Scaling

Horizontal scaling

Scale the runtime container to handle more concurrent executions:

docker compose up -d --scale runtime=3

The API round-robins requests across runtime instances.

Considerations

Each runtime instance respects MAX_CONCURRENT_EXECUTIONS
With 3 instances × 10 concurrent = 30 total concurrent executions
All instances share the same storage backend
No shared state between runtime instances (stateless)

Vertical scaling

Adjust resource limits in docker-compose.yml:

runtime:
  deploy:
    resources:
      limits:
        cpus: '8'
        memory: 16G
      reservations:
        cpus: '2'
        memory: 4G

Also increase concurrent executions:

MAX_CONCURRENT_EXECUTIONS=20

Database scaling

For higher database throughput:

Option 1: Tune PostgreSQL

Add these settings to your PostgreSQL service:

postgres:
  command:
    - "postgres"
    - "-c"
    - "max_connections=200"
    - "-c"
    - "shared_buffers=256MB"

Option 2: External database

Use a managed PostgreSQL service (AWS RDS, Azure Database, Cloud SQL):

# Remove the postgres service from docker-compose.yml
# Update DATABASE_URL to point to external database
DATABASE_URL=postgresql://user:pass@external-host:5432/flowlike

Load balancing

For multiple API instances, use an external load balancer:

docker compose up -d --scale api=2

Then configure nginx or similar:

upstream flowlike-api {
    server localhost:8080;
}

server {
    listen 80;
    location / {
        proxy_pass http://flowlike-api;
    }
}

Production recommendations

Workload	API replicas	Runtime replicas	Runtime resources
Development	1	1	2 CPU, 4GB
Small team	1	2	4 CPU, 8GB
Medium	2	4	4 CPU, 8GB
Large	3+	6+	8 CPU, 16GB

For large production workloads, consider Kubernetes deployment for better orchestration, auto-scaling, and isolation.