قِس قبل أن تحسّن

في هذا الدرس

apply a repeatable performance loop.

الموضوعات المخطط لها

• Loop: measure, hypothesize, change one thing, re-measure.
• Intuition about hot spots is often wrong; the profile decides.
• Define the goal first (p95 latency, throughput) and measure under realistic load.
• Profile in production-like mode, not debug mode; run long enough to pass warm-up and repeat runs for variance.
• Keep an investigation log: what was measured, what changed, the result.