Currently, we have a hodge-podge of different backup strategies. Most of the problem stems from consolidating 9 different sites into a single (primary and backup) datacenter.
Current environment includes 9 IBM Power systems running AIX with their own tape drives in each server, 18 Red Hat Enterprise Linux Systems with 5 tape drives spread between them (4 of the drives are in two robotic libraries), and 37 Windows servers with 14 tape drives between them (most of the servers without tape drives are not currently backed up at all).
Backup software is AIX Sysback for the Unix Servers, and varying versions of Backup Exec from 8.5 to 11d for the Windows servers, with One instance of HP Data Protector and one old scripted tar/cpio/dd backup thrown in for good measure .
Tape hardware includes 9 LTO2 drives, 4 LTO3 Drives (in libraries), 6 8mm Mammoth Drives, 2 8mm Mammoth2 Drives, 1 8MM AIT drive, 3 4mm DDS/DAT drives, and two DLT drives.
Total data is in the realm of 2.2TB currently.
I'm wanting to consolidate it all down to a single standard if possible, though I'll probably still end up with two (the Data Protector install is backing up a mission-critical Oracle RAC database, and I probably won't mess with that right now, though in the future, I'll want the ability to bring it into the centralized backup scheme), for easier administration, fewer tapes to change and take offsite, and easier DR planning (One type of drive available instead of 7 currently).
I am buying the backup solution to support a new VMware project that we're currently implementing, but I'd like to build it big enough to eventually be able to move everything into the fold.
sounds to me like you are prime for "one stop shopping"
we went from AIX servers with their own tape drives (servers that did not have a tape drive were setup to use the drive on another aix server, so we really had to watch the cron jobs to make sure the frist server was done before the second tried to use it) we also had windows using backup exec.
we go a robot silo (L700) and 20 tape drives and netbackup.
took work to move everything over, but now we can "see" our backups from the gui ( and nom when I get it setup) and you can get reports as well as set up emails.
At some point you just get to big to go around putting tapes in tapedrives every night.
you just need to make sure you know how you want to backup your databases. either to disk then that backup file to tape or stright to tape with a database agent.
we use the exchange agent to backup stright to tape.
I went from spending 2 hours a day unloading tapes and loading new ones, and checking last nights tapes to spending less then an hour. and with the robot, if the backup goes to more then one tape, it does not have to wait for me to change the tape, the robot does it.
Netbackup is powerfull and configurable and is an excellent choice for an enterprise solution to your backup needs. Here is a link to a document that will give you pointers on how to spec out your environment.
Veritas NetBackup (tm) Enterprise Server / Server 6.0 Backup Planning and Performance Tuning Guide for UNIX, Windows, and Linux
Not a problem at all. I had pretty much already decided on Netbackup, but because of the way we do our purchasing, I basically have to spec everything I need in the request for quotes. I'll take a look at the planning guide posted above and see if I can figure it all out, thanks for that! I guess I'm looking for recommendations as far as this: Since this was originally just going to be a VMware backup (one more to add to the list), would I be better off just setting up for that, and then adding modules and servers as time goes on? Is that something that's easy to do with Netbackup? I just worry about getting the funds to do it right in the future, but don't know if I have the budget to do it right, right now.
Guess I'll just spec it out and see what kind of quotes I get.
One other thing:
If I license NetBackup based on a capacity model (IE, for my situation, license 3TB), I get basically everything else without additional license fees, correct? Like, it wouldn't matter if my 3TB came from 10 servers or 100, I wouldn't need to license an agent for each server. (I do realize that there are different tiers of the capacity based licensing that change which parts I get).
Message Edited by John Hoover on 04-10-200805:14 AM
Hi I think you need a full new architecture there, you can try something like this:
1. Avoid using your stand alone AIX/Windows/RH drives for your new backups architecture, will be pain to manage those ones.
2. Based on the amount of data that you need to backup (2.2TB) with the LTO2 drives will be enough for your regular clients, based on the speed of the drives that is 35MB/Sec and the number you have, means that you can backup 1.44TB in a 12 hours time window, so if you run Weekly Fulls in 24hrs run you will get your 2.2TB and during the weekdays differentials I think will be enough, then you can use the LTO3 for critical applications has the Oracle ones and add a dedicated media server for this ones, that way you will have a better speed for restores and your RTO and RPO will do better, also will reduce the time window for the LTO2 drives.
3. You can configure the rest of your Drives at least one of each kind on a single media server for any future restore, this regarding your backexec images, because Im sure you will need to left at least one of the AIX boxes up and running for any Sysback restore you could need
4. Netbackup is a great solution and covers everything, so you will be fine with your VMware boxes
Thanks for all the replies so far. I've been reading the Capacity Planning guide from above, and I think that what has become abundantly clear is that I need to bring in a consultant to assist with the design and specification of the new backup environment, and then make sure to get some Netbackup training for the future.
There are a lot of things brought up that I hadn't though of, and I want to make sure as many contingencies as possible are planned for. Backups tend to be somewhat important :)
You're absolutely right, users could care less if you've got a good backup, right up to the point they fat finger something and delete that "really important" document. In a former life, AIX sysback was used to backup Oracle DBs. Oracle had to be down for this to complete. If the backup failed, the office was told to suspend all work and log off their terminals, while I would bring everything down and re-run the backup. There were some days that it wouldn't complete until noon, and then they'd have to not only keep up with work coming in, but enter all the paper that they'd had to use all morning. I was glad to sit in the computer room rather than the office those days.