Deep dive handbook

Reliability and performance tests

Runnable example of spike test

Script and how to run

The following script, executed with tool k6 , simulates a basic spike test:

script.js

import http from 'k6/http';
import { sleep } from 'k6';

export const options = {
  stages: [
    { duration: '1m', target: 1000 },
    { duration: '30s', target: 0 },
  ],
  thresholds: {
    http_req_failed: ['rate<0.01'],
    http_req_duration: ['p(95)<1000'],
  },
};

export default function () {
  http.get('https://quickpizza.grafana.com/contacts.php');
  sleep(1);
}

Install tool k6 and download script.js , then run it locally with:

  k6 run script.js

Analysis

After being run k6 produces result .

Test summary

Spike profile:

• Up to 1000 virtual users (VUs).
• 2 stages over 1m30s, followed by a 30s graceful ramp-down.
• Total duration: ~2m (including graceful stops).

40,124 iterations completed.

Observed behavior

• The system handled an abrupt surge to 1000 concurrent VUs without failures.
• No request failures (http_req_failed=0,00%).
• Response times remained stable and well below defined thresholds during the spike and recovery.

Metrics highlights

Requests: 40,124 total (~441 requests/sec).

Response time:

• Average: 129,22ms
• Median: 127,6ms
• 95th percentile (p95): 141,03ms (well below 1000ms threshold).

Iteration duration:

• Average: 1,13s
• 95th percentile: 1,15s.

Data transferred

• Total received: 45 MB (~495 kB/sec).
• Total sent: 2,1 MB (~23 kB/sec).

Overall analysis

The system demonstrated excellent resilience during the spike test, absorbing a sudden and significant traffic surge without performance degradation. Response times remained consistent, and no errors occurred, suggesting good scalability and robust architecture.

Suggestions and takeaways

• The system is well-prepared to handle short-term traffic spikes.
• Next steps could include stress tests with longer sustained periods or testing simultaneous spikes to confirm consistency under repeated surges.
• Backend monitoring during spikes could provide additional insight into capacity limits and areas for optimization.