One big backup job vs. one server per job

[jchien]

Well, I got very limited resource (available tapes & help & ,,,, ) but lots data is on this.

What will be the benefit for one or the other?

I got a 2 backup drivers, 24 slots robotic libraries but 22 are usable for scratch media and Partitioned the 22 slots to 10 x 12.

The 10 slots partitioned is used for infrastructure backup and the other 12 slots for studio test file backup.

Total 4 servers (all win2k3) are included in the infrastructure backup but it covers 2T for weekly backup under one backup job.  I believed this can save me some tapes.  The job rate is at 1.081.00 MB/min so far and could be dropped.

Don't know I will gain anything if use one job per server instead? 

The 5th server will be on line for backup and it will use the 2nd partitioned for studio test files backup. AND it is 10T of data.  

Any suggestion will be grateful!

 Thanks!

[Dave6]

I would create seperate selection lists for each server to make it easier to administrate your enviroment (and diagnose problems), and run two policies one for your infrastructure jobs and one for your studio jobs.

[Policy - Infrastructure]
Selection List - Server 1
Selection List - Server 2
Selection List - Server 3
Selection List - Server 4

[Policy - Studio]
Selection List - Server 5

[Gilly-Bhoy]

Personally I wouldn't be setting up single jobs per server.
I would find that an administrative nightmare!
The way that the logs are structured means it is easy to see at a glance if there is a problem and with which server it lies!
The other thing is you'd only need one selection list to adjust.

But hey that's me..
Gilly

[jchien]

Thanks for the response so quickly.  

I did a quick test for a job per server. It used both drives at the same time.
 

Look like the advantage for this might be utilized my 2 drives units and cut short of my backup time needed for Weekly backup (23hours). 

The disadvantage for this is takes more tapes to run this unless I set the overwrite protection to long number.  But there is more for that!  

This is my idea (BTW, no more new tapes no budget):    suggestion are welcome!

Daily still run as one big job, 4 of my servers can run just one tape per 6 days – 4 tapes per month.

Weekly will cut down to 2 servers per job.  I think this is call load balance (hope it works).
 

This will take up to 4 tapes daily/month and 16 tapes (Sorry! LTO4) for a month of weekly data. 

Schedule the Studio Test backup away from the Infrastructure backup hours (night). 

Well, more test to do before I can safely say it save time and data.

One little story, our backup was on backup to USB Drive and take 24 + hours per server (600G).

And catalog a drive will take up 9+ hours for a size of 750G USB drive (got that to retrieve one deleted file).  

The new Robotic Libraries is almost heaven (but also nightmare just all by myself plus so many other my part-time job :D)

[teiva-boy]

Try to break up jobs by server OS, and Server roles.

jobs for the SQL servers, email, AD server, File servers, Linux vs. Windows, etc...  

This would help make troubleshooting easier, and if a job bombed out, you wouldn't have to start from the beginning like one large job.


Lastly, tape costs are something I've found to be so minuscule in the larger scheme of IT budgets, that it was no different than the Office Admin, buying reams of paper.  That would be the least of my worries, unless you had hundreds to thousands of tapes.  But 48 tapes a year?  Thats petty change compared to buying new storage, switches, workstations, etc...

[jchien]

GOOD SUGGESTION! THANK YOU! And I just ran into failed on weekly backup and have to re-run it with 2 T of backup.

[Gilly-Bhoy]

Just run a backup of the resources that were skipped and append them to the appropriate tape(s).
I prefer a job to back up all!

Anyway each to their own I guess!

Gilly