[
https://issues.apache.org/jira/browse/SOLR-3393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13261203#comment-13261203
]
Shawn Heisey commented on SOLR-3393:
------------------------------------
Hoss, thanks for your comments.
Since I was the one who wrote the previous version of this, I just opted to
completely replace LFUCache rather than go with a "Fast" appendage. I hadn't
considered the naming problem in quite the same light as "yet another fast*"
name, but it did seem like a bad idea.
Yonik had the same concern about stats being preserved forever on SOLR-2906,
and he helped with a decay option to deal with that. I think the decay is a
good idea. There was only one kind of decay before, applied to all elements
anytime there were evictions, defaulted to on.
In a new version of the patch for this issue (which I have not yet uploaded) I
have now included two kinds of decay. There is the kind applied at eviction,
now defaulting to off, and one applied at warming, defaulting to on. I will
expand the documentation on the Wiki, making it clear that turning off the
decay option will probably lead to an undesirable cache state. Currently the
decay is implemented with a bit shift (>>> 1), I may make another option
available that just subtracts one, and we can bikeshed about which option
should be default.
> Implement an optimized LFUCache
> -------------------------------
>
> Key: SOLR-3393
> URL: https://issues.apache.org/jira/browse/SOLR-3393
> Project: Solr
> Issue Type: Improvement
> Components: search
> Affects Versions: 3.6, 4.0
> Reporter: Shawn Heisey
> Priority: Minor
> Fix For: 4.0
>
> Attachments: SOLR-3393.patch, SOLR-3393.patch
>
>
> SOLR-2906 gave us an inefficient LFU cache modeled on
> FastLRUCache/ConcurrentLRUCache. It could use some serious improvement. The
> following project includes an Apache 2.0 licensed O(1) implementation. The
> second link is the paper (PDF warning) it was based on:
> https://github.com/chirino/hawtdb
> http://dhruvbird.com/lfu.pdf
> Using this project and paper, I will attempt to make a new O(1) cache called
> FastLFUCache that is modeled on LRUCache.java. This will (for now) leave the
> existing LFUCache/ConcurrentLFUCache implementation in place.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
