Comprehensive data protection for all workloads
Post Reply
TimTheEnchanter
Novice
Posts: 7
Liked: never
Joined: Jan 09, 2014 3:24 pm
Full Name: Matthew Hatfield
Contact:

Netapp Storage as VEEAM Repository

Post by TimTheEnchanter »

I was wondering if anyone has experience using Netapp storage as a VEEAM respository. We are trying to reorganize our VMWare environment and SAN infrastructure to get the most bang for our buck and at the same time taking advantage of the new tape backup features in VEEAM 7 so we can finally get rid of Backup Exec completely.

My plan is to switch all of my reverse incremental jobs to incremental ones and copy them to tape either weekly or bi-weekly. The problem is that this requires twice the disk storage for my VEEAM respository. My question is, does anyone out there have experience using Netapp storage as a VEEAM repository? Does the deduplication on the Netapp work well enough with vbk files where I could potentially store multiple synthetic fulls and only really be storing one plus the differences? Also, if I do end up using NetApp as the repository is there an advantage to any particular type of connection to the VEEAM server? i.e. CIFS, iSCSI, Fiber Channel?

Thanks in advance.
foggy
Veeam Software
Posts: 21069
Liked: 2115 times
Joined: Jul 11, 2011 10:22 am
Full Name: Alexander Fogelson
Contact:

Re: Netapp Storage as VEEAM Repository

Post by foggy »

Matthew, while you're waiting for other NetApp users to chime in, I encourage you to search these forums for NetApp and review the existing topics regarding that, there are quite a few of them. Btw, here's the one that popped up in the quick search to me: VEEAM 7 backup speeds with NetApp.

Regarding deduplication, I take the liberty of quoting Tom Sightler's post, which, I bet, will be useful:
tsightler wrote:There's a couple of things that are important when using Netapp dedupe. First of all, contrary to popular opinion of Netapp fanboys, there really is a performance penalty when enabling dedupe on a volume. When you enable dedupe on a Netapp volume you immediately get a penalty on writes of approximately 7% (this is Netapp's own number). This is because, while the dedupe is "post-process" it still calculates the fingerprint of each block on write. It then uses the fingerprints later during the post-process portion. That's the next thing to remember, there comes a time when this "post-process" dedupe must happen. It will use CPU resources of the filer, and I/O resources of the volume, but, because it is largely just updating metadata to eliminate duplicate fingerprints, it's typically pretty fast. You must make sure that this process occurs outside of the backup window and it will impact read performance of the volume while it is in progress.

If using forward incremental, synthetic full performance is mostly impacted by the fact that you can't use Veeam compression if you expect to get decent dedupe from the Netapp. This means that synthetic full processing must read and write uncompressed data, which is typically 2-3x more than if Veeam compression is used, thus making synthetic full processing take very long, typically longer than an active full.

Honestly, with the customers that I've worked with that have tried both, almost all have ended up simply using Veeam with reverse incremental and not using Netapp dedupe. Veeam Reverse incremental backups, with compression are very space efficient, and remove the complexity of scheduling full backups/synthetic fulls, the retention requirements, and the added space required to store uncompressed full backups prior to the Netapp dedupe process.
Also, here's a huge topic with general recommendations on backup storage: recommendations for veeam backup storage.

Thanks.
joergr
Veteran
Posts: 391
Liked: 39 times
Joined: Jun 08, 2010 2:01 pm
Full Name: Joerg Riether
Contact:

Re: Netapp Storage as VEEAM Repository

Post by joergr » 1 person likes this post

I´d go 100% with Tom´s suggestion. Please don´t expect ANY san vendor to do a better online/scheduled dedup than veeam itself can do. Thus, i´d suggest to turn off dedup on netapp side completely and use veeam dedup+compressions. You will notice soon enough, the veeam algorithms are extraordinary efficient.

If additional dedup is a must because of cost reasons please don´t expect efficiency miracles and please don´t expect performance miracles. External dedup engines usually are optimized for non dedup and non compressed stuff. There may be nice and cost effective things like e.g. storage spaces with w2012/w2012r2 (which are pretty cool, btw) but then again - no system can do magic and no system can abrogate simple physics&logic ;-)

For your special situation i´d suggest using reverse incrementals. If you for whatever reason only need to backup the latest vbk and always a full backup you could play around with something like this here http://forums.veeam.com/tape-f29/backup ... tml#p85284 BUT please do that only after very careful testing in a test lab and (of course) on your own risk. this is ps stuff and this is not tested at all the way i would love to have it tested (my time is sadly very limited, if i find the time i will probably improve all this and test it ;-) ).

Best regards,
Joerg
TimTheEnchanter
Novice
Posts: 7
Liked: never
Joined: Jan 09, 2014 3:24 pm
Full Name: Matthew Hatfield
Contact:

Re: Netapp Storage as VEEAM Repository

Post by TimTheEnchanter »

Thanks all for the info. I'm currently using reverse incremental backups, and I'm getting some very good performance from the VEEAM dedup engine. I'm able to store 31 days of reverse incremental backups of two 4 TB file servers in 9TB of disk space. The reason that I am even considering other alternatives is tape backups. We have a requirement to keep certain data for one year and other data indefinitely, so I've got to use tapes for long term storage. The first thing I noticed is that the tape job doesn't finish in 24 hours and then when the production job kicks off it stops the tape job because it needs the files to create the reverse incremental. This is easy enough to get around with scripts. I've read tons of posts and blogs that say incremental is the way to go when using tapes and it makes sense. I can take as long as I need to backup the full / synthetic full file because it won't be needed again until the next week. Then once a week I can go label and store those tapes. If I could get good performance out the NetApp dedup engine I could potentially store more recovery points on disk and also be able to use the built in scheduling to run the tape backups.

My second problem is which type of tape job to run. I've run the "backups to tape" option and it seems that with reverse incremental it only backs up the vbk file on each run and ignores the vrb's. If I run a "files to tape" job I can store the vrb files as well, but VEEAM doesn't track the recovery points onto the tape so I'll have to manually track which tapes contain which data. This is not a huge deal, but I'd prefer to be able to have VEEAM tell me which bar code to get rather than put them all in, restore the backup files to disk, catalog them, and then restore the file. We regularly have users who request restores of files and they can't remember when the last time they saw the was so I need a reasonably workable way to get them back.
Vitaliy S.
VP, Product Management
Posts: 27055
Liked: 2710 times
Joined: Mar 30, 2009 9:13 am
Full Name: Vitaliy Safarov
Contact:

Re: Netapp Storage as VEEAM Repository

Post by Vitaliy S. »

Hi Matthew,

How often do you need to offload backup to tapes? In order to address your current challenges you can either switch to forward incremental job (this would allow you to track VMs on tapes) or create another backup job that would run only once a week and will be used as a source for backup to tape job.

Thanks!
joergr
Veteran
Posts: 391
Liked: 39 times
Joined: Jun 08, 2010 2:01 pm
Full Name: Joerg Riether
Contact:

Re: Netapp Storage as VEEAM Repository

Post by joergr »

Hi Matthew,

me personally, i see backups to tape as very essential and very useful but only as a sorta last line of defense, thus, we only see requests for restore of old files once in a blue moon. If you have users who often need older files recovered i´d suggest using backup copy jobs and applying a fitting-for-your-needs gfs retention scheme.

If you´d do this, i´d stay with the backup to tape anyways in parallel if i were you - having an additional backup on good old tape is always a good idea. But using backup copy jobs with gfs it would speed up your whole recovery process massively.

Best regards,
Joerg
TimTheEnchanter
Novice
Posts: 7
Liked: never
Joined: Jan 09, 2014 3:24 pm
Full Name: Matthew Hatfield
Contact:

Re: Netapp Storage as VEEAM Repository

Post by TimTheEnchanter »

The only factor driving our current offload to tape scheme is backup to disk space. We currently have two backup jobs running nightly doing reverse incrementals. One job backs up the two large file servers and the other backs up the remaining critical servers (SQL, Exchange, syslog, etc). Each job uses a separate 9TB backup repository and we have found that by using VEEAM dedup and compression that we can store 31 days of reverse incremental backups on disk. Therefore we have chosen to archive to tape once every four weeks. This provides for the production backup to be paused over a weekend and allows enough time for the tape backups to complete. Someone then logs in remotely on Sunday night and re-enables the production backup before the backup window opens. This is a manual process for now, but I'm not concerned with that at this point as it's fixable with scripting as such.

We also run replication jobs to a separate Intel Blade-in-a-box server with a SAS attached Promise disk array. These jobs replicate the critical servers and also store the only backups of our non critical servers (the ones we could potentially rebuild from scratch without much trouble). We store 7 days of recovery points on the replicas. We plan to use this setup for disaster recovery scenarios as we are right in the middle of hurricane country.

I have a few places where I could possible create enough space to store backup copies if that is the way to go. If I was to create a backup copy job of our two main backup jobs that stored 12 monthly recovery points would VEEAM treat that like a reverse incremental with only the differences or would I need the full space that our vbk files take up (8TB) for each recovery point? If I could recover 31 days worth from disk and one point from each month for 12 months, then I could continue my tape backups as stated above, use those only for emergencies, and modify my SLA to state that recoveries from beyond one year will take a week to process.
joergr
Veteran
Posts: 391
Liked: 39 times
Joined: Jun 08, 2010 2:01 pm
Full Name: Joerg Riether
Contact:

Re: Netapp Storage as VEEAM Repository

Post by joergr »

Hi Matthew,

if you use backup copy jobs first of all you must know they are always forward increment, even if the source is reverse. Second, if you apply a gfs scheme, gfs backups are full backus (the ones written weekly or monthly) and thus, need the full space. Conlusion: Yeah, you need space for gfs ;-)

Best regards,
Joerg
foggy
Veeam Software
Posts: 21069
Liked: 2115 times
Joined: Jul 11, 2011 10:22 am
Full Name: Alexander Fogelson
Contact:

Re: Netapp Storage as VEEAM Repository

Post by foggy »

Correct. However, if not enabling GFS rotation scheme, you could set the copy every interval to 30 days and keep 12 restore points in a forward incremental chain.
joergr
Veteran
Posts: 391
Liked: 39 times
Joined: Jun 08, 2010 2:01 pm
Full Name: Joerg Riether
Contact:

Re: Netapp Storage as VEEAM Repository

Post by joergr »

Yep. And if you go foggy´s suggestion (which is a good and sound one) you also could enable scheduled compression (you can do that when using standard scheme).
TimTheEnchanter
Novice
Posts: 7
Liked: never
Joined: Jan 09, 2014 3:24 pm
Full Name: Matthew Hatfield
Contact:

Re: Netapp Storage as VEEAM Repository

Post by TimTheEnchanter »

Great! Thanks for the suggestions. I think I'll try using backup copies every 30 days in addition to tape backups every 30 days. I'll schedule them on alternate weekends so they don't interfere with each other.

Thanks again for all the help. I think I have a good way forward.
veremin
Product Manager
Posts: 20270
Liked: 2252 times
Joined: Oct 26, 2012 3:28 pm
Full Name: Vladimir Eremin
Contact:

Re: Netapp Storage as VEEAM Repository

Post by veremin »

TimTheEnchanter wrote:This is a manual process for now, but I'm not concerned with that at this point as it's fixable with scripting as such.
You can write a simple PS script that enables/disables a given backup job and schedule this script via Task Scheduler.

Disable:

Code: Select all

asnp VeeamPSSNapin
$Job = Get-VBRJob -name "Name of backup job"
$Job.DisableScheduler()
Enable:

Code: Select all

asnp VeeamPSSNapin
$Job = Get-VBRJob -name "Name of backup job"
$Job.EnableScheduler() 
Thanks.
Post Reply

Who is online

Users browsing this forum: ante_704, Bing [Bot] and 29 guests