9 replies
by tomber
We use dlt tapes on a bi-weekly basis.
Everyday at lunch we do a differential and every night we do a full backup.
This way we have a 2 week window to fall back on. This is not only in order to save data but also to be able to backtrack if misuse as occured.
by knannest122
A backup strategy be created by the need for restore in a normal or disaster recovery situation.
Tapes should be stored securely outside the computer rooms for the servers that are backed up.
The restores has to be tested at least twice a year for each server, and tested for each operating system upgrade, database upgrade and backup software upgrade. This also implies testing
the restore times and update the DR plan.
Backups has to be stored according to the
demands from the company and tax laws.
This implies storage of tapes in 10 years.
Tapes has to be tested and if necessary copied to new tapes. Monthly backups has to be duplicated and one copy stored outside the tape library.
Each weekly backup has to be moved outside the library, and inserted the next week.
by tturner33
We are in the process of moving from DLT storage to VXA-2. We recently purchased an autoloader from Exabyte w/1.6T capacity. Their promotional materials boast "Two weeks of unattended backups", however, I can seem to grasp the safety in that. I was taught to change tapes daily and take them off-site. Can some point me to a resource for the rationalization of changing tapes every two weeks? (Other than Exabyte's sales pitch!)
In the event of disaster, it would cripple my company to have to re-create a week or more worth the work. Any help is appreciated.
by jadium
We are currently in the process of evaluating an off-site disk-based backup solution. As I understand it, performing backups in this manner will alleviate many of the primary concerns with current Tape Backups. Disk backups are more reliable, and space is cheap. The only major downfall is the speed of the connections. If we could use gigabit ethernet I would, but I am looking to use fiber between two buildings, so this should be fast enough.
-J
by dvanbeve
As a Veritas Consultant concentrating on Netbackup, I see many DR strategies. Some good and some bad. The idea of doing backups across sites is a good one assuming the sites are far enough away. As mentioned by someone else, fire is one concern if buildings are too close. Consider a regional disaster where blocks or miles are evacuated. Without access to the datacenter your plan is shot.
I have implemented a solution with a customer where they do have 10 miles of fibre to a second site. Backups from site being written to the other. A node of a clustered Netbackup master at each site, ready to take over at a moments notice. They still offsite their backups to a secured vault.
Using disk as a cache for your backups is a great idea. It will help reduce the need for too many tape drives to complete your backups in a given backup window. It?s not meant to be a all encompassing DR solution. Well, maybe when someone comes up with a library that will eject hard drives.
I didn't look up the model of equipment that you are using but this should apply.
My last job had a tape library. We made backup tapes that were intended to remain in the library. Backup jobs ran at different times and there was idle time between backups. We were using Veritas Netbackup Datacenter software and the tape library had six tape drives, so it was pretty sweet. Anyway, I had a script that would run and when one batch of backups was finishing the script would instruct the Veritas software to make copies of the backup tapes. The way that Veritas works allowed me to put a certain number of tapes in a 'pool' that would be sent off site. That way all of the tapes in this pool would have backups made on the same date and all of the backups would expire at the same time making them available for Veritas to use again. Once the backups were copied to this second set of tapes designated for off site storage the Veritas software would eject the tapes. In the morning I would just pick up the stack of off site tapes from the tape library output box and put them into an Iron Mountain storage box with appropriate documentation. The Veritas software also created the documentation for each box. This was all done using the command line functions of Veritas in scripts. The main backup jobs were managed within the Veritas GUI job management interface. Overall it was great. That was the first time that I had used a tape library and I loved it. If a tape was bad the Veritas software would just eject it from the tape drive and replace it with another tape without any loss of backups. Using Veritas Netbackup Datacenter in this way with a tape library is the best way to approach mission critical backups. You have protection from bad tapes ruining your backups as mentioned above. You can make duplicate tapes of your backups to keep off site while keeping the principle backup tapes on site. I can't say enough good things about it. Mind you there were always little problems that came up with the Veritas software or the tape library hardware. I found the Veritas telephone support along with the Veritas user documentation to be excellent. If I couldn't fix some hiccup by looking up the error code in the documentation then the telephone support people usually did a good job helping me. I spent a lot of time initially writing these scripts to duplicate backup tapes because my scripts could do a better job of it than the Netbackup software, but once the scripts were written and working the whole thing was a blessing.
So, overall I believe very highly in using a tape library with intelligent backup software. My experience using Veritas is very good. I highly recommend the Veritas Netbackup product line. It has its quirks but you learn to work with it. For instance the first time that a backup failed because Netbackup said that there were no free tapes even though it looks to you like there are plenty of free tapes is a bitch. I learned to manually expire the backups on the oldest tapes every day.
If your backups are important to you, and if you want to make duplicates of your backups to store off site, then a robotic tape library and a good backup software solution are absolutely critical.
by advancedgeek
by lindfx
by knannest122
Generally, servers should always be stored in more
than one place, and servers shold never be backed
up to local tape libraries.
What is the acceptable down time in a disaster situation.
How many days will it take for your hardware vendor to install new SAN/disks, new backup servers, new backup library, restore the data, install routines. configure network equipment?
When are you allowed to enter a building after a fire? Is the safe where the tapes are stored in
a room with little or no materials that will burn?
Always try to install servers equally in two locations, and the tape library in a third location. It the servers are not clustered,
use test servers as 'standby' production servers.
This discussion is closed to new replies.
Trending Storage posts
Sort by Trending