When is the Cache Warm? Manufacturing a Rule of Thumb
summary
Derives a simple rule-of-thumb formula that estimates cache warm-up time using only cache size and a tolerance parameter, validated by simulation on CDN and storage traces; uses a new interval hit ratio metric to capture dynamics, fits linear/log-linear regression models with R^2 > 0.8 for accuracy, and shows warm-up can complete up to 40% sooner than full-fill time. Provides an offline estimator workflow, discusses limits (empty vs partially populated restarts) and plans to open-source the simulator and package.
AI-written summary · about
chapters (10)
- 0:00Intro
- 0:23Distributed Caches are Dynamic
- 1:55Understanding Cache Warmup
- 3:33Cache Dynamics
- 8:55Defining Warmup Time
- 9:42Solution: Rule of Thumb
- 10:28Deriving a Rule of Thumb
- 13:18Applying the Rule of Thumb
- 14:11Discussion
- 15:23Conclusion
related talks
42mCaching for Cash
05mCache Warming: Warm Up The Code
27mCaching for Cash
38mLazy Defenses: Using Scaled TTLs to Keep Your Cache Correct
34mOptimizations in Multiple Dimensions
45mCaching in Applications Still Matters
31mCaches All the Way Down!
30mBuilding Faster, More Resilient Apps with Service Worker: A Caching Strategy Deep Dive