cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chris Goffinet (JIRA)" <>
Subject [jira] Commented: (CASSANDRA-337) Make BinaryMemtable work
Date Thu, 27 Aug 2009 21:24:59 GMT


Chris Goffinet commented on CASSANDRA-337:

I can live with merging the executors for BMT flush and normal MT. Though I want the ability
to specify a min and max for the threads in storage-conf.xml. During an import, we would want
the ability to have a certain quality of service (I want a min of 4 threads during normal
usage, but when we start binary import, I want a higher thread count). 

Merging both executors makes it a little harder, but the benefit is less code/cruft to manage
-- that's a win for sure. When we were benchmarking our 7 node cluster with 600-800Mb/s per
node we saw lower memory usage/better throughput on the flushing by having 2 threads per disk.

If this is acceptable (above) I can submit a patch.

> Make BinaryMemtable work
> ------------------------
>                 Key: CASSANDRA-337
>                 URL:
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Jonathan Ellis
>            Assignee: Chris Goffinet
>             Fix For: 0.4
>         Attachments: 0001-Working-version-of-BMT.patch, 0002-Added-flushAndshutdown-v2.patch,
0003-Removed-decorateKey-in-writer.append-in-BMT.patch, 337-rebased.patch, 337.patch

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message