Large VMs 50TB

[kgoucher]

Hi All

I'm looking for some good advice and discussion around this topic.
I am running some large VMs as file servers. Currently my largest has 2x 15TB VMDK disks attached as seperate file shares on same VM. I need to scale up to larger file server with 50TB-60TB. This is for data for one department that generates very large datasets for thier projects, so I can't separate it in to smaller servers.
Is anyone running VMs of this size and backing them up with Veeam? Is this the best way to go for large file servers?
I'm concerned about restore time and RPOs. How are others handling very large file servers.

regards
Kieron

[Gostev]

Honestly, I would not consider it a "very large" file server. I regularly meet customers who have file servers of 100TB and more.

[kgoucher]

That's very interesting to me. Are they using single VMDK to back the file share or multiple VMDKs with DFS Namespaces

[Gostev]

Well, it's always multiple VMDKs even just because the single VMDK is limited to 62TB in size. And how they present this capacity always depends on the workload, but that's not something we usually discuss since for Veeam, it does not matter. The usual topic is that guest file system indexing takes too long on such monster file servers.

[Regnor]

The main problem I see with such large VMDKs is the performance. You can achieve great numbers in parallel (multiple disks) but single disk backup will often only get limited performance. It will depend on your storage type and backup mode so just do some tests in your environment.
This also counts for restores, so be sure your compliant with the defined RTO.

The other topic I often see at customers is the tape backup. A single large VM can only use a single tape drive when offloaded to tape. This will probably be even slower than your backup speed.

I often recommend seperate the data into smaller VMs or disks. It will be a bit more complicated to build but better to handle in terms of backup.