Latency Principles

Based on "Latency: Reduce delay in software systems" by Pekka Enberg (Manning, 2026).

This skill provides a systematic framework for minimizing delay in software systems, covering the entire stack from Physics and Hardware to Application Architecture and User Experience.

Foundational Concepts

• What is Latency?: The time delay between a cause and its observed effect.
• Latency vs. Bandwidth: You can always add more bandwidth (more links), but you are stuck with bad latency unless you optimize the path.
• Impact on User Experience (UX): < 100ms (immediate), < 1s (instant), > 10s (slow).
• Measuring Correctly: Use percentiles (p95, p99) to capture tail latency. Avoid averages. Avoid Coordinated Omission by using fixed-interval benchmarking.

Modeling Performance

Use these laws to establish theoretical bounds and size systems: