Application servers rely on caching and clustering to achieve high performance and scalability. While queries benefit from middle-tier caching, updates introduce a distributed cache consistency problem. The standard approaches to this problem, cache invalidation and cache replication, either do not guarantee full cache consistency or impose a performance penalty. This paper proposes a novel approach: Freshness-Aware Caching (FAC). FAC tracks the freshness of cached data and allows clients to explicitly trade freshness-of-data for response times. We have implemented FAC in an open-source application server and compare its performance to cache invalidation and cache replication. The evaluation shows that both cache invalidation and FAC provide better update scalability than cache replication. We also show that FAC can provide a significant better read performance than cache invalidation in the case of frequent updates.
展开▼
