What size is the largest vm you backup ?

[thuizenga]

We pulled our large file server out of veeam a couple of years (about 3tb). It didn't make much sense to us because we are still using traditional file agents to back it up and replicating that data offsite. The whole snapshot process on such a large and active server created it's own issues. Also fighting against all the changed blocks that shadow copy services created was to much of a hassle. The kicker for us was that our fileserver was created with GPT partitioning which is not supported (or at least wasn't at the time) by veeam.

Instead we built an additional fileserver in our dr environment and configured DFS replication. We already had all our shares in the DFS so there wasn't a major migration issue. It took some time to create all the file shares and duplicate all the permissions but it has been very reliable and it replicates 24x7. IT can immediately access files in case of a disaster and in order to activate the shares for the rest of the company all we need to do as change the referral status on the dr links.

[slam]

I have used DFS before but I am not sure about using it for such large file server. Our file server full backup is 8.4TB and daily incremental is 60GB. Last time I tried DFS, it had problem keeping up with much smaller data size.

File servers of 8TB may be unimaginable a few years ago. But not being able to support it in present time is definitely an oversight.

[dellock6]

Also, DFS can be a solution to guarantee availability, but what about backups of the files it holds? If a file is corrupted replication between DFS nodes will replicate the corrupted files, and there is no protection from user deleting files except for using also Shadow Copies (that adds extra space costs).
A backup strategy is required even when using redundant services like DFS (ora AD, or DNS...)

[beovax]

What is the best practise for backing up a VM with 10TB of data?

a) Volume spanned across VMDKs using GPT
b) vRDM is there a 2TB limit on these as read in the forums?
c)Does File level recovery work with VM of this size?
d) Any restore issues with a VM of this size as in forums read max VM size someone has tested is 7TB.

Can anyone confirm what the maximum tested size backed up and how this was acheived.

[Cokovic]

Biggest VM i'm backing up has provisioned space of 13.5TB with a total of 29 VMDKs. FLR works without any issues (only thing i ran into is a limitation in the number of VMDKs that were mounted by FLR. The limit is 20 VMDKs but can be changed with a registry key). Using B&R v.6.1.

[stuartmacgreen]

Out of curiosity...
Can you tell us any information about this 13.5TB VM?
OS, application running on this - what is the VM used for.
What method of backup using in Veeam.

Thanks.

[Cokovic]

It's a member of an Exchange 2010 DAG cluster holding passive copies of all databases within this DAG. OS is Windows 2008 R2 Enterprise. Backed up in Reverse incremental mode.

[jimkeops]

hi all,

i have a big vm that is a file server, we migrated all the data and now we are looking to back it up, do you guys have any of you experience backing up and VM like that?
any tips? considerations?

thanks!!

[dellock6]

H Jim, nothing special about that size, Veeam can do it, I've seen as big as 12 Tb VM and I read around these forums of 16 and more.
For the design part, remember it's a single VM so it will be processed sequentially, so requiring some time to complete, depending on the underlying infrastructure.

Luca.

[jimkeops]

thanks! i will have the job running tonight and let you know how it goes. thank you again

[captainflannel]

Was curious if anyone had any experience with VEEAM B&R and large windows file share servers. Currently we use an enterprise NAS dedicated for our SMB file sharing which is about 20TB total. A new project were considering moving to a virtual MS server and 2012R2 and take advantage of the deduplication engine.

My question regards backup of the 20TB of data. For performance of backup and restore - I would either plan for VEEAM to perform all of the backup and restore functions of the VMDK (I know there are some requirments for B&R and 2012R2 dedupe) or use our other backup vendor (EMC Networker) software with a client for the 20TB of file shares.

We use VEEAM for everything else virtual, but I've never had to deal with such a large VMDK, CBT and VEEAM. Our backup target is an EMC DD2500, we're following best practices with an NFS Gateway and weekly synthetic fulls at the moment. (Hope to see what VEEAM8 and ddboost offer!)

[foggy]

Please review other users' experience in backing up large VMs. Basically, if you're able to fit in the required backup window, you're fine with such a large VMDK. Thanks.

[bdoe]

What's the current recommendation for file servers of around 20TB in a single volume? In ESXi 5.1, obviously the only choice was to use multiple .vmdk's and use Windows spanning, but 5.5 can go way bigger. I've heard two recommendations - that Veeam would have an easier time with a single 20TB .vmdk, and that Veeam's parallel processing would make sticking with a span better.

I already have one VM with 3TB x7 that works fine, struggling to add an upgraded host with a 2TB x 10 span.

[bdoe]

It turns out my issue was VAAI enabled on an Areca controller - with that disabled, I'm now backing up two 20TB file servers with Veeam, with a third host with a third 20TB VM waiting to be upgraded.

[ITSSCloud]

Good morning.

I wanted to inquire about a recommended maximum size of a machine for backup on it.

now we have in our VMware platform with several machines that exceed 10TB and I wonder if there is any best practice or recommendation of veeam about the recommended maximum sizes for backup.

Regards.

[foggy]

Hi, your post has been merged into existing thread discussing similar questions. Please review for some considerations and hints given above. Generally speaking, there are no any limitations on the VM size, provided your environment is capable of backing it up within the required backup window. 10TB is quite common among our customers these days.

[pizzim13]

The largest we backup are a pair of 20TB VMs with 30 restore points and a weekly transform. 15TB+ VMs are fairly common place in our environment. Large change rates during the backup window and the subsequent snapshot commit can be a challenge .

[lando_uk]

When dealing with large file servers, isn't NTFS chkdsk issues more of an problem than actually backing them up?

[Gostev]

By the way, chkdsk was vastly improved in Server 2012.

[GarethUK]

Hi,

What currently is the maximum VM size that Veeam will support.

If I go and create and backup a 55TB VM as a single disk will Veeam fully support that? Can I undertake an instant VM recovery and do individual Windows file recovery form a machine that big?

I agree a full restore will take ages to recover. But, there has to be a downside to organising things this way.

Regards,

Gareth

[foggy]

Gareth, technically there're no limitations on the VM size, everything depends on the underlying hardware and assumed backup window.

[GarethUK]

Thanks Foggy. That's good to know. The backup wIndow is a good point. However, I'm already doing 5TB VM's and some of them can be done in a few minutes . Over 200 TB's of VM's complete in a 2-3 hour window in my environment.

I've hit a 'business decision' issue where I'm unable to influence the outcome to the extent that a big file server is necessary. I'm not really happy about the idea of huge VM's (for the sake of it) but that's the situation I find myself in. Any issues with restore times etc will be for the 'internal customers' to deal with.

Regards,

Gareth

[chrisdearden]

Is there no way to run multiple volumes on that box ? it might help with some parallel processing.

[GarethUK]

Hello Chris,

Nice to hear from you.

That's a good point. We've steered away from doing that because some of the resulting VM configs where a bit brain dead. But, I may need to revisit that - it would definitely solve my problem of getting a 25TB VM into a 20TB VSA cluster . 5TB spanned volumes helps for a number of reasons - SAN load distribution etc.

For restores the only limiting factor is the 1Gbps interfaces our ESXI hosts have. The extra large VM's will be on a specific set of VMWARE infrastructure so I think I need to look a reworking that. However, I also have to investigate thick provisioning to enable direct to SAN restore but this isn't something I've had a play with because the mere mention of it will have our VMWARE team thick provisioning all our VMWARE infrastructure because they don't *LIKE* the idea of thin provisioning. Which just results in capacity management having to be done on the VSA level which is particularly fun game - not.

Anyway, some ideas I need to go away and consider.

Regards,

Gareth

[davecla]

Interested in knowing if anyone is backing up any really big file servers using B&R?

Currently the largest filer I backup with Veeam is about 25TB and while it's normally pretty good, if it ever decides to do a full backup (CBT corruption or whatever) it takes a couple of days.

I've got 500TB on another filer (single server) that I need to protect, I'm thinking I'll need to take a different path for this, but keen to hear if B&R has been used for this sort of size.


ta
dave

[Shestakov]

Hello Dave,
500TB sounds challenging, but there is no limitation of VM size to be backed up.
How much time does full backup of 25TB VM take? What is the bottleneck?
Thanks!

[nielsengelen]

Is your VM 500TB or are you using something like NetApp filer for it?

[davecla]

@Shestakov - a full backup of 25TB takes around 36 hours (from memory, might be closer to 60 hours, I must check the logs again). B&R reports the bottleneck is the source. I've never looked further into it. There isn't a huge amount of data change, daily's only take 15 mins of so, so once that full is done it's all good.

The bigger filer is physical - ZFS on x86.
Was thinking about virtualising the storage and server so I could use B&R. Logistics of testing are a bit daunting (about a month of copying for 500TB) so wanted to see if anyone else has that approach.

thx
Dave

[Shestakov]

Thanks for the reply, Dave.
Answering your initial question, I`ve not heard of users backing up 500TB VM, but theoretically it should work.
I would make sure everything is tuned in accordance with best practices before start the backup job.
Out of curiosity, do you leverage storage snapshots/direct storage access mode?

[dellock6]

The biggest VM I've seen personally was 90 TB, and backup was going good, with obviously some amount of time required to complete it, especially a full.

My two humble suggestions:
- leverage parallel processing as much as possible by having multiple small VMDK rather than a few huge vmdk. multiple disks can be processed in parallel.
- avoid as much as possible active full backups, except obviously for day1

Luca