Veeam Best Practises?

[kennyt2000]

Hi All,

Below is a guide i wrote from my own experiences using and setting up veeam within our environments. My question is do you guys have anything to add that would make the backups / replications run smoother and keep backup file sizes down to a manageable level? please note Im not complaining or suggesting we have a problem here just trying to see if we can learn / improve anything.

We do this for each VM before adding it to our backup job or when we notice something strange occuring.

cheers in advance

Veeam Tips and Tricks for a clean and simple backup environment

Step 1 - Indexing
Unless needed for specific reasons disable file indexing on the servers:
1. Server 2003
• Run ciadv.msc and check where the index file locations are and if the service is running
• Run services.msc and set the indexing service to be disabled if you do not want it to run.
• You may also want to delete the indexes in the ciadv.msc console to free up used space on the server.

2. Server 2008
• Check if the windows search service or 2003 indexing service has been added to the file services role. If so consider removing the role services or disable the windows search service using services.msc
• Check the current status by opening the ciadv.msc (indexing service) or running the indexing options menu from the control panel (windows search service)

Step 2 – Volume Shadow Copies

Veeam treats these the same as any file within a vm and as such due to the nature of CBT (Change Block Tracking) as they change disk blocks which can add a large amount of data into your backup files. If you do not need the volume shadow copies to be retained by each backup then follow these simple steps:
1. Calculate the amount of space needed for your volume shadow copies across all drives within your VM, a rule of thumb is normally around 10% of the original disk size but this can vary depending on your environment and frequency of shadow copy schedules.
2. Add a new disk to your VM and based on the size calculated below
3. Prepare the disk in the VM so it is ready for use (I normally label it accordingly Volsnaps no backup)
4. On each disk disable the current shadow copy (Warning this will delete any existing shadow copies for that drive)
5. Re setup the volume shadow copy ad schedules for the drive but in the storage location when setting up the shadow copies select the new volsnaps drive.

Step 3 – Server Clean up

Windows retains a large amount of files especially when service packs have been applied, in most cases these can be safely removed as rolling back is not normally required after a certain period of time. To reclaim this space and reduce the space burden on your server do the following.
1. Server 2008 x96 /x64 SP1/SP2
a. Run an elevated command prompt
b. Enter Complcln.exe
c. Say yes and let the process complete
2. Server 2008 R2
a. Run an elevated command prompt
b. DISM /Online /Cleanup-Image /SPSuperseded

Step 4 – Page File

The page file is included in Veeam backups for any version below V6, it is recommended to move the page file to a separate Hard drive and exclude it from the backup job. Windows will always create a new page file on C:\ if one isn’t found during bootup and restore from a backup image.

Step 5 – Defrag

Once you have cleaned up the servers time to defrag each drive ready for use. Follow usual procedures.

Step 6 – Sdelete

I encountered an issue with a 2003 server which was utilising only 14 GB of space total yet would consistently track and transfer 70GB+ into the backup set. On investigation it was found that old deleted data was still being tracked by the CBT and included in the jobs, this is even months after the files were removed. To overcome this and now part of my best practise do the following: (Warning this will expand and thinly provisioned disks)
1. Download sdelete from the Microsoft site
2. Run sdelete –z driveletter:\ on each drive you have within your server.
This will reset all empty blocks to Zeroes so they are not tracked in the backup, or at least not backed up. This reduced my backups of this VM from 70GB per night to around 350mb
Note: this will use up all free space for a brief period of time so be careful with any production servers.

General Notes:

• VM builds and disk additions: where possible use Thick provisioned Eager Zeroed disks as this will basically act like an sdelete on the disk before it is used. It also makes disk usage faster if you’re creating new data.

• IIS Servers:
o Disable logging unless the logs are required for a specific purpose

[dellock6]

Sorry but I disagree on some points:
2. sounds really over-complicated and both time/space consuming. What if I have 2000 VMs?
5. Defrag is one of the worst activity you can do for CBT, since moving around all the files creates lots of new CBT blocks. If the VM is not showing really bad performances, I always avoid defrag activites.
7?. Veeam Backup is a portion of a VMware infrastructure design, but you also have to consider the requirements of the production storage. Unless you have VAAI capable storage and it has large amounts of data, eager zeroing disks is I/O intensive and cannot allow space savings.

Others sound ok.

[deduplicat3d]

There has also been extensive discussion on sdelete. My personal opinion is that it should be use only if
1) You need to maintain a HD that is much bigger than the amount of data you are using
2) You had previously filled up the harddrive with data that is never going to be duplicated again (i.e. you put a bunch of temp data on the harddrive and then deleted it)

Making a practice out of using sdelete will make increments larger than they should be due to all of the new zeros.

So if you want to do a one time sdelete, it's OK in certain circumstances. It's a lot of manual labor (especially with lots of servers) and new data consumes any recently freed up blocks so not really worth it.

[Daveyd]

Sorry but I disagree on some points:
5. Defrag is one of the worst activity you can do for CBT, since moving around all the files creates lots of new CBT blocks. If the VM is not showing really bad performances, I always avoid defrag activites.
Others sound ok.

That is true but there is really no harm defragging right before a full backup. I do full backups once a week and have my VMs defragged the night before the full backup. It keeps the incrementals reasonable during the week.

[kennyt2000]

Sorry but I disagree on some points:
2. sounds really over-complicated and both time/space consuming. What if I have 2000 VMs?
5. Defrag is one of the worst activity you can do for CBT, since moving around all the files creates lots of new CBT blocks. If the VM is not showing really bad performances, I always avoid defrag activites.
7?. Veeam Backup is a portion of a VMware infrastructure design, but you also have to consider the requirements of the production storage. Unless you have VAAI capable storage and it has large amounts of data, eager zeroing disks is I/O intensive and cannot allow space savings.

Others sound ok.

Hi there,

This is a procedure we follow before adding a machine into our backup job not something we do on a daily/nightly basis. These steps saved massive amounts of space and money when you factor in the cost of backup tapes and having to incresae bandwidth for replication.

I agree this may not work for all environments but i was simply looking for ideas to help streamline our setup further. Thanks anyway for your feedback.

[chrisdearden]

I'd also be careful with defrag & thin provisioned disks ?

[vBPav]

A good thread on defragging guest OS is below. I agree with Chris, defrag on thin-provisioned disks is no bueno. As a matter of fact, unless the defrag technology is VMware Ready i would 'personally' advise against it.

http://blogs.vmware.com/vsphere/2011/09 ... st-os.html

[kurt2439]

I also always heard to stay away from defrags on guest OS machines and that article provides a number of reasons.

Interesting idea about the Volume Shadow Copies. Did you cut back your backup sizes significantly doing that?

[Vitaliy S.]

For the future readers - this topic explains why sdelete shouldn't be used for thin disks: Veeam v6 and sdelete/defrag optimisations for CBT.

[kennyt2000]

I also always heard to stay away from defrags on guest OS machines and that article provides a number of reasons.

Interesting idea about the Volume Shadow Copies. Did you cut back your backup sizes significantly doing that?

sorry for the slow response, i was just re-reading my posts.

With one of our active file servers we were seeing 20GB transferred to the job on a daily basis, this dropped down to about 1.5GB once we moved the Volume Shadow Copies to another disk. For us the volume shadow copy versions were not critical but its best you consider if you would want theses in the even of a failure on your site.