Printable View
Has anyone managed to get a good replication performance in an active active setup. We have 2x10GB ports in a RR setup and the maximum we get for replication traffic is about 300MBs which is less than 50% of the write speed of the VG when replication is disable.
You probably mean resync speed, or initial sync speed. Speed is adaptive and managed by internal mechanism of volume replication not to interrupt other services. It is possible to control some of volume replication parameters in console -> Ctrl+alt+W -> Drbd tuning, but usually its best to leave default ones.
I am talking about the normal replication traffic in an active-avtive cluster when all is working correctly. I am using iometer on a windows machine to test the performance of the iscsi volumes, when running in degraded mode I am getting 1939 MBs writes with iometer, when running in normal mode it drops down to 500MBs
Have you verified that any of the things mentioned here is not the factor?
http://kb.open-e.com/Data-access-is-...task_2483.html