Hi all,
We're about o refresh our Netapps and move to OnTap 7.0.1 and back them up with Legato Networker 7.2, and hopefully NDMP, though we currently do NFS mounts to the (Solaris 5.8, E450) backup server.
We'd like to setup one large aggregate, with a small root FlexVol and a large large large data FlexVol, with a bunch of qtrees inside the flexvol.
So what are others doing to make their backups run quicky, and their restores to run reliably? I've got the release notes for Networker 7.2 and I see the limitation of one (1) million files in DAR restores, and since I know our volumes have lots more files than that... it could be a problem.
But maybe if we broke down our backups on a per-qtree basis, that would work?
What are other people doing for backups? Any gotchas or performance tuning hints would be appreciaited.
Thanks, John John Stoffel - Senior Staff Systems Administrator - System LSI Group Toshiba America Electronic Components, Inc. - http://www.toshiba.com/taec john.stoffel@taec.toshiba.com - 508-486-1087
We're using Legato and NDMP to backup our filers. While we aren't on 7G yet, we do have a couple of "large, large, large" data volumes. To work around the DAR limitation of 1 MM files, we use the qtree method you propose below. All data is stored in qtrees and each qtree is listed individually on the client in Legato. It creates a little more management overhead in that we have to be sure that all new top-level directories are added as qtrees (we've had a few sneak in that weren't) and add them to Legato, but it works well for us.
We're looking forward to Legato 7.2 to allow us to do multiple NDMP streams to a single tape device. Our filers backup to an Overland Neo 4100 with 4 x LTO-2 drives and we can't push the drives as fast as we'd like to.
Jeff Mery - MCSE, MCP National Instruments
------------------------------------------------------------------------- "Allow me to extol the virtues of the Net Fairy, and of all the fantastic dorks that make the nice packets go from here to there. Amen." TB - Penny Arcade -------------------------------------------------------------------------
"John Stoffel" john.stoffel@taec.toshiba.com Sent by: owner-toasters@mathworks.com 05/05/2005 03:52 PM
To toasters@mathworks.com cc
Subject OnTap 7.0.1 and Legato Network 7.2
Hi all,
We're about o refresh our Netapps and move to OnTap 7.0.1 and back them up with Legato Networker 7.2, and hopefully NDMP, though we currently do NFS mounts to the (Solaris 5.8, E450) backup server.
We'd like to setup one large aggregate, with a small root FlexVol and a large large large data FlexVol, with a bunch of qtrees inside the flexvol.
So what are others doing to make their backups run quicky, and their restores to run reliably? I've got the release notes for Networker 7.2 and I see the limitation of one (1) million files in DAR restores, and since I know our volumes have lots more files than that... it could be a problem.
But maybe if we broke down our backups on a per-qtree basis, that would work?
What are other people doing for backups? Any gotchas or performance tuning hints would be appreciaited.
Thanks, John John Stoffel - Senior Staff Systems Administrator - System LSI Group Toshiba America Electronic Components, Inc. - http://www.toshiba.com/taec john.stoffel@taec.toshiba.com - 508-486-1087
jeff> We're using Legato and NDMP to backup our filers. While we jeff> aren't on 7G yet, we do have a couple of "large, large, large" jeff> data volumes.
Can you define large? I've got a 3.5Tb volume we have to backup, and it takes forever.
jeff> To work around the DAR limitation of 1 MM files, we use the jeff> qtree method you propose below. All data is stored in qtrees jeff> and each qtree is listed individually on the client in Legato. jeff> It creates a little more management overhead in that we have to jeff> be sure that all new top-level directories are added as qtrees jeff> (we've had a few sneak in that weren't) and add them to Legato, jeff> but it works well for us.
I think this is the step we're going to take. We really want the flexibilty of FlexVols in 7G, along with lots of qtrees, we should be ok.
jeff> We're looking forward to Legato 7.2 to allow us to do multiple jeff> NDMP streams to a single tape device. Our filers backup to an jeff> Overland Neo 4100 with 4 x LTO-2 drives and we can't push the jeff> drives as fast as we'd like to.
We've just upgraded all our sites to Legato Networker 7.2, now we need to do some upgrades to 7G and start testing out NDMP backups and see how well they work out.
We have a small number of SuperDLT drives at each site, so not being able to Multiplex multiple NDMP data streams to a single tape drive was a really problem for us. Hopefully we won't run into too many problems with this in Networker 7.2.
Are any people out there planning on setting up filedevices for their NDMP backups, and then cloning them to tape? That seems like a good way to do it, esp if you have a big enough disk farm. It might only work well for incrementals though, not for Fulls.
John John Stoffel - Senior Staff Systems Administrator - System LSI Group Toshiba America Electronic Components, Inc. - http://www.toshiba.com/taec john.stoffel@taec.toshiba.com - 508-486-1087
jeff> We're using Legato and NDMP to backup our filers. While we jeff> aren't on 7G yet, we do have a couple of "large, large, large" jeff> data volumes.
Can you define large? I've got a 3.5Tb volume we have to backup, and it takes forever.
We have several large file systems based on size and/or file count. One vol is 3.5 TB with 9.5 MM files, one is 1TB with 7.4 MM files, one is 2.6 TB with 5.9 MM files. We have a few other volumes as well, but don't consider them to be "large". Fulls take about 36 hours on that 3.5TB guy (only volume on that filer). Fulls on the other volumes take about 18 hours (other volumes both on 2nd filer). We do three-way backups to the Overland library which is FC-attached to one host.
jeff> To work around the DAR limitation of 1 MM files, we use the jeff> qtree method you propose below. All data is stored in qtrees jeff> and each qtree is listed individually on the client in Legato. jeff> It creates a little more management overhead in that we have to jeff> be sure that all new top-level directories are added as qtrees jeff> (we've had a few sneak in that weren't) and add them to Legato, jeff> but it works well for us.
I think this is the step we're going to take. We really want the flexibilty of FlexVols in 7G, along with lots of qtrees, we should be ok.
jeff> We're looking forward to Legato 7.2 to allow us to do multiple jeff> NDMP streams to a single tape device. Our filers backup to an jeff> Overland Neo 4100 with 4 x LTO-2 drives and we can't push the jeff> drives as fast as we'd like to.
We've just upgraded all our sites to Legato Networker 7.2, now we need to do some upgrades to 7G and start testing out NDMP backups and see how well they work out.
We have a small number of SuperDLT drives at each site, so not being able to Multiplex multiple NDMP data streams to a single tape drive was a really problem for us. Hopefully we won't run into too many problems with this in Networker 7.2.
Are any people out there planning on setting up filedevices for their NDMP backups, and then cloning them to tape? That seems like a good way to do it, esp if you have a big enough disk farm. It might only work well for incrementals though, not for Fulls.
File devices are an intersting idea. Unfortunately we don't currently have the space to do that. We are investigating disk-to-disk and virtual tape backups which may offer up some alternatives, but we're not far enough along there to provide useful comments.
Jeff Mery - MCSE, MCP National Instruments
------------------------------------------------------------------------- "Allow me to extol the virtues of the Net Fairy, and of all the fantastic dorks that make the nice packets go from here to there. Amen." TB - Penny Arcade -------------------------------------------------------------------------
Just throwing out an option for those of you looking for a OnTap 6.x to 7.x migration strategy. I am not going to try to do a sales job. Just offering a solution.
I have been told by a couple NetApp guys that if you upgrade your filers from 6.x to 7.x, everything currently in place will work, you just won't get flexvol support, right? Any new volumes you create could be a flexvol.
RainStorage gives you the ability to perform transparent data moves from CIFS or NFS source to a new destination, within the same filer or to a new filer, while clients have full read-write access. We retain all original date and time stamps, handle all open and locked files and, if desired, can perform full SID/ACL translation if you want to move data between Windows authentication or resource domains, such as an NT4 to Active Directory migration.
We can integrate into your global name space (DFS for CIFS/Automount for NFS if you have them) so that we can create, update or modify existing resources. We can also tell you who all the users are accessing the source and confirm that they have all been migrated over to the destination. If you don't have a global name space, we can, if given the ability to do so, interact with your clients and based upon configurable idle time, force them to unmap/unmount and/or remap/remount to the new destination, again reporting who is still using the source and who has migrated to the destination.
We also automate the process of creating the destination qtrees, exports and shares and perform those preparation steps, including assigning and activating quotas on the destination as part of the move process.
We can eliminate the downtime and extended maintenance windows associated with the migration as well as consolidations, capacity, performance and tiered storage management.
This is a very small part of the functionality delivered by RainStorage.
See the website for more info. Or don't be afraid to ask me directly.
E.
John, et al,
Be aware - Legato (EMC) does not yet support DOT 7.0 - we have been told that DOT 7.0 with Legato NW 7.1.3 is slated for certification Q3 or Q4 of this year. DOT 7.0 with NW 7.2 may be sooner. No real commitment.
Go figure.
Later John
----- Original Message ----- From: "John Stoffel" john.stoffel@taec.toshiba.com To: toasters@mathworks.com Sent: Thursday, May 05, 2005 3:52 PM Subject: OnTap 7.0.1 and Legato Network 7.2
Hi all,
We're about o refresh our Netapps and move to OnTap 7.0.1 and back them up with Legato Networker 7.2, and hopefully NDMP, though we currently do NFS mounts to the (Solaris 5.8, E450) backup server.
We'd like to setup one large aggregate, with a small root FlexVol and a large large large data FlexVol, with a bunch of qtrees inside the flexvol.
So what are others doing to make their backups run quicky, and their restores to run reliably? I've got the release notes for Networker 7.2 and I see the limitation of one (1) million files in DAR restores, and since I know our volumes have lots more files than that... it could be a problem.
But maybe if we broke down our backups on a per-qtree basis, that would work?
What are other people doing for backups? Any gotchas or performance tuning hints would be appreciaited.
Thanks, John John Stoffel - Senior Staff Systems Administrator - System LSI Group Toshiba America Electronic Components, Inc. -
john.stoffel@taec.toshiba.com - 508-486-1087
John> Be aware - Legato (EMC) does not yet support DOT 7.0 - we have John> been told that DOT 7.0 with Legato NW 7.1.3 is slated for John> certification Q3 or Q4 of this year. DOT 7.0 with NW 7.2 may be John> sooner. No real commitment.
That's a pretty sucky thing to remember. We need to go with Data OnTap 7.0.1 for the FlexVols, it's the only way we can move forward. Since we're currently doing NFS backups, we could just stick with that until support becomes official, but we'd like to move to NDMP if we can, just to get performance improvements.
Thanks for the reminder, I'm sure we'll all be pushing EMC/Legato to support DOT 7.0.1 ASAP.
John John Stoffel - Senior Staff Systems Administrator - System LSI Group Toshiba America Electronic Components, Inc. - http://www.toshiba.com/taec john.stoffel@taec.toshiba.com - 508-486-1087