10-07-2008 05:48 AM
I've inherited a horrible mess of a backup solution. Netackup 5.1 running on Windows Server 2000. The policies are horrid. The tape naming convention is a mess, there are about 20 failed backups when I come in each morning, etc, etc.
The 3 choice I have are:
1) Ditch everything. Install a new 2003 server with NB 6.5 and start over.
2) Try and tidy up 5.1 as much as possible.
3) Migrate to a new server.
Now, ideally I'd like to migrate to a brand new 6.5 server. But would I be best of simply migrating 5.1 off onto a new server and THEN upgrading to 6.5?
I don't think scrapping everything is really the way to go, would lose the catalogue. So, what's the best approach and how much pain is involved?
Thanks
:)
10-07-2008 06:08 AM
Thinking now that I could keep the 5.1 server and use it a media server until all the old images expire. I could then implement a 6.5 master server. Would that work?
:)
10-07-2008 08:20 AM
10-07-2008 01:36 PM
if the problem is with policy, pool, volume and any naming convention, then you can create all what you need without deleting anything, then you can just migrate or move your clients to this new policies with the new pools and STU's, after a while and if everything is working fine, go ahead and delete the old stuff, migrate to 6.5 is a must, becauseyou need to be at the top of technology, but is a different process and if you start from scratch a new master then you can expirience some issues on restores, other option can be upgrade your old master to 6.5, inststall a new 6.5 then migrate the images this will bring in only what you need for restores with a clean configuration.
hope this helps.
resgards.
10-08-2008 04:29 AM
Cool, thanks. I think the way forward is to get a new 6.5 server. I'll keep the 5.1 and fire it up if I ever need to do a restore. 1 question with that though: if I install the 6.5 client can I restore from the same box that previously used 5.1?
:)
10-08-2008 07:56 AM
Been there, done that.
I had a 5.1 MP5 master media and all clients.
I went to a new 6.5 master and media. Moved the robot to this master and media
Keeping my old 5.1 master ( moved the robot off and put a standalone on it so I could do restores)
The issue is that you cannot (are not suppose to so I did not try) to restore to a client that is at a higher level then the master.
so you clients have to stay at the same level as the old master. Until all your old images expire.
So I had any new clients go to the level of my new master.
kept all old clients at 5.1 MP5.
Kept the old master name in the list of servers on each client. ( this is so the client will allow the old master to do a reastore to it.)
But changed the master server to the New master on all clients.
so all my clients are now being backed up by the new master media.
I could do restores from the old master just fine.
once all my images on the old master expired.
I removed the old master from the list of servers on all clients, and upgraded all clients to the same level as the master media servers.
10-08-2008 12:26 PM
I'm doing the same as Mr Hinchcliffe.
Only difference is, once the last client has been migrated from old env to new env AND thus the old libraries are effectively read only, then we're going to zone in the old libraries to new env, and zone in a small 2U single tape drive LTO3 library to the old env. AND, import just the long retention backups from LTO1, and then duplicate to VTL and LTO4 - and keep the old master/media and small library for doing emergency restores for old images that only the old env knows in its catalog - until all the imports are done - at which point we can bin the old servers and the old catalog AND only then begin upgrading clients to v6.5.x - and just duplicate at our leisure - before finally disposing of the old libraries.
Old env: Win2000 MSCS stretch cluster, NBU ES v5.1 MP6.
New env: Solaris 10 Sparc VCS stretch cluster, NBU ES v6.5.2A.
Haven't migrated any of 250+ production clients to new env yet, only 250+ dev/test/UAT servers have been migrated so far - as v6.5.2A is too buggy - core dumps from nbpem, nbstsrv - so waiting for stability, hopefully when v6.5.3 arrives before moving the important clients over.