Hyper-V Cluster replication

[slamjam99]

Hi guys

I need some help and advice.

We run a Hyper-v Cluster using 6 x Windows 2008 R2 NODEs & 12 x 2TB CSVs (2 per node).

All of the 2TB CSVs are stored on the SAN storage and accessed over a 1GB redundant network. We run a separate LAN for our backup traffic and would use this for the replication traffic.

Currently we are using the Software VSS to backup the system etc and we are running about 120 VMs, 20 per NODE. The current backup using DPM 2010 takes about 14 hours.

We plan to install a 1GB link between our 2 data centres for the replication traffic.

What sort of performance can we expect from Veeam?

We would like to be able to replication each VM to the other datacenter within a hour and so if we switched to our 2nd data centre, we would only lose 1 hours work.

Can this be achieved using Veeam? How long will it take to perform incremental replication updates on a 250GB VM?

Thanks

[foggy]

Hello! It will depend on the disk activity of this VM actually (the amount of changes it will produce during this period). Also, source and target storage performance and compression/dedupe rates do come in as well as some environmental factors. Hard to predict not knowing all the details. Thanks.

[slamjam99]

Are we able to setup a test to answer the questions?

We are very keen if this would work as described?

[foggy]

Sure, you can test everything you want - this is what the trial version is all about.
Consider using hardware VSS provider to minimize the time required for replication. And take a chance to review the product user guide, available on the product page under Resources tab.

[slamjam99]

We can't use the hardware vss at this time due to a software issue within the firmware.

Will this make much difference ?

On average do you know how long it would take to incrementally backup a 250GB VM which has only 10GB data change on good SAN storage?

[foggy]

With software provider, the CSV volume acts in redirected I/O mode for the longer period of time increasing the overall backup time (especially for VMs with large disks). You can get more details from the user guide.

Regarding the backup time, you can try to upload the expected amount of changes to the DR and see what you get. This could be taken as a rough estimate of the required time.

[slamjam99]

OK, we have installed the trial and I must say the replication seems to work very well apart from this issue:

We are using Software VSS and so only seems to be able to replication 1 VM at a time per cluster, is this correct?

Using DPM and software VSS we could backup 1 VM per clustered NODE at a time as each NODE has its own CSVs and they do not overlap with other node (i.e. backup 6 compared to 1 at a time).

[foggy]

VBR6 acts the same way. In case when each node stores its VMs on a separate volume, you can backup several VMs at a time. However, to achieve that, you need to set up several jobs - one per VM, as within a job, VMs are processed sequentially.