Skip to content

PointH.md

Latest commit

 

History

History
94 lines (66 loc) · 2.1 KB

PointH.md

File metadata and controls

94 lines (66 loc) · 2.1 KB

PointH: Load Testing & HPA Scaling Validation

Overview

K6 load testing validated Horizontal Pod Autoscaler (HPA) automatically scales Order Service and Payment Service based on CPU and memory utilization under sustained load.

1. Load Test Configuration

Test File: load-tests/k6/order-service-load-test.js

Duration: 4.5 minutes

  • Ramp-up: 30s (0→15 VUs)
  • Sustained: 3min (25 VUs)
  • Ramp-down: 1min (25→0 VUs)

Endpoint: POST /api/v1/orders

HPA Thresholds (lowered for testing):

  • Order Service: CPU 15%, Memory 30%
  • Payment Service: CPU 15%, Memory 30%

2. Test Results

Load Test Metrics:

  • Total Requests: 661
  • Error Rate: 0% (all successful)
  • Response Time: 264ms average, p95=584ms
  • Duration: 4m7.9s
  • Peak VUs: 25

Status Code Distribution:

  • 201 Created: 661 (100%)
  • 4xx/5xx Errors: 0

3. HPA Scaling Results

Order Service:

  • Initial: 2 pods
  • Scaled to: 4 pods during load
  • CPU per pod: 18% (2 pods) → 6-8% (4 pods normalized)
  • Pod creation time: ~25 seconds

Payment Service:

  • Remained at 2 pods (load below thresholds)
  • Metrics actively monitored

Scale-down:

  • Conservative 300s stabilization window
  • Gracefully scaled down to 2 pods after test

4. Key Findings

HPA Triggered: Thresholds exceeded, scaling activated automatically Pods Scaled: 2 → 4 replicas confirmed during load Load Balanced: Requests distributed across all pods Zero Errors: 0% error rate throughout test Graceful Scaling: No disruption during pod creation/deletion Performance Stable: Response times remained consistent

5. How to Run

# Terminal 1: Run load test
cd load-tests/k6
k6 run order-service-load-test.js

# Terminal 2 (optional): Watch HPA
kubectl get hpa -n ecommerce -w

# Terminal 3 (optional): Watch pods
kubectl get pods -n ecommerce -l app=order-service -w

6. Requirements Compliance

Load testing tool (K6) Sustained traffic (25 VUs, 3+ minutes) HPA scaling confirmed (2→4 pods) Resilience demonstrated (0% errors) Automatic scaling (no manual intervention) Zero downtime during scaling