Feature request - Recycle Bin

[ejenner]

Hi guys,

I've always wondered if there was a way to protect against deleted files using a regular backup. We recently found out when a user deleted the only copy of a folder 530 days ago. We have good backups for the area where the file was deleted but we've not been performing incrementals on that drive for anywhere near that length of time. Only discovering the problem this week has meant the folder is long gone.

My idea is that Veeam could very easily determine the last time a file was seen in a backup job and output it to a deleted files cache on a rolling basis. This would be a special repository the backup administrator would maintain according to his organisations requirements and the feature could be enabled per-job with a check-box.

I imagine this feature would be useful on file server backups and you'd be less likely to enable it on a domain controller backup (for instance)

Various other controls could be included such as notifications for the condition (i.e. capacity, ect) of the deleted files cache.

I think this would be a neat way around the problem identified above and removes the requirement for keeping extremely long history to cover this eventuality.

[HannesK]

Hello,
I feel NAS backup in V10 will do exactly what you are asking for.



For VM backup (snapshot based block based backup) this is not possible because the backup does not know about files. It's just blocks.

Best regards,
Hannes

[ejenner]

That's great except our main fileserver is a Windows Agent backup of a clustered fileserver. Does the functionality cover that kind of backup or is it just for NAS boxes?

[Egor Yakovlev]

Hi EJ.
Mentioned file archival feature works only for File Share Backup. However you can combine both of two worlds - have Agent job to backup system drives, and File Share job to protect your file servers data - in case server dies you can do system restore using Agent backups, in case files restore is needed you can use File Share Backups!
/Thanks!

[ejenner]

So it targets any file share, even if it's not specifically a NAS device... i.e. it could be a Windows file server.

If so then that's great news and I withdraw my feature request as it's built into the new version.

[HannesK]

yep - it does not care whether it's a Windows SMB share or an SMB share from any NAS vendor (same for NFS)

Actually you could even backup any managed server (Windows / Linux) with NAS backup. But you have a cluster, so adding the SMB share is the way to go.

[ejenner]

Would there be any issue with the cluster node changing from one place to another and confusing situation by appearing as new data every time it switches node?

[Egor Yakovlev]

No challenge, as we track file share name and track tree structure underneath. Which will not change no matter the file share owner node.

[ejenner]

Is that a v10 improvement as I believe 9.5 thinks it's a new backup if targeting a DFS copy for instance?

[HannesK]

9.5 does not have NAS backup capabilities. Everything in 9.5 is block based (except file-to-tape)

[ejenner]

I'm having trouble saying what I mean I think... we'll try the new feature when we get it and see how it goes.

[RaymondKN]

I found a good way to cover deletions was to have a mirror folder separate from the backup and backup software that was mirroring all the files, if changed, but would not delete any files so I could look here if needed. It has been easier to find files here than try to dig them out of backups, only downside is if the file is changed it updates and you loose the original.

[randy16randy]

I've been using Condusiv's Network Undelete Server for almost 10 years in my environments. It is essentially a network recycle bin for recovering files. Allows multiple previous versions to be kept. I run it on my main file servers specifically for cases where users save files with incorrect changes over good copies, or accidentally delete files or folders. Given enough working space for the recovery bin, previous versions of files can be held for years. This is always my first stop when recovering files as it is even quicker and easier than Veeam - improbable as that may seem

[Egor Yakovlev]

Veeam NAS backup is able to keep up to 999 "Active" and 999 "Deleted" file versions for every single file protected, as long as you have space available for up to 999 years in the archive!
And secondary archival storage could be quite cheap(cloud?) too. I encourage you to check it out with v10 Release!
/Cheers!

[Gostev]

Veeam NAS backup is able to keep up to 999 "Active" and 999 "Deleted" file versions for every single file protected

Actually, unlimited versions (by default).

[Gostev]

This is always my first stop when recovering files as it is even quicker and easier than Veeam - improbable as that may seem

Actually, this is quite expected. File-level recovery from a file-level backup has got to be much faster than one from an image-level backup, because the former does not have to mount the file system before performing the restore.

I too am blown away by the initialization time of file restore from v10 NAS backup, which is basically instant - especially after getting so used in 12 years to how long it takes with an image-level backup

[ejenner]

My main fileserver takes about 45 minutes to load for recovering individual files. It's also my most commonly requested restore source. I posted about this separately: veeam-backup-replication-f2/guest-file- ... 64116.html

[Gostev]

v10 will partly help, as mount is now fully isolated to the mount server (there's no backup mounting to the actual backup console by default). You will see this in the What's New document, once it's published.

[YouGotServered]

I feel NAS backup in V10 will do exactly what you are asking for.

Hannes,
Couldn't Veeam theoretically integrate this with the guest processing capabilities that Veeam already has? We already have guest file system indexing, wouldn't this just be a further evolution of that? If we are already indexing all the files, I imagine it would be pretty simple to do a simple comparison on them. When I have to do file server migrations, I use some pretty lightweight software that is able to compare millions of files between two directories in just a few minutes. I imagine this would be a pretty simple post-processing task for Veeam. Of course, everything is more complicated than it sounds, but I think it is definitely doable.

[HannesK]

Hello,
not really. That would be going back to 1990s style file backup vs. 21st century block based backup.

If we are already indexing all the files,

for VM based backup we are not indexing anything per default. The guest indexing is an additional task we only do to allow search capabilities. Indexing has nothing to do with VM / image based backup

Up to a certain point we can exclude files from block based backup (BitLooker). But that costs performance. And building something similar to NAS backup capabilities would be the opposite of what we do today.

Best regards,
Hannes

[YouGotServered]

Hannes,
I understand that we don't index anything by default, and that guest indexing is an additional, optional task. If we opt-in to the indexing feature, which already goes through and indexes all the files, couldn't we simply mark which files have been deleted if the indexer can't find them?

Indexing is a process that is deployed by the guest interaction policy as a runtime process during the backup into the guest VM, right? I understand the overall backup is block level, which is better than the 1990s style file-based backup, but my understanding is that the guest processing gives us capabilities into the guest VM over and above block level backup - such as file indexing and potentially, keeping track of and recording deleted files

Does that make sense, and does that sound like something that is feasible with the guest interaction proxy? Heck, maybe we wouldn't even need to make any adjustments to the proxy, but the backup server could then compare the index built from the most recent job run vs the previous runs, provided that indexing is turned on. Does that seem feasible?

[YouGotServered]

@Hannesk, am I crazy, or would my suggestion be possible?

[HannesK]

I would never say you are crazy But not possible...

[YouGotServered]

Haha I appreciate that. Could I ask why it wouldn't be possible to compare guest file indexing indexes if we have enabled that feature?

[infused]

Just out of interest, has this option changed? I can't see it in v10

[Egor Yakovlev]

No changes. Feature request noted, however there is no ETA on arrival.
/Thanks!