performance - cassandra massive write perfomance problem -
i have server 4 gb ram , 2x 4 cores cpu. when start perform massive writes in cassandra works fine initially, after couple hours 10k inserts per second database grows 25+ gb, , performance go down 500 insert per seconds!
i find out because compacting operations slow don't understand why? set 8 concurrent compacting threads cassandra don't use 8 threads; 2 cores loaded.
appreciate help.
we've seen similar problems cassandra out-the-box, see:
http://www.acunu.com/blogs/richard-low/cassandra-under-heavy-write-load-part-ii/
one solution these sort of performance degradation issues (but no means only) consider different storage engine, castle, used in above blog post - opensource (gpl v2), has better performance , degrades more gracefully. code here (i've pushed branch cassandra 0.8 support):
https://bitbucket.org/acunu/fs.hg
and instructions on how started here:
http://support.acunu.com/entries/20216797-castle-build-instructions
(full disclosure: work acunu, may little biased ;-)
Comments
Post a Comment