[ https://issues.apache.org/jira/browse/CASSANDRA-11452?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15240834#comment-15240834 ]
Benedict commented on CASSANDRA-11452: -------------------------------------- Nice catch, that's a really pernicious property. The paper not only doesn't mention it, it seems to me that by comparing against the eviction victim the TinyLFU is no longer an _admission_ policy as stated, but both an admission _and_ an eviction policy - debatably in opposition to the paper. It seems that we could instead be comparing against a cohort of near-to-eviction candidates, or some other dynamic threshold. The eviction candidate isn't particularly special, as far as I can tell, it's simply a proxy for "this is the threshold above which a value is likely to be reused" It would also be great to support more than 32-bit hashes for seeding the sketch hashes, to reduce the incidence of this. > Cache implementation using LIRS eviction for in-process page cache > ------------------------------------------------------------------ > > Key: CASSANDRA-11452 > URL: https://issues.apache.org/jira/browse/CASSANDRA-11452 > Project: Cassandra > Issue Type: Improvement > Components: Local Write-Read Paths > Reporter: Branimir Lambov > Assignee: Branimir Lambov > > Following up from CASSANDRA-5863, to make best use of caching and to avoid > having to explicitly marking compaction accesses as non-cacheable, we need a > cache implementation that uses an eviction algorithm that can better handle > non-recurring accesses. -- This message was sent by Atlassian JIRA (v6.3.4#6332)