[
https://issues.apache.org/jira/browse/CASSANDRA-11452?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15238051#comment-15238051
]
Benedict commented on CASSANDRA-11452:
--------------------------------------
bq. retain hash collisions
For this to be a problem it would have to be a collision on all hashes/buckets,
which is probabilistically unlikely - with a large enough cache and uniform
hashes we can even calculate an expected incidence. Given the LRU arena will
evict it in a predictable time we can also calculate the fixed overhead for
such mispredictions; it is unlikely to be very large (but would be interesting
- perhaps something the paper could have covered in its analysis of the
overhead per item).
Bear in mind also that - as opposed to LIRS which is an _eviction_ strategy -
LFU only prevents promotion to the LRU; the eviction strategy is still LRU, so
it will not keep collisions, only fail to filter them before they hit the
(main) cache.
> 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)
