Slow Lefthand Remote Snapshot

[Dave_M]

We have a P4300 running SAN iQ 8.1 doing a remote snapshot to a DR P4300G2 SAN iQ 8.5 the remote bandwidth is set to 1Gb however the remote snapshot only runs at 15Mb/sec.

We also have a second P4300G2 SAN iQ 8.5 doing a remote snapshot out to the sameDR  P4300G2 as above, this is set as 1Gb remote bandwidth but this barely goes above 15Mb/sec sometimes hitting 20Mb/sec

All the Lefthands are connected directled to 2 x Cisco 3020 switches in teh back of our c7000 blade centre only the lefthands are on these switches no connection to our main network.

If we move a VM from one SAN to the DR P4300G2 via the VMWare software we can transfer a 50Gb VM in approx 1 hour, so we know the speed is not a problem.

Any one have any suggestions on what we can try?

Thanks

Dave

[oikjn]

What is your actual bandwidth at the local and remote sites for your WAN? Are they both 1Gb links? If not, then you will never get 1Gb speeds even if you sent the CMC to limit your bandwidth to 1Gb. If I had to guess, you have a 15Mb upload speed at the local site and that is your limiting factor.

BTW, 50Gb in one hour is 13.8Mb/sec so everything is consistent with a network speed limitation somewhere of 15Mb and I would guess its the upload speed limit or maybe your remote download speed.

[Dave_M]

Thanks for your reply.

Currently the our DR Lefhand is in the same datacentre as our other lefthands, I ahve attached an image to illustrate how our current system is setup.

There is nothing in teh Cisco configs to limit port speed.  The Lefthands are on their own subnet the Cisco 3020 only deal with the Lefthand traffic they are not connected to our primary network.

Thanks

Dave

[oikjn]

as long as there are no problems with things such as flow control or jumbo frames on your switches, I would check the switches to make sure the error rates on the ports are all on or close to zero and then I would monitor the performance of that DR VSA.  It sounds like your network morst likely isn't your bottleneck and its one of the VSAs.  Watch the CPU usage along with IO+Latency/throughput on each VSA paying close attention to the remote one (as that is most likely the bottleneck.

The speed numbers yoiu said are consistent, so I dont' think the problem is the remote snapshot function as much as your effective throughput on a node. 

As a reference, I have a "remote"site also on my local switches that consist of a single VSA running in hyper-v on a Xeon E5630 w/ 2vCPU and using two 2TB "green" SATA Hdds in raid 1 as storage.  I limit my bandwidth to 100Mb and the remote snapshots hit that rate.

I would suggest creating a temporary LUN target and doing a benchmark on the remote VSA to see what that it can actually do.

[Dave_M]

Managed to resolve this, by increasing the remote bandwidth slider

[shaketheclouds]

Suffering a similar kinda speed issue. Could you elaborate more on the fix?

Which bandwidth slider bar?