Snapshotting at the hardware level - EMC VNX

[evil3]

Hi Guys

At the moment when a backup fires off, vCenter snapshots at the VM level, but this periodically causes the VM to stutter for a couple of seconds (i.e. longer pings, occasionally lost packet, which is not good for a webserver ). I've been advised that letting the vCenter snapshot at the LUN/Volume level would help to avoid this as its quicker.

The problem is we use 2 different types of array, an EqualLogic, configuring this is to do the snapshotting is easy enough and well documented (By using the Host Integration Tools). But what I can't find is anything on how to do this with an EMC VNX5300. I've contact EMC and to be honest, their reply was useless, they suggest using SNAPCLI, which I don't think is quite what we are looking for.

Does anybody have an EMC VNX or Celerra and use hardware level snapshots, if so how did you get it working?

Thanks

[veremin]

Hi, Kevin,

VMware community might be a better place to ask this sort of questions, as this is not connected with Veeam products anyhow.

Thanks.

[kjstech]

I have an NX4 Celerra and also a VNX5200 but the VNX is not in service yet. We are doing some power distribution work before we can bring the VNX online safely with our UPS.

I am curious if this works with NFS as well and how you would go about configuring this. We see this stutter as well and its the main reason why I have backup windows set in Veeam. All the backups must complete before the start of the next workday or else theres performance issues when the snapshots commit. I even have one machine with a 58gig snapshot file because Veeam never committed it, so now I have to schedule off hours time to consolidate it.

Our NX4 is using dual 1gig connections and the VNX will be dual 10gig connections off each data mover. So I'm curious if the bandwidth also comes into play here with vm's going offline for a few seconds during the last part of the snapshot commit.

If you get any answers I would love to hear!

[Vitaliy S.]

vCenter snapshots at the VM level, but this periodically causes the VM to stutter for a couple of seconds (i.e. longer pings, occasionally lost packet, which is not good for a webserver

We've heard about similar behavior for Exchange server VMs, and allocating either more resources to this sever or migrating to a less loaded datastore usually helped in these cases.