[
https://issues.apache.org/jira/browse/SOLR-667?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yonik Seeley updated SOLR-667:
------------------------------
Attachment: SOLR-667-alternate.patch
OK, here's the implementation based on my previous pseudo code, along with a
very quick performance test.
The test uses random keys over a slightly bigger range than the table.
It also uses an upper water mark 10% higher than the lower water mark, and an
acceptable water mark half way inbetween. I haven't experimented to see what
the best acceptable water mark is for either impl. If anyone wants to do a
more realistic test with real queries, they are welcome to it.
I did 4 runs and took the lowest number for each sub-test. Java6 -server,
WinXP, P4. Times in milliseconds.
{code}
doPerfTest(2000000, 100000, 200000); // big cache
noble=17063 yonik=9157
doPerfTest(2000000, 100000, 120000); // smaller key space increases
distance between
oldest, newest and makes the first passes less effective.
noble=8391 yonik=5812
doPerfTest(6000000, 1000, 2000); // small cache, smaller hit rate
noble=17578 yonik=12515
doPerfTest(6000000, 1000, 1200); // small cache, bigger hit rate
noble=11500 yonik=8219
{code}
> Alternate LRUCache implementation
> ---------------------------------
>
> Key: SOLR-667
> URL: https://issues.apache.org/jira/browse/SOLR-667
> Project: Solr
> Issue Type: New Feature
> Components: search
> Affects Versions: 1.3
> Reporter: Noble Paul
> Assignee: Yonik Seeley
> Fix For: 1.4
>
> Attachments: ConcurrentLRUCache.java, ConcurrentLRUCache.java,
> ConcurrentLRUCache.java, SOLR-667-alternate.patch, SOLR-667-updates.patch,
> SOLR-667.patch, SOLR-667.patch, SOLR-667.patch, SOLR-667.patch,
> SOLR-667.patch, SOLR-667.patch, SOLR-667.patch, SOLR-667.patch, SOLR-667.patch
>
>
> The only available SolrCache i.e LRUCache is based on _LinkedHashMap_ which
> has _get()_ also synchronized. This can cause severe bottlenecks for faceted
> search. Any alternate implementation which can be faster/better must be
> considered.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
