[DRBD-user] Write performance tuning of protocol C

Sean McCreadie smccreadie at CanyonPartners.com
Tue Jan 4 01:46:43 CET 2011

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Hello,

I have setup two Vmware virtual machines to replicate their storage with DRBD using protocol C.  So far the write performance isn't as good as I had hoped.

I have tested write performance using fio and right now the best I can get is write throughput and IOPS that are about 50-75% as good as with drbd replication disabled.

My setup is two identical Centos 5.5 VMs running drbd 8.3.8.  The link between them is two dedicated 10Gb adapters, no switch in between.  I feel the link is not the bottleneck but something in my drbd config.

I have followed the tuning recommendations on the linbit site and have seen some small improvements, but still don't have the performance I was hoping for.  I thought I would be able to get a write penalty more like 10%, not the 25-50% I am seeing.

I added no-disk-flushes; and no-md-flushes;.
I set max buffers and max-epoch-size to 8196, and also experimented with several unplug-watermark sizes.
I also set my hardware RAID controller to 100% write cache.

I attached my drbd.conf.  Any other ideas would be most appreciated.  Thank you

Sean

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20110103/3f83ccef/attachment.htm>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: drbd.conf.txt
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20110103/3f83ccef/attachment.txt>


More information about the drbd-user mailing list