Looking tio get an idea on what others are getting for bandwidth utilization for replication compared to what I'm seeing in my environment. All my hosts have 10 GbE and the vmnics are dedicated for replication traffic. However, I am consistently pushing around 160 Mbps from the source host, per replication job.
Is anyone else getting better than that? If so, did you find that there were any gotchas in the config of replication, hosts or even in the physical hosts?
Thanks in advance.
Thanks! I had seen that one before but hadn't been able to test until now. I doubled both values as suggested and although my Read Rate seemed to jump significantly the it did not change the bandwidth utilized for replication.
The stats are in KBps but I converted them for my own sanity. Here's a graph. I had 2 replications going at the same time. One had 2 disks and the other was a single disk VM. And to really make me look like a liar, this time it jumped for a short while to just under 400Mbps. Storage is Hitachi but not sure of the model number.
So just to add here, I have been doing some reading on improving NBD backup performance for another set of machines and learned that there is a 32MB combined buffer limit for NFC connections/traffic streams. Because NFC is used to transfer from Replication appliance to host, could this be the wall I am hitting for the speed limit?
If so, why would I be getting above that 32MB?
Any help would be appreciated.