LinuxQuestions.org
Share your knowledge at the LQ Wiki.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Newbie
User Name
Password
Linux - Newbie This Linux forum is for members that are new to Linux.
Just starting out and have a question? If it is not in the man pages or the how-to's this is the place!

Notices


Reply
  Search this Thread
Old 01-03-2012, 10:29 AM   #1
jorran
Member
 
Registered: Dec 2011
Location: Iowa
Distribution: RHEL 5.X
Posts: 54

Rep: Reputation: Disabled
Back Up


What is the easiest way to back up all our data for our Linux Clusters. Right now it takes almost 11 days to back ALL of our data up, yes we have a lot but there should be a faster way to do this? I am not exactly sure how we are doing it now but when I find out the exact procedure I will post it. Any suggestions will help.

Thanks
 
Old 01-03-2012, 12:31 PM   #2
unSpawn
Moderator
 
Registered: May 2001
Posts: 29,335
Blog Entries: 55

Rep: Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535Reputation: 3535
Quote:
Originally Posted by jorran View Post
I am not exactly sure how we are doing it now but when I find out the exact procedure I will post it.
Finding out IMHO should be the first step. (To paraphrase whats-his-name from Blackhawk Down: nothing takes eleven days ;-p) How can anyone offer suggestions if you don't know what's in use? Also please try to be factual instead of saying stuff like "all our data" (we don't have a clue what the data comprises of) or "our Linux Clusters" (we don't have a clue how many sets of machines in whatever locations using whatever fat pipes you're talking about).
 
Old 01-03-2012, 12:33 PM   #3
MensaWater
LQ Guru
 
Registered: May 2005
Location: Atlanta Georgia USA
Distribution: Redhat (RHEL), CentOS, Fedora, Debian, FreeBSD, HP-UX, Solaris, SCO
Posts: 6,713
Blog Entries: 14

Rep: Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038
You'll need to post details to get good responses. A lot of data to mean means TB but to you it might mean MB and the answers I'd give would depend on what "lot" means. If you're doing backups by simply copying to NFS mounts without compression or deduplication over a 10 MB NIC it would be incredibly slow. On the other hand if you're doing direct backup to fibre tape drives via a fibre SAN and they are relatively new (e.g. LTO5) then it would be incredibly fast. There are far too many variables to provide you with good idea.

So as you said you will - post what you're doing to "backup" now. Also post key details such as:
1) How much data exactly are we talking about?
2) Is it a few large files or thousands of small files?
3) What is your network topology (e.g. 10 MB, 100 MB, GigE etc...)?
4) Are you doing any sort of compression before sending the data to backup storage?
5) What is the backup storage (e.g. NFS mounts, tape drives, deduplication appliances)?
6) Are you using any backup application (e.g. NetBackup, Bacula) or tool (e.g. rsync, scp)?
7) What kind of storage is the original on (e.g. internal hard drives in software raid, hardware raid, SAN disks, NFS disks)?
8) Are you scheduling the backup to run at a time when most processes are quiescent?
9) Are you always doing a full backup? Have you thought about staggering it so that the OS is backed up on one day, the application binaries on a different day and the database files on yet another?
10) What is your plan for restoring the data if necessary? This is a key consideration many people forget about when planning backup strategies. It may not do you any good to backup all your data in 10 minutes if it takes two weeks to restore.
 
Old 01-04-2012, 09:11 AM   #4
jorran
Member
 
Registered: Dec 2011
Location: Iowa
Distribution: RHEL 5.X
Posts: 54

Original Poster
Rep: Reputation: Disabled
Quote:
Originally Posted by MensaWater View Post
You'll need to post details to get good responses. A lot of data to mean means TB but to you it might mean MB and the answers I'd give would depend on what "lot" means. If you're doing backups by simply copying to NFS mounts without compression or deduplication over a 10 MB NIC it would be incredibly slow. On the other hand if you're doing direct backup to fibre tape drives via a fibre SAN and they are relatively new (e.g. LTO5) then it would be incredibly fast. There are far too many variables to provide you with good idea.

So as you said you will - post what you're doing to "backup" now. Also post key details such as:
1) How much data exactly are we talking about?
2) Is it a few large files or thousands of small files?
3) What is your network topology (e.g. 10 MB, 100 MB, GigE etc...)?
4) Are you doing any sort of compression before sending the data to backup storage?
5) What is the backup storage (e.g. NFS mounts, tape drives, deduplication appliances)?
6) Are you using any backup application (e.g. NetBackup, Bacula) or tool (e.g. rsync, scp)?
7) What kind of storage is the original on (e.g. internal hard drives in software raid, hardware raid, SAN disks, NFS disks)?
8) Are you scheduling the backup to run at a time when most processes are quiescent?
9) Are you always doing a full backup? Have you thought about staggering it so that the OS is backed up on one day, the application binaries on a different day and the database files on yet another?
10) What is your plan for restoring the data if necessary? This is a key consideration many people forget about when planning backup strategies. It may not do you any good to backup all your data in 10 minutes if it takes two weeks to restore.
1) How much data exactly are we talking about?
* There is a total of about ~8TB
2) Is it a few large files or thousands of small files?
* Hundreds of Thousands of small files
3) What is your network topology (e.g. 10 MB, 100 MB, GigE etc...)?
* GigE
4) Are you doing any sort of compression before sending the data to backup storage?
* Not from the O.S. perspective.
5) What is the backup storage (e.g. NFS mounts, tape drives, deduplication appliances)?
* Previously we used EMC snap shots to back the data up then the snap shot would be mounted to a NBU media and a backup to tape would be performed. Since we moved the cluster everything has been going straight to tape.
6) Are you using any backup application (e.g. NetBackup, Bacula) or tool (e.g. rsync, scp)?
* NetBackup 6.x.
7) What kind of storage is the original on (e.g. internal hard drives in software raid, hardware raid, SAN disks, NFS disks)?
* Not sure I understand the question could you please restate it?
8) Are you scheduling the backup to run at a time when most processes are quiescent?
* Scheduling it when least processes are running however, it takes 11+ days so all processes run at some point.
9) Are you always doing a full backup? Have you thought about staggering it so that the OS is backed up on one day,
the application binaries on a different day and the database files on yet another?
* Always doing a full backup at this point but may change to staggering it.
10) What is your plan for restoring the data if necessary? This is a key consideration many people forget about when
planning backup strategies. It may not do you any good to backup all your data in 10 minutes if it takes two weeks to restore.
* Still getting details for this question at this point.
 
Old 01-04-2012, 03:22 PM   #5
MensaWater
LQ Guru
 
Registered: May 2005
Location: Atlanta Georgia USA
Distribution: Redhat (RHEL), CentOS, Fedora, Debian, FreeBSD, HP-UX, Solaris, SCO
Posts: 6,713
Blog Entries: 14

Rep: Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038
You said it is going straight to tape. Does this mean the cluster server is a NetBackup media server? If you're using a separate media server then it means you're pushing 8 TB over GigE and that would be a problem.

Thousands of small files are actually slower to backup than several large files so this could be a bottleneck as well.

Are the tape drives attached via fibre or SCSI? What kind of tape drives (e.g. DLT, LTO1, LTO5)? The drive can make quite a difference.

Do you have multiple drives? For large backups setting multiple streams is often what you need to do.

There is a quite active NetBackup email forum you can subscribe to at http://mailman.eng.auburn.edu/mailma...nfo/veritas-bu. Many NBU experts there and even some Symantec folks respond occasionally.
Also there is an online forum that I don't use at Backupcentral.com that is somehow linked into the above mail forum.
 
Old 01-05-2012, 11:00 AM   #6
jorran
Member
 
Registered: Dec 2011
Location: Iowa
Distribution: RHEL 5.X
Posts: 54

Original Poster
Rep: Reputation: Disabled
Are the tape drives attached via fibre or SCSI? Fibre
What kind of tape drives are they? (DLT, LT01, LT05)? LTO5
How many drives are there?4
Are you using a seperate media server to backup?multiple


Our content manager backups have the largest issue with the file system and it is GFS and netbackup doesnt support GFS...
 
Old 01-05-2012, 01:10 PM   #7
MensaWater
LQ Guru
 
Registered: May 2005
Location: Atlanta Georgia USA
Distribution: Redhat (RHEL), CentOS, Fedora, Debian, FreeBSD, HP-UX, Solaris, SCO
Posts: 6,713
Blog Entries: 14

Rep: Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038Reputation: 1038
If NBU doesn't support GFS how are you backing up on the "multiple" media servers? Via NFS mounts to those media servers? If so your bottleneck is almost certainly your GigE network since NFS is a network filesystem. You'd be much better off having fibre cards (i.e. Qlogic) in the host that has the GFS mount and making it a media server so you could backup straight to tape from it. While we don't do GFS here we do have fibre to LTO5 backups done on multi-TB databases from one of our RHEL5 hosts.

I can't find anything at Symantec that suggests GFS is not supported by NetBackup. The following technote suggests that it must be supported otherwise their resolution would simply have been to say they don't support it.
http://www.symantec.com/business/sup...=1325789973995

Some others do say GFS isn't supported by NBU but no one who could point to anything that definitively said so. I did see one mention that said that only Veritas (VxFS) snapshots (i.e. GFS snapshots aren't). Are you backing up snapshots?

If you post to the list I mentioned yesterday you might find someone who has been able to do this.
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
LXer: Back In Time, Automatic Back-Up Tool For Linux LXer Syndicated Linux News 0 12-12-2011 09:21 AM
[SOLVED] how to run 2 bash command back-to-back in bourneShell nass Slackware 7 10-10-2011 05:23 PM
Minicom not working I connected linux machines back-to-back Prabhunath Linux - Networking 2 06-30-2010 04:39 AM
Can't log back in after changing runlevels from init 3 to 1 back to 3 manwichmakesameal Slackware 5 06-29-2009 11:36 AM
Connecting 2 linux computer back to back with cross cable. michalng Linux - Networking 8 02-11-2009 07:00 AM


All times are GMT -5. The time now is 03:16 AM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration