Hi All,
I have a Dell TL2000 library with 2 LTO-6 drives and one file server that has 9tb of data. Can I backup the full VBK file to tape and use both tapes to get 320 megabytes per second? I need to find a way to backup this server faster as 9tb takes too long today and would take too long to restore.
GFS media pool Options Enable parallel processing for tape jobs
Thanks,
Joe
-
- Enthusiast
- Posts: 56
- Liked: 2 times
- Joined: Nov 10, 2020 8:07 pm
- Full Name: Joe G
- Contact:
-
- Product Manager
- Posts: 9846
- Liked: 2605 times
- Joined: May 13, 2017 4:51 pm
- Full Name: Fabian K.
- Location: Switzerland
- Contact:
Re: 9tb file server to dual LTO-6
Parallel processing doesn‘t work for a single machine or vm.
There are two „parallel processing“ methods:
- parallel processing of different tape jobs
- parallel process of different source backup chains
A single vbk file will always be using a single tape drive.
If you want to leverage both drives simultaneously, you need to split your file server into multiple servers with „per machine backup chains“. Using „per machine backup files“ will also help to reduce recovery time from tape, because not the entire data needs to be copied back to the backup repository in case of a Guest file restore. Only the vm, from which you need todo the restore.
Or if you have multiple drives inside the fileserver, create an additional backup jobs for a few of this drives. Not really recommended, but it will work for parallel processing.
Veeam - Parallel Processing of Tapes
There are two „parallel processing“ methods:
- parallel processing of different tape jobs
- parallel process of different source backup chains
A single vbk file will always be using a single tape drive.
If you want to leverage both drives simultaneously, you need to split your file server into multiple servers with „per machine backup chains“. Using „per machine backup files“ will also help to reduce recovery time from tape, because not the entire data needs to be copied back to the backup repository in case of a Guest file restore. Only the vm, from which you need todo the restore.
Or if you have multiple drives inside the fileserver, create an additional backup jobs for a few of this drives. Not really recommended, but it will work for parallel processing.
Veeam - Parallel Processing of Tapes
Product Management Analyst @ Veeam Software
-
- Veteran
- Posts: 643
- Liked: 312 times
- Joined: Aug 04, 2019 2:57 pm
- Full Name: Harvey
- Contact:
Re: 9tb file server to dual LTO-6
Hey Joe,
Out of curiosity, what sort of RPO are you expecting on 9 TiB for a tape archive? 17 hours for the backup on a single LTO6 drive isn't really that out of the ordinary, and it's pretty common (or used to be anyways) to just start the tape backups on the weekends and make sure the box to "Prevent the source job from interrupting the tape job" is checked. (This won't stop your source job, it just makes it "wait" for the tape job to finish)
As for restores, I guess I again have to ask what your expectation is; Tape is meant to be archival and spared for those "oh fsck" moments when every other option has been exhausted for restoring the data. I wouldn't be anticipating "fast" restores from tape especially since at max you can probably expect close to 100 MiB/s from LTO6 (under absolutely ideal conditions, you can get 150 MB/s) so really you're looking at a day in and a day out for LTO 6 no matter how you do it.
If you truly want to consider workarounds, you can archive the backup file further with something like 7zip and split those archives (be sure to just use "store" compression, don't actually try to compress it further) and at least that way you can stream from two drives with file to tape, but keep in mind this increases the complexity of your restores as now you need to do a lot more manual tracking of the individual files you backup and making sure you recover the correct files from tapes. To stream to two drives would mean two file to tape jobs, and frankly I think this just gets very messy and potentially introduces even __longer__ restore times if you make a mistake and get the wrong tapes/files.
Out of curiosity, what sort of RPO are you expecting on 9 TiB for a tape archive? 17 hours for the backup on a single LTO6 drive isn't really that out of the ordinary, and it's pretty common (or used to be anyways) to just start the tape backups on the weekends and make sure the box to "Prevent the source job from interrupting the tape job" is checked. (This won't stop your source job, it just makes it "wait" for the tape job to finish)
As for restores, I guess I again have to ask what your expectation is; Tape is meant to be archival and spared for those "oh fsck" moments when every other option has been exhausted for restoring the data. I wouldn't be anticipating "fast" restores from tape especially since at max you can probably expect close to 100 MiB/s from LTO6 (under absolutely ideal conditions, you can get 150 MB/s) so really you're looking at a day in and a day out for LTO 6 no matter how you do it.
If you truly want to consider workarounds, you can archive the backup file further with something like 7zip and split those archives (be sure to just use "store" compression, don't actually try to compress it further) and at least that way you can stream from two drives with file to tape, but keep in mind this increases the complexity of your restores as now you need to do a lot more manual tracking of the individual files you backup and making sure you recover the correct files from tapes. To stream to two drives would mean two file to tape jobs, and frankly I think this just gets very messy and potentially introduces even __longer__ restore times if you make a mistake and get the wrong tapes/files.
Who is online
Users browsing this forum: No registered users and 28 guests