Re: [Veritas-bu] VTL & NetBackup Best Practice

OK….since I did not hear from any of my former co-workers at my previous job on this subject, I will chime in.

We were using VTL’s and we did do multiplexing. We kept the number of drives per VTL down to 8 and we had 6 VTL’s. All of the VTL’s were connected to one ADIC i2000 tape library with LTO-3 tape drives. The physical library was split up into partitions, one for each VTL.

At first we were using Vault to go to physical tape and we had set up our virtual tapes to be smaller than physical tape. This did not work very well (it was slow and it did not scale) and we ended up switching to a method where we did the copy to physical tape off of the back end of the VTL (NetApp, in case you were wondering). This Direct Tape Copy method has worked very well and we were getting tape drive speeds around 50MB/sec as opposed to Vault which was around 10MB/sec avg. (BTW, at last count, we were doing over 1PB per month to our VTL’s or somewhere around 300TB/week.)

As another poster stated, you do need to over-subscribe, but that’s really not a problem. Restores from VTL are very quick if the tapes have not expired off. If they have expired off, all you need to do is to start the import of the physical tape (the barcodes of the virtual tapes are the same as the physical tapes that they get copied off to) and as soon as the import has started you can begin the restore. I set up a script to check for available tapes per VTL and then assigned new tapes as they were available in the physical library.

-Stuart Liddle

From: veritas-bu-bounces AT mailman.eng.auburn DOT edu [mailto:veritas-bu-bounces AT mailman.eng.auburn DOT edu] On Behalf Of Mike Ferlote
Sent: Wednesday, March 05, 2008 5:35 PM
To: Kevin Whittaker; veritas-bu AT mailman.eng.auburn DOT edu
Subject: Re: [Veritas-bu] VTL & NetBackup Best Practice

Thanks for all the replies... Good feedback (I wasn't looking for a right or wrong answer).

I have looked into the Disk staging units (i.e. openstorage APIs etc..) but don't feel that this paradigm is mature or ready for production. I am also unclear about the API (and overall using NDMP for a new purpose) and the vendor support when it will come to using some of these disk features on 6.5. Anyways, that is a whole discussion on its own. Also I feel the API will lock us into a single solution... so I am with Curtis on that one.

I am testing a couple vendors, both NetApp and EMC. Currently on the EMC one but the message is fairly consistent from both vendors. For your first issue they now support libraries that can have 100+ drives... Also if you use the fact that they put a Media Server on their EDL then you don't have to use your Master server to clone everything and that would drastically clean up your solution. I think if the price is correct and I architect this correctly then the VTL solution is worth every dollar... especially with the dedup roadmaps for these products that will uncover a lot of potential!

Thanks.

----- Original Message ----
From: Kevin Whittaker <Kevin.Whittaker AT syniverse DOT com>
To: Mike Ferlote <merked AT rogers DOT com>; veritas-bu AT mailman.eng.auburn DOT edu
Sent: Wednesday, March 5, 2008 1:54:42 PM
Subject: RE: [Veritas-bu] VTL & NetBackup Best Practice

Mike,

Last January, I implemented a VTL (EMC CDL720) with 35 usable TB. We did much study on the multiplex issue in most instances the backup did perform faster when it was not multiplexed, but with out a doubt the restore AND duplications speed is around 4 times as fast. So, I am working towards having enough tape drives to stop all multiplexing in my environment.

Here are my 2 issues....

I have a heavily clustered environment with 27 Media Servers. Well, an ORACLE DB could fail over from one server at any time. Well, when I originally setup the servers with 1 or 2 drives each, there was not enough on a robot with 20 drives to have the same robot on each media server. Well, what happens when the server A goes down that points to robot 1, and the DB fails over to server B and it only sees robot 2?!? I would have to scramble to make robot 1 visible on server B. I did see something in the VTL that might allow me to transfer the tape over from one robot to another and then I guess I would inventory the robot.... but alas I am unsure.

Issue 2.... I do all my vaulting on my master server. Since my media servers are also production servers, I do not want to hit them with so much IO for duplication. So the master server needed to have tape drives on each robot.

So, I after much thought and upgrading to 6.5 I realized that SSO is not so bad. In fact it is wonderful! Media servers and join the SSO and pull out of it with no need to shutdown netbackup on the master server! Configuring tape drives is one step away by typing "tpautoconf -a".

Also, now I share the robots with in the VCS clusters and make sure that each server in the cluster can see all the robots. So, I can do restores with out any issues.

Lastly, I don't have to generate 13 robots with 20 tape drives each to turn off multiplexing! Yes, that is how many tape drives I would have needed to meet that demand.

So, I recommend that you don't turn away from SSO. I believe it is still a great feature and runs so much smoother in 6.5.

P.S..... Actually I would say NOT to purchase a VTL. I believe with all the new great features in NB 6.5, make the Disk Staging Units to be the best options for most people.

P.S.S.... Remember when you turn off multiplexing on the VTL, you will HAVE TO over subscribe the number of tapes, so that you don't run out of media!

Please feel free to e-mail me any other questions.

Kevin

From: veritas-bu-bounces AT mailman.eng.auburn DOT edu [mailto:veritas-bu-bounces AT mailman.eng.auburn DOT edu] On Behalf Of Mike Ferlote
Sent: Wednesday, March 05, 2008 6:25 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] VTL & NetBackup Best Practice

Hi,
I am currently testing my NetBackup environment to backup to VTL. We are seeing great backup performance but it seems like it requires a slight re-architecture to the current environment. I am being told to eliminate multiplexing... You can multiplex to a virtual drive but when it comes time to restore and to duplicate off to physical tape from a virtual cartridge that was multiplexed the performance is not very good. The Vendor said that multiplexing should be eliminated and instead of increasing multiplexing you can simply create additional virtual drives when needed. Since adding virtual tape drives in theory has no cost associated with it why not do this and eliminate multiplexing and SSO all together (that is the approach/mentality).

So my question to the forum is have others deployed VTLs in a similar fashion (i.e. MPX = 0) and how have things scaled and is there any associated management headaches? The couple con's of the above approach in my mind would be:
* Slower responding NBU GUI because there will be so many devices it will have to manage/query
* More BPTM processes since each tape drive in use requires one an additional BPTM process
* Slower backups on a per stream basis (which is OK because you have many drives I guess)

Can anyone comment if they foresee other drawbacks to this approach or have best practice recommendations?

Thank in advance for any feedback...
-Bill

_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu