[jbosscache-dev] new JDBC cache loader impl ready / final figures and design
Manik Surtani
manik at jboss.org
Sat Feb 3 07:07:11 EST 2007
Just a clarification, I presume "old imp" == "JDBCCacheLoader" and
"new impl" == "RedcdRecJDBCCacheLoader", as per your class diagram?
Interesting to note that even in the case of a shallow tree (say, not
more than 3 nodes deep) there still is a considerable improvement in
the put() operation at least, even though remove() and loadState()
don't produce that great an improvement.
--
Manik Surtani
Lead, JBoss Cache
JBoss, a division of Red Hat
Email: manik at jboss.org
Telephone: +44 7786 702 706
MSN: manik at surtani.org
Yahoo/AIM/Skype: maniksurtani
On 2 Feb 2007, at 03:12, Mircea Markus wrote:
> Hi,
>
> Attached is a comparison between old and new impl on JDBC cache
> loading. Basically recursion was eliminated in the case of
> loadState and remove, and the put had a 50% optimization. All other
> operation are more or less a single db call, so nothing can really
> be programmatically improved.
>
> The second image contains the design. Any comments are extremely
> welcomed :)
> Nothing is committed yet on CVS.
>
> An open issue is the backward compatibility - working on that.
>
> Cheers,
> Mircea
>
> P.S.
> How I've tested the performance (phps more easy to understand it
> from code :-)
> a tree with a TREE_DEPTH depth is created. Each node in the tree
> has a CHILDREN number of children. By default I've used
> TREE_DEPTH=7 and CHILDREN=3 which gave a total number of 3280
> children. A node is selected at each depth from disjunct
> subtrees. ( i.e. in the benchmark used 7 nodes are selected). On
> all those nodes a certain operation is performed: loadState, remove
> etc. For benchmarking the put operation, the in memory tree nodes
> are Collections.shuffle(List<Fqn>), then they are (randomly) added
> to the database/classLoader.put. This procedure was repeated sever
> times. Also additional testing was performed in the case of put for
> deep trees (100 - 200 nodes)
>
>
>
> <performance_comparison.PNG>
> <design.png>
> _______________________________________________
> jbosscache-dev mailing list
> jbosscache-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/jbosscache-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/jbosscache-dev/attachments/20070203/081a65b5/attachment.html
More information about the jbosscache-dev
mailing list