Currently all replication tasks use only one NIC but You can create bonding and run replication on it.
Currently all replication tasks use only one NIC but You can create bonding and run replication on it.
Thank you for the Tip.
I have create a blance-rr bond for the Volume Replication with two Nics, but this does not increase the throughput, the write-io with active replication is is still at 110 MB.
During creation of volume replication task you can set option: " Bandwidth for SyncSource (MB)". By default it's set to 40 MB. To change this value you have to delete this task and create it again.
There is a knowledgebase article: Why Open-E request to specify bandwidth value for volume replication? http://kb.open-e.com/Why-Open-E-requ...ation_172.html
"Once the destination volume show consistent, replication will use maximum possible bandwidth."
So the Option "Bandwidth for SyncSource" should not impact the Bandwidth, when the Volume is consistent.
hmm, I didn't know that.
Ok, could you try one more thing? Shutdown secondary machine and see if speed will also increase to 220 MB.
When the secondary Machine ist shutdown, the speed with IOMeter also increase from 110 to 220 MB. Apparently only the Replication slow down the througput.
Ok, You can try change some tuning options. In console go to Hardware Configuration (ALT + CTRL + W) and choose "DRBD tuning".
Here you can find some info about available options http://www.drbd.org/users-guide/re-drbdconf.html
I learned today that balance-rr and replcation doesn't always work as expected. The perfect solution would be using 10Gb ethernet cards, but probably this is not an option?