File Share Backups

[perjonsson1960]

Folks,

Are File Share Backups something new that has been added to B&R recently? Or has it been there for a long time? I haven't noticed it before...
Is this a better way to backup a file server cluster, rather than backing it up in a standard backup job as a Failover Cluster?
I see that the File Share Backup has version handling for files, which could be a good thing to have...

Sincerely,
PJ

[Egor Yakovlev]

Hi Per,

NAS Backup came out with VBR v10 in Q1 2020. Yes, it is totally different approach for file-level protection with many additional features. If you are backing up file shares with image level backup today, definitely consider NAS Backup as an alternative approach(per-file versioning instead of block level change tracking, per-file archival strategies based on file age\versions history, per-share rollback in time just to name a few!

You can find full guide on the feature here.

/Cheers!

[perjonsson1960]

Thanks!

If I want to backup file shares in physical Windows servers, must the backup proxy also be a physical server, or could it be a virtual machine in VMware?
And have I understood correctly when I say that the backup proxy and the cache repository are ideally assigned to the same machine?

[Gostev]

It could be any machine really, including the backup server itself if you share is not too big. In fact, I believe we do create the default proxy on the backup server automatically.

The key for both proxy and cache is to be on machines in the network proximity to the protected file share (same data center). I can't think of any reasons not to have both on the same machine.

[NightBird]

Ok for file share backups, but how to respect 3-2-1-1 rules ???
File share on prem, offload (copy) to S3 not possible ( (immutable or not), only archiving to S3, no File to tape job (as backup to tape job).

So how can we respect 3-2-1-1 rules with file share backup actually ?

Agent backup of file share server repository, so you can offload it ?

[Gostev]

No need to jump through the hoops there, we support NAS backup copy to a regular repository natively. And more options will come in the future

[perjonsson1960]

The "cache repository" used in a File share backup job, is that something that is used only during the backup, or is the data stored there needed for the backup chain and restores to work?

[Egor Yakovlev]

Only for backups.
It keeps changed files metadata to make incremental backups more efficient.

[perjonsson1960]

Okay. Is there any way to determine how much diskspace the cache repository must have? Is it possible to calculate the size needed, perhaps based on the amount of source data and number of files? I am trying to locate a physical machine in our infrastructure that can take the roles of cache repository and proxy.

[Egor Yakovlev]

It actually more dependent on number of folders, however in any case cache size numbers are negligible.
Cache for 11,6M files in 2,4M folders will take roughly 250MB(numbers from internal labs).

[perjonsson1960]

Okay. Not very much, then.
And what about backup proxies? Are there any prerequisites there for disk and RAM?

[Dima P.]

Hi,

Proxies do require local storage ~300MBs for component installation. You need at least 4GB of RAM per task (task would be a single file share processed by the proxy), more RAM = more task to process in parallel. Thanks!

[perjonsson1960]

Only for backups.
It keeps changed files metadata to make incremental backups more efficient.

I am beginning to think that perhaps you misunderstood my question. In the help center it says:

"When the cache repository receives a new CRC tree structure from the proxy, it compares it with the CRC tree created during the previous run of the backup session. If any files or folders of the file share have changed since the previous backup session run, the cache repository instructs the backup proxy to start reading changed data from the source file share."

It sounds to me as if the data stored on the cache repository is needed from one session to the next? Or have I misunderstood? What happens if I empty the cache repo between sessions?

[Egor Yakovlev]

No, file proxy cache has no strict requirement to be present between backups.
It is meant to increase performance\optimize calls to the file share in the beginning of the session. If you delete cache between runs, you will still get normal incremental backup from last run, it might just take a little longer for us to get full list of changed objects since last run (due to no cached directory tree to compare with).

[perjonsson1960]

Okay, I understand! Thanks!

[tedsteenvoorden]

With image level backups of file servers only the changed blocks are being backupped. E.g. if 1 MB of a 500 GB file is being changed, only the 1 MB will be transferred. Does file share backup also have a mechanism to backup only the changed part of a file or is the complete file being backupped/transferred if only a part of the file has changed?

[Dima P.]

Hello Ted,

File level backup does not operate on block level, instead it tracks the modified files and backs up entire file version whenever it's changed. Thanks!

[tedsteenvoorden]

Hello Dmitry, thanks for the feedback. Is there any form of compression/dedup to reduce the file share backup storage?
Also, is it correct when I say that file share backup is not an optimal backup for shares with many container files (e.g. FSLogix containers), due to the complete containers being backupped by each change?

[wishr]

Hi Ted,

We provide various compression levels, see here.

Regarding deduplication, please refer to that post.

File backup is the most convenient solution for situations when you have a lot of files on a file share and have a requirement to maintain backups of several versions of each file while also being able to perform granular file restores with low RTO. Also, if the amount of files is massive, file backup would give you a better backup speed in comparison with image-level backups. You may use a deduplication appliance as a target and also archive aging files data to an object storage to minimize storage costs.

I would also recommend giving our file backup a shot at least in a sandbox environment and then decide whether it fits your needs.

Thanks

[tedsteenvoorden]

Hi Fedor, thanks for the usefull information. I will certainly give the file backup a try!

[mberrey]

How often are file shares scanned for updates in VBR?

[rdixon01]

We are running V12.1.1.56 and are looking into possibly backing up our SMB file shares to our Scality Ring. Could someone guide me to documentation on Best Practice, Sizing of file proxies, etc.
Thank You