REDIS_ADDR, GOMA_REDIS_PASSWORD, and GOMA_REDIS_DB take precedence over the configuration file.

Features Enabled by Redis​

  • Distributed Rate Limiting: The rateLimit middleware throttles requests globally across instances.
  • Shared Caching: The httpCache middleware stores responses in Redis, shared between nodes.
  • Shared OIDC sessions: The oidc middleware can keep sessions in Redis, so a login works on every instance.
  • Real-time visitors gauge: gateway_realtime_visitors_count counts distinct visitors across all instances (see Monitoring).
  • Analytics: The per-request analytics event stream uses Redis as its transport.

Notes​

  • If Redis is not configured, or the gateway cannot connect to it at startup, rate limiting and caching fall back to in-memory state local to each instance, and a warning is logged.
  • Redis must be reachable from all Gateway instances for consistent behavior.
  • Redis Sentinel and Redis Cluster are not supported; addr points to a single Redis endpoint.