Veeam protection for unifed systems

[zak2011]

Is it possible for to use Veeam to protect filesystems if we plan to incorporate in the vnx 5300 unified system so that i dont need to use the snap functionality for block data at array level ?

[dellock6]

VNX unified storage is simply block and nas storage at the same time. What is your use case? Creating nas luns on the VNX and use them as file servers?

Luca.

[zak2011]

Yes..that is exactøy the case of creating nas luns on the VNX and use them as file servers.
As we trying to get hold of the VNX from EMC..the offer came with something called local protection suite which does the snaphots and cloning on the array level.
Isnt it possible for Veeam to take checkpoints of the filesystems or use the snap functionality for block data at array level rather than using the local protection suite of vnx?

Thanks!

[dellock6]

Veeam can backup only something managed by VADP libraries, even with SanSnap restore on HP Lefthand the files in the snapshots are regular vmdk files. So, I think the answer is no. If you want to use Veeam to save those file, you better create a virtual file server on VMware and save it as usual, and run only block storage on the VNX.
Also remember, storage snapshots ARE NOT backups, saved data are still inside the storage. You loose the storage, you loose both data and their copies...

Luca.

[zak2011]

So actually there isnt a real need to buy local protection suite on the VNX as i would rather virtualize the file servers and use Veeam to backup it up.
Thanks Luca.

[zak2011]

I was just trying to justfiy with EMC that i wouldnt actually depend on storage snapshots as backups and they do agree with it. But they have some claims on storage snapshots.
"Storage snapshots are in no way meant to replace the need to take regular backups, but they provide a shorter time to recover after a disaster and also less data lost.
Imagine that you take 1 backup every 24 hours (this is quite a standard). This means that in the worst case scenario you can lose 23 hours 59 minutes and 59,9999 seconds of production data.

Snapshots make it possible to shorten this time significantly. If you take a snapshot, say every 4 hours, then in an event of a failure (either complete system failure or component failure), you will lose maximum 3 hours 59 minutes and 59,9999 seconds of production data.
Considering the fact that snapshots on VNX aren’t exactly space consuming, you can take snapshots every 1 hour if you’d like – can you take backups just as fast as this?
Also note that snapshots are fully customizable down to which luns/volumes to snapshot, when to take snapshots, how often, for how long they should be saved, and after what time they should be automatically deleted.

All snapshots will survive a failure on the system (they are crash consistent) and besides – our systems never really go down."

If i use the near CDP option of Veeam, wont i be able to take backups much lesser than 1 hour?

Thanks!

[dellock6]

Well, for sure EMC has to sell its products, I cannot blame them for trying to say SAN snapshots are better. To me, they are simply different, so there is a use case for both:
- pro: faster snapshots than vsphere, no VM stun caused by snapshot commit on high I/O virtual machines
- cons: copies are saved inside the same storage, you have to restore the whole LUN instead of a single VM or even file and is not that easy

Funny they are saying snapshots do not consume space... where are the new bits supposed to be saved, in the air?

Ideally, you can use both if your budget allow for it, but remember san snapshots restore are not so easy as doing snapshots themselves. I do not know how many VMs you have and how big they are, or how much IO they produce, but you can start to separate them in different groups:
- critical: those you need so save as often as you can
- important: maybe two copies per day are enough
- normal: one backup every 24 hours
- application with no data: once per week, or even only when modified

Probably you will find out not all your VMs need to be saved so often, and you can apply Veeam near-CDP only to those. If among those VMs there are "heavyweight" VM producing high IO, and thus having problems with vSphere snapshots during working hours, probably you can also consider san snapshots.

Luca.

[zak2011]

thanks for that comment. If i am going to use Veeam's CDP , how often will the snapshots be taken...i dont find it convincing enough to buy storage snapshots funtionality with EMC when Veeam can do near CDP.

[tsightler]

Well, to be completely fair, near CDP functionality in Veeam can be quite intensive and I normally only suggest it for a handful of servers, and even then only if it is properly architected (good, fast disks for snapshot storage, plenty of IOP overhead). Remember that Veeam is completely dependent on Hypervisor level snapshots, and in the case of VMware, this can cause significant impact. If you have 100's VMs, and you attempt to run near CDP on them, you will be taking and removing 1000's of snapshots every hour, which is probably not realistic.

The reality is that the combination of snapshots and backups are ideal, it's exactly why we are introducing the integration of Veeam with SAN snapshots. SAN snapshots can provide low impact snapshots to use for restore points, while Veeam can still be leveraged for proper backup to secondary storage, while providing a unified user interface for both restore types. For now this feature is only available with HP Lefthand based storage (StoreVirtual/P4000).

So EMC is not incorrect, their solution has benefits, as well as ours. We consider these two solutions complimentary.

[zak2011]

Thanks Tom. It makes more sense and seems more practical now to have a combination of both.