The optimizer may now choose a new caching strategy called “fetch and discard” strategy. Standard caching strategy reads pages in at the head of the MRU/LRU (most recently used/least recently used) chain in cache. This flushes other pages out of cache. Fetch-and- discard (or MRU) strategy reads pages into cache much closer to the end of the cache, so it does not flush other pages. But these pages remain in cache a much shorter time, so subsequent user queries that might get a “cache hit” by finding these pages in cache under the standard strategy will be less likely to find the page in cache.
When the optimizer estimates that a particular query will read a significant number of pages, and only need those pages once during a query, it may choose fetch and discard strategy. If your disk I/O increases in release 11.0, check the I/O strategy for queries using showplan to see the cache strategy in use. If the queries are using fetch- and-discard (MRU) strategy, you can add a clause to select, update, and delete commands to specify standard caching.
See Chapter 10, “Advanced Optimizing Techniques” in the Performance and Tuning Guide.
