Veeam Active Full Backup

[penfold]

We have just had an issue with our SAN and as a result or Veeam Backups failed at the weekend which is when they are set to run a synthetic full. Anyway I am looking at changing our backup to tape and did a little more reading on Veeam but I think I've managed to confuse myself on the Veeam forums and wanted to check that I'm not doing something which I shouldn't be.

 

Currently our backups run a synthetic full at the weekend and transform previous full backup chains into rollbacks to save space. Now I've read that as this relies on the previous incremental/rollbacks being good you could end up in a situation where (if used for a long time) the synthetic full may contain corrupt data. So should I be using the Perform Active full backup option as currently I've been relying on the Synthetic fulls to created our Full backup.

 

One consideration I've got is should I be running an Active Full once a month to limit any corruptions? We are trying to reclaim some space on our servers by performing some archiving but until this is done if I need to run Active Full Backups I am going to start running out of space!! But equally I don't want to rely on Synthetics if they shouldn't be relied upon long term. Can anyone clarify if I need to run Synthetics & Active Full?

[Cache]

From the recommendations I've read, Synthetic Full backups are fine, but as you mention if a corruption/problems rolls it's self into the backup then the synthetic will keep that defect within it - the last recommendation I saw on the Veeam forum was every 1 - 3 months setup the job to automatically perform an Active Full backup to ensure that a fresh, full backup is taken without faults (But I manually run it every so often to manage the space as I haven't got enough currently like you).

 

To be honest, this wasn't explained to me at the time and for nearly a year we were running Synthetic backups until I started to question why one backup group was so large when I'd removed a load of VM's out of it. Turned out it was because our setup was set not to remove VM's nolonger backed up until an Active Full was performed.

[lmgtfy]

We tried synthetic full like you on a weekly schedule and although during the week days the backups were fast at the weekend the synthetic full was taking 48 hours plus to complete or sometimes failing altogether. We then went back to reverse incremental, it's slower on average but still completes all our backups within our backup window, takes less space and you always have a latest full backup so recovery should be faster.

[zag]

I'd look at why the backup was corrupted in the first place and fix that first.

[penfold]

We don't have an issue with synthetic fulls taking too long (well it seems the only time is due to an issue with our VM). But I was more concerned about keeping backups as synthetic fulls thinking all was well only to discover that our backups become useless due to a corruption being rolled into a full backup. I think I may need to start looking at performing Active Full's but not sure where I'm going to find the space from to be honest!!

[penfold]

I'd look at why the backup was corrupted in the first place and fix that first.

 

How do I do that? If I see the job complete successfully I'm not sure how I can see if it is corrupted unless we test the system regularly. We do test our systems, probably not as often as we should, but if I see the job complete I don't want to be testing every backup we take but I do want to know it can be used.

[zag]

How do I do that? If I see the job complete successfully I'm not sure how I can see if it is corrupted unless we test the system regularly. We do test our systems, probably not as often as we should, but if I see the job complete I don't want to be testing every backup we take but I do want to know it can be used.

 

Try a regular restore to a sandbox environment in Veeam.

 

I do this once a month or so.

[lmgtfy]

@penfold Is there any reason why you do not use reverse incremental?

[penfold]

Try a regular restore to a sandbox environment in Veeam.

 

I do this once a month or so.

 

We do this, but not once a month...more like every 6. That's why I am looking to see if I am doing something wrong as I wouldn't like to go 6 months and find that our backups are useless. We only keep disk based files for 1 week and then they go off to tape so if we need to go to tape to restore I would like to know it would work.

[penfold]

@penfold Is there any reason why you do not use reverse incremental?

 

We backup to tape and I read it wasn't recommended.

[stujham]

I changed our backups from Incremental with synthetic fulls to reverse incremental and have not looked back, these are all backed up to our 100TB backup SAN and then off to tape from there. I managed to save approx 35TB off the Backup SAN and therefore able to increase our restore points on the File servers to over 3months allowing for instant file recovery of any deleted/lost data within 5minutes.

 

Again instant VM recovery is a breeze and works a treat.

[penfold]

I changed our backups from Incremental with synthetic fulls to reverse incremental and have not looked back, these are all backed up to our 100TB backup SAN and then off to tape from there. I managed to save approx 35TB off the Backup SAN and therefore able to increase our restore points on the File servers to over 3months allowing for instant file recovery of any deleted/lost data within 5minutes.

 

Again instant VM recovery is a breeze and works a treat.

 

If I had 100TB to play with I don't think I would be having a problem, unfortunately we just don't have that kind of space

[zag]

Yeh we do Reverse Incremental as well onto 1tb of SSD disk.

 

I now have 60 work day restore points on our main file server so we can go back nearly 3 months to restore any file in seconds.

[stujham]

We did have to increase memory on our backup server/Veeam proxy as reverse incremental is quite IOP intensive but overall the space saving was worth it. Like I said we have 90+ days on fileservers along with 90+ days on exchange we can then do quarterly archiving from disk to tape and just checked we still have 28TB to increase further if required.

[penfold]

So does reverse incremental save space then? I thought it would still require the same storage space as Full+ Incrementals but it wasn't recommended if backing up to tape? I'm just trying anything

[stujham]

A good place to start is: Veeam backup methods and the impact on destination storage I/O | Virtual to the Core this will help you decide what is best for your situation and explains the pro's and con's.

[penfold]

Just to bump this thread with Version 8 now. I am looking at moving all our jobs over to forever incremental. I have read that the requirement for a regular active full backup is reduced in later versions and therefore you can move over to 1 full + incrementals. From the Veeam Forum about Forward Incremental- Forever Backup part way down veeam have posted

Specifically to Active Fulls: I don't recommend scheduled Active Fulls at all, but perhaps doing them manual only, as needed only in the following circumstances:

a) SureBackup detects unrecoverable backup, or backup file data corruption.

b) You want to reduce VBK size (after rearranging jobs or deleting lots of VMs), or its fragmentation

 

So if we run Surebackups to test our backups can we remove the need for performing scheduled full backups now? I'm looking into implementing this on a few systems as it would mean we could keep much more restore points on disk. Is anyone already doing this and if so how have you found it?

[Steve21]

@penfold We still have active fulls running on all our backups periodically, although surebackup checks that the physical VM will load, it doesn't check anything in terms of data corruption etc, so you could have for example your whole SIMs drive corrupted, and as long as the SIMs server/SQL loads it reports it's happy.

 

Steve

[penfold]

@penfold We still have active fulls running on all our backups periodically, although surebackup checks that the physical VM will load, it doesn't check anything in terms of data corruption etc, so you could have for example your whole SIMs drive corrupted, and as long as the SIMs server/SQL loads it reports it's happy.

 

Steve

 

That's a similar thought to the take I had on it which is why we run active fulls. But as I was reading Veeam are saying you no longer need to do active fulls I am thinking we can change the way we backup. I'm currently looking at running some powershell scripts which will attempt to pull data from databases with the theory being that if it can do that, not only it is happy that it loads but you can actively check the data too (not sure if that is possible yet).

 

I still think we will still need to do active fulls but I'm hoping we can do it much less often.

[Steve21]

That's a similar thought to the take I had on it which is why we run active fulls. But as I was reading Veeam are saying you no longer need to do active fulls I am thinking we can change the way we backup. I'm currently looking at running some powershell scripts which will attempt to pull data from databases with the theory being that if it can do that, not only it is happy that it loads but you can actively check the data too (not sure if that is possible yet).

 

I still think we will still need to do active fulls but I'm hoping we can do it much less often.

 

It's not like you need a lot of them, we have like 1 per month on "low risk" machines, and 1 per week on the high. Over 30 point retention it's 1 point for low machines, 4 for big etc.

 

Just honestly don't see the point in risking not using them for 1-4 points of full backups etc And to do a custom surebackup script for every server etc seems faff

 

Steve

[penfold]

@Steve21 While you don't need a lot of them they soon take up space when you have systems closing in on 1TB usage:( which is why I am trying to change the way we backup. I don't want to create a script for every server, just for a couple of systems which would be nice to know there was a problem straight away rather than waiting for a new active full to be created on schedule. We wouldn't really have a problem if we had a larger storage area for backups but unfortunately we don't have that and it seems a low priority at the minute as we can currently backup all we need for a week and then also go to tape.

 

I'm just trying to make my life easier and move towards longer retention on disk in addition to arching to tape.

[Steve21]

@Steve21 While you don't need a lot of them they soon take up space when you have systems closing in on 1TB usage

 

What's your backup server running on? We're using 2012R2 so using Dedup on the backup files, and getting very nice results. The server supports 13TB of backups, and currently using 10TB with 8TB additional in Dedups. So 21TB total in reality.

 

But I mean you can always delete the ActiveFulls via retention policy etc? We're using reverse incrementals so the latest backups are always the big ones, all the rest are the smaller ones. It just ensures a full backup is done and the chain reset, rather than incrementing bad data etc. Don't actually need to keep the Full Actives long term etc, as the data is used to refresh the normal one. If that makes sense?

 

Steve

[penfold]

We are running on Windows 2012 R2 and have dedup enabled but to be honest I think our overall backup needs updating as we have live systems that are doing their own backup which then is being backed up via Veeam. This is one of the things I am trying to sort out but it's a slow process as I'm not getting all the information I need. The main problem we have is we need to update our VMWare version as we are stuck with 2TB limit on the drive. It's something I want to look at upgrading at some point but it's been put on hold for other more important things by those above.

[Steve21]

Are you using a virtual veeam server then? Aye guess the 2TB would be an issue then in regards to storing the additional fulls :s Guess only other option currently would be attaching more drives to the server so it has more space etc to store the files, but even then you'll struggle to fit multiple 1TB files on it

 

Steve

[penfold]

Are you using a virtual veeam server then? Aye guess the 2TB would be an issue then in regards to storing the additional fulls :s Guess only other option currently would be attaching more drives to the server so it has more space etc to store the files, but even then you'll struggle to fit multiple 1TB files on it

 

Steve

That's exactly where we are at. But you know these short term workarounds....they tend to become the working practice