Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
That's what I'm saying. Makes more sense to license VTL as VTL by capacity. It's also possible, I believe, I'd have to check the licenses, to license VTL as VTL by virtual drive. Paul -- -Original Message- From: Curtis Preston [mailto:[EMAIL PROTECTED] Sent: July 30, 2007 4:23 PM To: Paul Keating; veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. FWIW, The official answer from Symantec is that if you're using a VTL, then you should buy the VTL license. Their logic is that when you bought your tape drive license, you didn't buy a license to back up to virtual tape, you bought a license to back up to real tape. Now that you're backing up to virtual tape, you need to buy the virtual tape license. They also just had a major change in how they price things that you should look into. They announced it at vision, but I forgot it already. ;) As I recall, the whole server can be capacity based now. FWIW, I like capacity-based pricing. --- W. Curtis Preston Backup Blog @ www.backupcentral.com VP Data Protection, GlassHouse Technologies -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Paul Keating Sent: Monday, July 30, 2007 7:42 AM To: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Has anyone done benchmarking of any sort with.. Depends which licensing model you're following. If you have an Enterprise site license, fine. I think Jon and I are both licensing the robots and a per drive scheme, in which case we're both paying for our VTLs. However, since you can create a ridiculous number of virtual drives on a VTL it makes more sense to license by provisioned storage in almost any case, then you can have as many drives as you want. Paul -- -Original Message- From: Kevin Whittaker [mailto:[EMAIL PROTECTED] Sent: July 30, 2007 10:18 AM To: Dyck, Jonathan; Paul Keating; Justin Piszcz Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. Well, actually if your VTL masks itself as a regular robot... Licensing is non-existent. Because the EMC CDL/EDL that I have can look like an L700E and I already have the license for a robot like this, and I have enterprise licenses then I had to pay nothing extra thru Veritas. Kevin -Original Message- From: Dyck, Jonathan [mailto:[EMAIL PROTECTED] Sent: Monday, July 30, 2007 10:01 AM To: Paul Keating; Justin Piszcz; Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. Hey Paul, Have you actually tried to implement this before (same STU being used with multiple policies, two (or more) different MPX levels)? Yours is the same theory I figured on, being that the lowest MPX number wins out, but I've observed differently when actually implementing it (NBU 5.1 MP5). When I spat the failure (the fact that lower levels of MPX were obviously writing to higher levels MPX streams) out to support, they basically said I had 3 options: 1) different volume pools for the different MPX policies, 2) different retentions for the different MPX policies, 3) create separate STUs for the different MPX policies. Question on VTLs, the only reason I could see you wanting to MPX anything on them would be to save $$. Symantec has two options (I think?) license per virtual drive or license per TB used last I checked into it. Any technical reason anyone can think of to MPX data onto a VTL? Cheers, Jon -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Paul Keating Sent: Friday, July 27, 2007 9:24 AM To: Justin Piszcz; Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Has anyone done benchmarking of any sort with.. Yes they do. :o) I think what kevin was getting at, and what you were asking about, is: If you set the MPX=1 in the schedule, regardless of the MPX level of the STU, the lowest number is the one that is observed. F'rinstance, you've got 2 jobs runningschedule and STU MPX=4, so effectively, you could start two more jobs and they would start streaming to the same tape. If you start a single job in another policy with MPX=1, even though there are 2 streams available on that drive, the MPX=1 job will wait untill it can have sole use of the drive before it starts, as the sched limits MPX=1. Paul -- -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Justin Piszcz Sent: July 27, 2007 9:02 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Has anyone done
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
It's also possible, I believe, I'd have to check the licenses, to license VTL as VTL by virtual drive. Not according to Symantec. That's the point _I'm_ making. ;) ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
I'll back this up, too. We're going to install a couple large VTLs this year, and Symantec wants to license only by size. For the size we are getting, it would be cheaper to license by size only if we were going to configure over 100 virtual drives on each one. I blasted them over their VTL license as the value of the VTL comes from the VTL itself, not the extortion fee you have to pay Symantec to permit you to use it with NetBackup. I suggested that they over both options - license by size or license by number of drives - whichever is most cost effective for the customer. Jack L. Forester, Jr. UNIX Systems Administrator, Stf Lockheed Martin Information Technology (304) 625-3946 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Curtis Preston Sent: Tuesday, July 31, 2007 11:20 AM To: Paul Keating; veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Has anyone done benchmarking of any sort with.. It's also possible, I believe, I'd have to check the licenses, to license VTL as VTL by virtual drive. Not according to Symantec. That's the point _I'm_ making. ;) ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
VTLs seem nice in this regard then :) Justin. On Fri, 27 Jul 2007, Kevin Whittaker wrote: Ah... Yes I see where you are going But my way will just put the image on the tape, no multiplexing at all. I have a Virtual Tape Library, and I am in the process of creating 5 robots with 20 drives each. To avoid multiplexing you have to dedicated a drive to each channel, and that is what setting the MPX=1 does. When you are limited on tape drive resources You can not do this. I used to just have my L700 with 11 9940B drives, and could have never considered this with the fact that I backup over 35 databases. But now, with the VTL I can configure as many drives as I want to configure. So to answer the question, un multiplexing will commit 1 drive to each channel. You can not multiplex different backups onto the same drive at the same time. Kevin -Original Message- From: Justin Piszcz [mailto:[EMAIL PROTECTED] Sent: Friday, July 27, 2007 8:46 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. On Fri, 27 Jul 2007, Kevin Whittaker wrote: Oh... Sorry about that. That is the easy part. Under the Schedule, within the policy, I changed the media multiplexing to 1. The RMAN script starts 4 channels, and sees the schedule says only 1 multiplex then it sends them to different tapes. This is good in a sense, but say you have 10 tape drives for example. If you are multiplexing other backups with the DB backups, does the RMAN policy write with LEVEL 1 multiplexing to separatae tapes and the other backups still take place for the other (assume MPX=4) 3 levels of multiplexing on the drive? Or does the MPX=1 job utilize the entire tape drive? Does it make sense what I am mentioning? Essentially you'd want separate tape media IDs for each server but you still want other servers writing for the other 3 levels of MPX on that tape. Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
On Fri, 27 Jul 2007, Kevin Whittaker wrote: Oh... Sorry about that. That is the easy part. Under the Schedule, within the policy, I changed the media multiplexing to 1. The RMAN script starts 4 channels, and sees the schedule says only 1 multiplex then it sends them to different tapes. This is good in a sense, but say you have 10 tape drives for example. If you are multiplexing other backups with the DB backups, does the RMAN policy write with LEVEL 1 multiplexing to separatae tapes and the other backups still take place for the other (assume MPX=4) 3 levels of multiplexing on the drive? Or does the MPX=1 job utilize the entire tape drive? Does it make sense what I am mentioning? Essentially you'd want separate tape media IDs for each server but you still want other servers writing for the other 3 levels of MPX on that tape. Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
Ah... Yes I see where you are going But my way will just put the image on the tape, no multiplexing at all. I have a Virtual Tape Library, and I am in the process of creating 5 robots with 20 drives each. To avoid multiplexing you have to dedicated a drive to each channel, and that is what setting the MPX=1 does. When you are limited on tape drive resources You can not do this. I used to just have my L700 with 11 9940B drives, and could have never considered this with the fact that I backup over 35 databases. But now, with the VTL I can configure as many drives as I want to configure. So to answer the question, un multiplexing will commit 1 drive to each channel. You can not multiplex different backups onto the same drive at the same time. Kevin -Original Message- From: Justin Piszcz [mailto:[EMAIL PROTECTED] Sent: Friday, July 27, 2007 8:46 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. On Fri, 27 Jul 2007, Kevin Whittaker wrote: Oh... Sorry about that. That is the easy part. Under the Schedule, within the policy, I changed the media multiplexing to 1. The RMAN script starts 4 channels, and sees the schedule says only 1 multiplex then it sends them to different tapes. This is good in a sense, but say you have 10 tape drives for example. If you are multiplexing other backups with the DB backups, does the RMAN policy write with LEVEL 1 multiplexing to separatae tapes and the other backups still take place for the other (assume MPX=4) 3 levels of multiplexing on the drive? Or does the MPX=1 job utilize the entire tape drive? Does it make sense what I am mentioning? Essentially you'd want separate tape media IDs for each server but you still want other servers writing for the other 3 levels of MPX on that tape. Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
Oh... Sorry about that. That is the easy part. Under the Schedule, within the policy, I changed the media multiplexing to 1. The RMAN script starts 4 channels, and sees the schedule says only 1 multiplex then it sends them to different tapes. -Original Message- From: Justin Piszcz [mailto:[EMAIL PROTECTED] Sent: Friday, July 27, 2007 8:35 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. Yeah but specifically how did you get NetBackup to backup a separate stream to each tape drive as that's the kicker? ;) On Fri, 27 Jul 2007, Kevin Whittaker wrote: Oh, I had a spare 240 available, so I loaded ORACLE 10/Solaris 10 on it. I had a DBA created a database with 108GB of data in it. Then we did a Level 0 (full) backup, that took 39 mins. Then we destroyed the database and restored it. Then we just kept doing it back and forth, testing different ways to do the backup/restore. Kevin -Original Message- From: Justin Piszcz [mailto:[EMAIL PROTECTED] Sent: Friday, July 27, 2007 8:24 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. On Fri, 27 Jul 2007, Kevin Whittaker wrote: Recently I did quite a bit of benchmarking on restores with ORACLE RMAN multiplexing. We currently do 4 channels to 1 tape. Our test backup was taking 39 mins, with a full restore time of 2 hours. And we tested other channels, and it seems that the restore time is # of channels times the backup time. So my restores are 4 times a long as the backup. Well, we also tested sending each channel to different tape drives. Well the same backup was restored in 38 mins! How did you accomplish this? So, I am in the process of modifying my VTL to have enough tape drives so that we can un multiplex all of our backups. Also I will add vaulting a multiplex image takes longer than one that is not multiplex. And yes, my vault profile is set to keep multiplexing. Kevin -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Justin Piszcz Sent: Friday, July 27, 2007 6:50 AM To: veritas-bu@mailman.eng.auburn.edu Subject: [Veritas-bu] Has anyone done benchmarking of any sort with.. The delay on Multiplexed restores? Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
Yes they do. :o) I think what kevin was getting at, and what you were asking about, is: If you set the MPX=1 in the schedule, regardless of the MPX level of the STU, the lowest number is the one that is observed. F'rinstance, you've got 2 jobs runningschedule and STU MPX=4, so effectively, you could start two more jobs and they would start streaming to the same tape. If you start a single job in another policy with MPX=1, even though there are 2 streams available on that drive, the MPX=1 job will wait untill it can have sole use of the drive before it starts, as the sched limits MPX=1. Paul -- -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Justin Piszcz Sent: July 27, 2007 9:02 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: Re: [Veritas-bu] Has anyone done benchmarking of any sort with.. VTLs seem nice in this regard then :) Justin. La version française suit le texte anglais. This email may contain privileged and/or confidential information, and the Bank of Canada does not waive any related rights. Any distribution, use, or copying of this email or the information it contains by other than the intended recipient is unauthorized. If you received this email in error please delete it immediately from your system and notify the sender promptly by email that you have done so. Le présent courriel peut contenir de l'information privilégiée ou confidentielle. La Banque du Canada ne renonce pas aux droits qui s'y rapportent. Toute diffusion, utilisation ou copie de ce courriel ou des renseignements qu'il contient par une personne autre que le ou les destinataires désignés est interdite. Si vous recevez ce courriel par erreur, veuillez le supprimer immédiatement et envoyer sans délai à l'expéditeur un message électronique pour l'aviser que vous avez éliminé de votre ordinateur toute copie du courriel reçu. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
Yeah but specifically how did you get NetBackup to backup a separate stream to each tape drive as that's the kicker? ;) On Fri, 27 Jul 2007, Kevin Whittaker wrote: Oh, I had a spare 240 available, so I loaded ORACLE 10/Solaris 10 on it. I had a DBA created a database with 108GB of data in it. Then we did a Level 0 (full) backup, that took 39 mins. Then we destroyed the database and restored it. Then we just kept doing it back and forth, testing different ways to do the backup/restore. Kevin -Original Message- From: Justin Piszcz [mailto:[EMAIL PROTECTED] Sent: Friday, July 27, 2007 8:24 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. On Fri, 27 Jul 2007, Kevin Whittaker wrote: Recently I did quite a bit of benchmarking on restores with ORACLE RMAN multiplexing. We currently do 4 channels to 1 tape. Our test backup was taking 39 mins, with a full restore time of 2 hours. And we tested other channels, and it seems that the restore time is # of channels times the backup time. So my restores are 4 times a long as the backup. Well, we also tested sending each channel to different tape drives. Well the same backup was restored in 38 mins! How did you accomplish this? So, I am in the process of modifying my VTL to have enough tape drives so that we can un multiplex all of our backups. Also I will add vaulting a multiplex image takes longer than one that is not multiplex. And yes, my vault profile is set to keep multiplexing. Kevin -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Justin Piszcz Sent: Friday, July 27, 2007 6:50 AM To: veritas-bu@mailman.eng.auburn.edu Subject: [Veritas-bu] Has anyone done benchmarking of any sort with.. The delay on Multiplexed restores? Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
Oh, I had a spare 240 available, so I loaded ORACLE 10/Solaris 10 on it. I had a DBA created a database with 108GB of data in it. Then we did a Level 0 (full) backup, that took 39 mins. Then we destroyed the database and restored it. Then we just kept doing it back and forth, testing different ways to do the backup/restore. Kevin -Original Message- From: Justin Piszcz [mailto:[EMAIL PROTECTED] Sent: Friday, July 27, 2007 8:24 AM To: Kevin Whittaker Cc: veritas-bu@mailman.eng.auburn.edu Subject: RE: [Veritas-bu] Has anyone done benchmarking of any sort with.. On Fri, 27 Jul 2007, Kevin Whittaker wrote: Recently I did quite a bit of benchmarking on restores with ORACLE RMAN multiplexing. We currently do 4 channels to 1 tape. Our test backup was taking 39 mins, with a full restore time of 2 hours. And we tested other channels, and it seems that the restore time is # of channels times the backup time. So my restores are 4 times a long as the backup. Well, we also tested sending each channel to different tape drives. Well the same backup was restored in 38 mins! How did you accomplish this? So, I am in the process of modifying my VTL to have enough tape drives so that we can un multiplex all of our backups. Also I will add vaulting a multiplex image takes longer than one that is not multiplex. And yes, my vault profile is set to keep multiplexing. Kevin -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Justin Piszcz Sent: Friday, July 27, 2007 6:50 AM To: veritas-bu@mailman.eng.auburn.edu Subject: [Veritas-bu] Has anyone done benchmarking of any sort with.. The delay on Multiplexed restores? Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
On Fri, 27 Jul 2007, Kevin Whittaker wrote: Recently I did quite a bit of benchmarking on restores with ORACLE RMAN multiplexing. We currently do 4 channels to 1 tape. Our test backup was taking 39 mins, with a full restore time of 2 hours. And we tested other channels, and it seems that the restore time is # of channels times the backup time. So my restores are 4 times a long as the backup. Well, we also tested sending each channel to different tape drives. Well the same backup was restored in 38 mins! How did you accomplish this? So, I am in the process of modifying my VTL to have enough tape drives so that we can un multiplex all of our backups. Also I will add vaulting a multiplex image takes longer than one that is not multiplex. And yes, my vault profile is set to keep multiplexing. Kevin -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Justin Piszcz Sent: Friday, July 27, 2007 6:50 AM To: veritas-bu@mailman.eng.auburn.edu Subject: [Veritas-bu] Has anyone done benchmarking of any sort with.. The delay on Multiplexed restores? Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
Recently I did quite a bit of benchmarking on restores with ORACLE RMAN multiplexing. We currently do 4 channels to 1 tape. Our test backup was taking 39 mins, with a full restore time of 2 hours. And we tested other channels, and it seems that the restore time is # of channels times the backup time. So my restores are 4 times a long as the backup. Well, we also tested sending each channel to different tape drives. Well the same backup was restored in 38 mins! So, I am in the process of modifying my VTL to have enough tape drives so that we can un multiplex all of our backups. Also I will add vaulting a multiplex image takes longer than one that is not multiplex. And yes, my vault profile is set to keep multiplexing. Kevin -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Justin Piszcz Sent: Friday, July 27, 2007 6:50 AM To: veritas-bu@mailman.eng.auburn.edu Subject: [Veritas-bu] Has anyone done benchmarking of any sort with.. The delay on Multiplexed restores? Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
[Veritas-bu] Has anyone done benchmarking of any sort with..
The delay on Multiplexed restores? Justin. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Re: [Veritas-bu] Has anyone done benchmarking of any sort with..
The delay on Multiplexed restores? Benchmarking could be done for your setup, but there's no general answer. If you have slow tapes and fast clients, then multiplexing will tend to slow restores as the clients will be waiting on the drive to feed data. If you have slow clients and fast drives, then multiplexing will tend to have no impact on restore time, because making the drives go faster on restore will not be useful (you're waiting on the client, not the drive). -- Darren Dunham [EMAIL PROTECTED] Senior Technical Consultant TAOShttp://www.taos.com/ Got some Dr Pepper? San Francisco, CA bay area This line left intentionally blank to confuse you. ___ Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu