Hi Guys,
Basically I have a support called raised with compellent support. We have a 1Gb link between two sites, and we replicate some volumes across the link.
We have replays that run at 6am and 6pm and from about 6am-8am and 6pm-8m the link gets utlized as you would expect, but after then it slows right down adn as a result replication is falling behind.
I have done a test using 2 VMs, one has an iscsi target and the other is an iscsi initiator, and when I transfer data onto that iscsi virtual disk I totally max out that link, exactly as you would expect.
My question is why doesn't the compellent do this? Why does it pretty much stop maxing out the link at 8 on the dot every time like clock work?
There is nothing in the QoS profiles, they have no limits set, and they are set for 1gb link
Supports final assessment was to say that there was nothing they could do. It was a lot of data going to tier 3 and replication was running.
I told them I couldn't accept that because:
7K disks would not be a reason why the SAN could not max out the 1Gb link
If the link was being maxed out all the time, and then replication was falling behind I could fully understand and that would be a networking issue. But the SAN doesn't max it out, and doesn't come close.
I have just sent them a screen shot of the link showing that between 6am-8am and 6pm-8pm the link maxes out, this coincides with when the Replays are created, so clearly the SAN has the capability to max it out, but seems to stop doing so at around 8am/pm?
I am at a loss, Replication MB/Sec averages at about 7.5 which is nothing.
This graph shows that after 8am the transfer drops off, and at around 16:45 I start my iscsi test and max it out again
and this one shows that at around 6am and 6pm it peaks again thanks to the replays?