CacheSifter: Sifting Cache Files for Boosted Mobile Performance and Lifetime
summary
CacheSifter is a framework for Android that classifies app cache files online into three categories based on reuse patterns and whether they reside in memory or storage, then selectively drops cache files unlikely to be reused to avoid flash writebacks. Implemented on real devices and evaluated with representative mobile apps and ML models, it reduces cache writebacks (around 59–62% depending on model) and improves I/O-intensive write performance (about 18–25%).
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:25Executive Summary
- 1:55Extensive Data Are Written into Flash Every Day!
- 2:56Lots of Cache File Writes
- 4:27Source of Massive Cached Data Android performance optimization
- 5:22Observation: Cache Files Are Not All The Same
- 6:09Differentiating File Types
- 7:02CacheSifter's Goal
- 7:43Framework of CacheSifter
- 9:44Experimental Setup Real smartphones
- 10:22Evaluation Results: Cache Files' Writes Reduction in cache file writes and total writes to flash storage
- 13:08Evaluation Results: Overhead Overhead of CacheSifter.
related talks
29m10 Ways To Improve Your Android App Performance
38mOptimize your app performance for the next generation of Android
35mThe Impact of Micro-Optimizations on Large-Scale Apps
18mSIEVE is Simpler than LRU: an Efficient Turn-Key Eviction Algorithm for Web Caches
38mLow effort, High impact Android performance wins- 38mOptimize your app performance for the next generation of Android
43mWhere Is My Cache? Architectural Patterns for Caching ...
42mCaching for Cash