How to Do Load Testing with k6

Virtual users, load profiles, thresholds - the k6 setup that validates your system before real traffic hits it.

Most applications are not load tested before they receive real traffic. The result is discovered during a product launch, a press mention, or a Black Friday sale. Load testing with k6 is fast enough to run in CI and provides the data to answer the question every engineering team needs to answer before launch: how many users can this handle?

No fluff. Real test coverage from QA engineers who find bugs before your users do.

The Load Test Types and When to Use Each

Smoke test: 1-5 VU for 1 minute. Validates that the test script works and the application responds. Run before every more expensive test. Load test: expected peak load for 10-30 minutes. Validates the system handles normal traffic. Run weekly or before releases. Stress test: ramp to 200-300% of expected peak. Finds the breaking point. Run monthly or before major launches. Soak test: steady load for 4-8 hours. Finds memory leaks and performance degradation over time. Run quarterly.

At Valletta Software, we focus on:

VU profile: smoke (1-5) load (expected peak) stress (2-3x peak) soak (steady 4-8h) - use each appropriately

k6 script: import http from k6/http - define scenarios thresholds and groups

Thresholds: http_req_duration p95 < 200ms error_rate < 1% - fail the test if violated

Scenarios: ramping-vus (ramp up hold ramp down) constant-arrival-rate (constant RPS) - match real traffic

Correlation: extract dynamic values (session tokens) from responses - not hardcoded

Think time: add sleep(1) between requests - real users are not bots making 100 RPS per VU

Output: k6 cloud or Grafana for visualization - not just terminal output

The Monitoring Setup During Load Tests

A load test without monitoring tells you the application broke. Monitoring tells you why.

We give you more than just people. We give you top performers who drive results.

Prometheus metrics: CPU memory DB connections request queue depth - scrape during load test

Grafana dashboard: latency percentiles error rate saturation - watch in real time during test

DB query time: slow query log enabled during test - load reveals query problems unit tests miss

Error log: tail application logs during test - first error message explains the cascade

APM: Datadog New Relic or Elastic APM - trace which endpoint is the bottleneck

Connection pool: monitor DB and Redis connection pool exhaustion - common scaling limit

Baseline comparison: compare metrics to pre-test baseline - not just absolute values

You lead the work. We handle the rest. From EUR 45/h.

EU-incorporated in Malta - NDA on day one, full GDPR compliance. Trusted by startups and enterprise teams across 12+ industries.

View QA Engineers

Generate test cases from specs and user stories automatically

Run visual regression across hundreds of screens in minutes

Build CI test gates that catch regressions before merge

Analyze test results and prioritize fixes by business impact