Skip to content

Latest commit

 

History

History
46 lines (39 loc) · 1.06 KB

ARCHITECTURE.md

File metadata and controls

46 lines (39 loc) · 1.06 KB

Kubernetes Autoscaling Architecture

System Components

1. Metrics Server

  • Collects resource metrics from kubelets
  • Exposes metrics through Metrics API
  • Required for HPA functionality

2. Horizontal Pod Autoscaler

  • Monitors specified metrics
  • Calculates desired replica count
  • Adjusts deployment scale

3. NGINX Deployment

  • Runs containerized web server
  • Configured with resource constraints
  • Scales based on HPA decisions

Workflow

  1. Metrics Server collects utilization data
  2. HPA queries metrics every 15 seconds
  3. HPA calculates desired replica count
  4. Deployment scales pods up/down
  5. Process repeats continuously

Resource Calculations

  • CPU: Target 65% of requested (50m) 
    Target CPU per pod = 50m * 0.65 = 32.5m
  • Memory: Target 1Gi absolute 
    Scale if: avg_memory_usage > 1Gi

Scaling Logic

graph TD
    A[Metrics Server] -->|Collect Metrics| B[HPA]
    B -->|Check Thresholds| C{Scale?}
    C -->|Yes| D[Calculate Replicas]
    D -->|Update| E[Deployment]
    C -->|No| F[Wait 15s]
    F -->|Repeat| B
Loading