Scalability issues in RHQ 4.9 - A summary

Elias Ross genman at noderunner.net
Sun Mar 9 17:12:03 UTC 2014


On Fri, Mar 7, 2014 at 6:45 PM, John Sanda <jsanda at redhat.com> wrote:
> Events, traits, baselines, and OOBs are on the list of things to migrate to Cassandra. Are you still seeing it take over an hour with the performance changes in master that will go into 4.10? The schema changes targeted for 4.11 should have a big impact as they will dramatically reduce the number of reads performed during aggregation.

At around 1,800,000 schedules, aggregation (plus purge) couldn't
reliably complete in under an hour for a four node cluster with six
spinning disks each. I was seeing I/O wait peak at 100%. I cut about
half the number of schedules and basically things complete on time
now.


More information about the rhq-devel mailing list