This article explains Caffeine's high‑performance local cache features, presents JMH benchmark results that show it beating Guava, ConcurrentMap and ehcache, details its expiration, refresh and eviction mechanisms, compares memory usage and eviction algorithms, and provides migration code for Guava users.
This article traces iQIYI's five‑stage journey of Java caching—from simple database queries and in‑process HashMap caches, through Guava's advanced features addressing lock contention, expiration, and refresh, to the high‑performance Caffeine library with its W‑TinyLFU algorithm—while providing code examples and architectural insights.
This article traces iQIYI’s five‑stage evolution of Java caching—from simple database lookups and Redis synchronization to Guava Cache enhancements and finally to Caffeine’s W‑TinyLFU implementation—explaining each approach, its advantages, drawbacks, and the underlying algorithms that drive modern cache performance.
This article traces iQIYI’s five‑stage Java caching journey—from early database lookups, through Redis synchronization, in‑process HashMap and Guava caches, to modern Caffeine with W‑TinyLFU—explaining each phase’s advantages, drawbacks, eviction algorithms, and implementation details such as segment locking, expiration queues, and multi‑queue data structures.