Table of Contents
Table of Contents
Further Reading
No items found.
No items found.
No items found.
No items found.
No items found.
No items found.
What performance testing evaluates
Performance testing confirms that applications deliver reliable experiences at any scale. It covers application code, databases, networks, servers, and infrastructure.
Key metrics
- Latency: how long requests take to complete
- Throughput: requests handled per second
- Error rate: share of failed requests under load
- Resource use: CPU, memory, and network consumption
Types of performance testing
Each type stresses the system in a different way.
How to run a performance test
- Set performance goals, such as p95 latency under 300 ms.
- Model realistic user scenarios and traffic levels.
- Run tests in an environment that mirrors production.
- Collect latency, throughput, errors, and resource metrics.
- Compare results with goals and earlier baselines.
Why performance testing matters
Without performance testing, systems risk failing in production. Teams pay for it in user trust and cloud bills.
- User experience: slow responses drive users away
- Bottlenecks: slow code, queries, or infrastructure go unnoticed
- Scaling costs: inefficient scaling raises spend
- Regressions: new features quietly slow things down
For Kubernetes and CI/CD, performance testing lets teams ship quickly without sacrificing reliability. Consider a SaaS provider that runs k6 tests on every pipeline run. A new query that doubles API latency fails the build before release.
Best practices
- Test continuously: run performance checks in CI/CD, not only before launch
- Use thresholds: fail builds when key metrics cross limits
- Keep baselines: compare every run with past results
- Watch the system: track infrastructure metrics during tests
- Test realistic paths: include login, search, and checkout flows
Real-world examples
- API latency gates: a SaaS provider runs k6 tests in Testkube on every CI/CD cycle
- Mobile banking: a bank tests transaction times, login speed, and balance lookups across network conditions
How performance testing relates to Testkube
Testkube supports performance testing workflows inside Kubernetes. Load generators run close to the services they test.
- Native tools: k6, JMeter, and Artillery run inside clusters
- Pipeline automation: performance checks run in CI/CD
- Stored results: artifacts kept for trend analysis and compliance
- Observability: dashboards for latency, throughput, and errors