What is Performance Testing?
Software performance testing, told as a story about a publishing house sending in pretend readers before a book even launches.
On release day, readers all show up at once.The print shop, usually fine, is slow today of all days.
Without a practice rush ahead of time, nobody knows where the print shop's limit is.Not knowing how big a rush stops the printing, everyone fears printing will back up again on release day.
Performance testing sends in pretend readers ahead of time to see how much the print shop can take.Performance testing finds the print shop's limit before real readers ever show up.
Performance testing measures four numbers: response time, throughput, concurrent readers, and error rate.These numbers say how big a rush it takes to stop the print shop.
The test setup and the real world can differ.If the test setup differs from the real one, passing doesn't guarantee it holds up for real.
Software performance testing sends in a pretend rush before real users ever arrive, measuring numbers like response time and throughput to find the breaking point ahead of time.
Performance testing checks whether a system meets metrics like response time, throughput (TPS), concurrent users, and error rate under a given load. It splits into load testing (expected levels, typical to peak), stress testing (at or beyond the limit), endurance/soak testing (sustained over time), and spike testing (sudden bursts).
When grown-ups say it
- Performance Testing
- Rushing in pretend readers ahead of time. The goal is finding the limit early.
- Load Test
- Rushing in at the expected level. Checks it holds up from typical to expected peak.
- Stress Test
- Pushing to the limit or past it. Finds exactly where it breaks.
- Response Time
- How long it takes to get one book. What readers actually feel as speed.
- Throughput (TPS)
- How many books printed per second. Throughput measures speed; function points measure functional size. → the ruler that measures functions, not pages
- Bottleneck
- The narrow door that jams up first. Testing finds it before it's too late.
- Hypothesis Testing
- How to tell which of two covers truly sells better. Used to separate real differences from coincidence. → telling a real difference from coincidence
- High Availability
- Keeping a backup print shop ready. A different story — about never stopping, not about speed (another field). → when the spare generator takes over