Skip to main content

Deployment Models: RunMCP, OpenAPI, API-first

RunMCP supports a range of deployment models, from simple single-node gateways to highly available, multi-node, cloud-native clusters.

Common Deployment Models​

  • Single-node: All traffic routed through one gateway instance (good for dev/test)
  • Multi-node: Multiple gateways behind a load balancer for high-availability
  • Cloud-native: Kubernetes or container-based deployments with dynamic scaling
  • Hybrid: Mix on-prem and cloud, or combine with other API gateways
tip

For production, use at least two gateway instances behind a load balancer for redundancy and zero-downtime upgrades.

Example: Docker Compose (Dev)​

version: '3.8'
services:
  runmcp:
    image: larebelion/runmcp:latest
    ports:
      - "8080:8080"
    volumes:
      - ./config:/app/config

Example: Kubernetes Deployment​

apiVersion: apps/v1
kind: Deployment
metadata:
  name: runmcp-gateway
spec:
  replicas: 3
  selector:
    matchLabels:
      app: runmcp-gateway
  template:
    metadata:
      labels:
        app: runmcp-gateway
    spec:
      containers:
        - name: runmcp
          image: larebelion/runmcp:latest
          ports:
            - containerPort: 8080
          volumeMounts:
            - name: config
              mountPath: /app/config
      volumes:
        - name: config
          configMap:
            name: runmcp-config
caution

Use persistent storage for configuration and logs in production deployments.

Best Practices​

  • Use environment variables for secrets and config
  • Monitor gateway health and scale horizontally as needed
  • Automate deployments with CI/CD pipelines

Further Reading​