Differences between current version and predecessor to the previous major change of LFU.

Other diffs: Previous Revision, Previous Author, or view the Annotated Edit History

Newer page: version 2 Last edited on Wednesday, October 8, 2003 5:20:44 pm by StuartYeates
Older page: version 1 Last edited on Wednesday, October 8, 2003 9:13:01 am by AristotlePagaltzis Revert
@@ -2,4 +2,6 @@
 This is a [Cache] entry expiry strategy. On cache miss, the least frequently used record is discarded from cache to be replaced by the data read for the request which caused the cache miss. 
 While this approach leads to very efficient utilization of the [Cache]'s capacity, it requires complex bookkeeping. The overhead incurred is rarely worth the effort and only pays off if a cache miss is many orders of magnitude more expensive than a cache hit. Even HardDisk [Cache]s, where the disparity between hit and miss is a factor of about 1,000, [LFU] may not achieve significantly better performance than [LRU]. 
+One situation where [LFU] may be more efficient is where there are two classes of user, one (such as a WebCrawler) accesses every storage unit in the underlying data store in turn and one (such as real people) accesses the underlying data store with high locality of refererence. In this situation the real people will not see any distrubance in the cache caused by the WebCrawler if [LFU] is used, but the cache will be cleared by [LRU].