[DRBD-user] Unusual Performance. Faster one way.

Philipp Marek philipp.marek at linbit.com
Mon Apr 27 08:30:31 CEST 2015

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

> I'm doing some testing with Ganeti, a cluster-based VM manager, which
> uses DRBD as the default syncing system between nodes. My test cluster
> has two nodes, Alpha & Beta. Ganeti provides a way to migrate from one
> to another, handling DRBD automatically to sync to the node that isn't
> running the VM.
> I'm running into an issue that I'm having difficultly understanding.
> When benchmarking from within the VM, I'm getting drastically
> different results depending on the node that the VM is running from.
> The DRBD benchmarks are below;
> For comparison, I've done benchmarks with DRBD turned off. These are below;
> Alpha;
> Sequential Write: 206.250 MB/s
> Beta;
> Sequential Write: 207.551 MB/

As the direct write speeds look similar, I'd guess it's network related.

Try turning the various optimizations off (TCP checksum offloading etc.) 
- if one of the machines generates bad CRCs every so many packets, it will 
mostly limit the data stream in the Primary => Secondary direction
as that one runs much more packets.

More information about the drbd-user mailing list