Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Thu, 14 Apr 2016 09:23:25 +0000 (UTC)
From: "Benedict (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.12954281.1459256447000.228759.1460625805461@Atlassian.JIRA>
In-Reply-To: <JIRA.12954281.1459256447000@Atlassian.JIRA>
References: <JIRA.12954281.1459256447000@Atlassian.JIRA>
 <JIRA.12954281.1459256447492@arcas>
Subject: [jira] [Commented] (CASSANDRA-11452) Cache implementation using
 LIRS eviction for in-process page cache
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/CASSANDRA-11452?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15240856#comment-15240856 ] 

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

The simplest way would be to accept a hash function that at least operates over long values, which if not provided just returns the int, or possibly {{int | ((long)int << 32)}}.  It would be nice if we could operate over larger values, but I don't think that's possible without negatively impacting Caffeine's footprint.

We can also protect against the attack by simply comparing against a random member of the LRU for frequency in the LFU on admission, perhaps preferring (exponentially) candidates that are near to eviction, i.e. generating a number via an extreme value/exponential/zipf RNF and walking this distance from the head of the eviction queue.  This at least provides a mechanism that will rapidly prune collisions without affecting behaviour meaningfully.


> 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)