cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Benedict (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-11452) Cache implementation using LIRS eviction for in-process page cache
Date Thu, 14 Apr 2016 09:39:25 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-11452?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15240867#comment-15240867
] 

Benedict commented on CASSANDRA-11452:
--------------------------------------

Well, a RNF like I proposed would have a very short average walk length.  The nice thing about
that is it protects against any number of collisions (or we can define a bound - 20 would
probably be plenty, since you may not want to walk arbitrary distances, although a more general
solution would be a skip list of two levels that permits travelling further along the list
more rapidly) without impacting constant factors and having barely perceptible impact on outliers
(definitely not perceptible given other causes of variance in Cassandra) - just offering slower
protection the more collisions there are (which is acceptable, given they're also considerably
less likely)

> 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)

Mime
View raw message