Performance

Optimizing performance on Kova deployments.

Resource Allocation

CPU

• Fractional cores supported (e.g., 0.5, 1.5)
• Higher allocation = faster processing
• Over-allocation wastes money

resources:
  cpu:
    units: 1.5    # 1.5 CPU cores

Memory

• Specify in Mi or Gi
• Under-allocation causes OOM crashes
• Monitor actual usage to optimize

resources:
  memory:
    size: 1Gi    # 1 gigabyte

Storage

• Persistent storage has performance overhead
• Use for data that must survive restarts
• Container filesystem is faster for temp data

Container Optimization

Image Size

• Use Alpine-based images when possible
• Remove unnecessary dependencies
• Use multi-stage builds

# Bad: 1GB image
FROM node:20

# Good: 150MB image
FROM node:20-alpine

Startup Time

• Minimize initialization logic
• Use health checks appropriately
• Consider warm-up strategies

Monitoring

Track these metrics in the dashboard:

• CPU usage: Percentage of allocated CPU
• Memory usage: Current vs allocated
• Response time: Latency of your endpoints
• Error rate: Failed requests percentage

Scaling

For high-traffic applications:

• Increase replica count in SDL
• Each replica gets placed on potentially different providers
• Load balancing is automatic

deployment:
  web:
    anywhere:
      profile: web
      count: 3    # 3 replicas