B2D Best Practices

[Ldoodle]

Hiya,

We use B2D for everything. At present, we have single (non RAID) disks for each day's backup (Mon-Fri) but are switching to a 6 drive RAID 50 setup on new hardware.

One the current system, each day's disk has a parent folder for the server being backed up, so Monday's folder structure looks like;

Root
|-----Domain Controllers
       |---- bkf files
|-----E-mail Servers
       |---- bkf files
|-----File Servers
       |---- bkf files

etc. Now as we're moving to  RAID 50 there will only be one volume within Windows and so to keep the existing format, while adding parent folders for the days seems a bit excessive;

Root
|-----Monday
|      |-----Domain Controllers
|             |---- bkf files
|      |-----E-mail Servers
|             |---- bkf files
|      |-----File Servers
|             |---- bkf files
|
|-----Tuesday
|      |-----Domain Controllers
|             |---- bkf files
|      |-----E-mail Servers
|             |---- bkf files
|      |-----File Servers
|             |---- bkf files

What do other people do?

[pkh]

Don't bother with creating an elaborate structure.  Just lump everything into one big directory.  Does it really matter to you to know which .bkf file contains the backup for which day.  BE keep track of all these for you. With a single directory, you can simplify your job setup and probably reduced the number of jobs required.

[Ldoodle]

Thanks pkh

At the moment, each server has it's own backup job, and on each day of the week. So for example, Exchange has the following jobs;

Exchange Servers - Monday
Exchange Servers - Tuesday
Exchange Servers - Wednesday
Exchange Servers - Thursday
Exchange Servers - Friday 1
Exchange Servers - Friday 2
Exchange Servers - Friday 3
Exchange Servers - Friday 4
Exchange Servers - Friday 5

This is repeated for all types of server (File Servers, Domain Controllers etc). Each backup job also has it's own device (B2D folder) (Exchange Servers - Monday, Exchange Servers - Tuesday etc), but each job shares the same day's media set (Monday, Tuesday etc)

How would you suggest reducing the number of jobs to get the same level of backup?

[pkh · Accepted solution]

For each server, you need a minimum of 2 jobs: One with AOFO off for your databases and Exchange and another with AOFO on for your flat files.  With one job, you can backup alll or a couple of servers.  So theoretically, you only need 2 jobs to backup all your servers.

[Ldoodle]

Thanks.

If you only had 2 jobs backing up all servers (one with AOFO off and another with it on), aren't you prevented from doing a restore if the job is still running?

For example, say I had one job with AOFO on for all servers with flat files and it took 24 hours to complete. During the backup someone asks if I can restore a file from a server that has finished being backed up in that job, but the job is still running as it's backing up another server?

Make sense?

[pkh]

Yes.  You are correct.  Therefore, you got to be judicious in deciding how many jobs you have.  You can have one big job or several smaller jobs.  There are pros and cons for both approaches.  You can have several jobs targeting one media set, but you cannot have a single job targeting several media sets.

[Ldoodle]

OK, what I have done is reduced the number of devices to just one for each day instead of one per server, but I will keep to the seperate daily backup jobs as I'm doing a mixture of full/diff/inc backups and I understand these have to be seperate jobs.

On another note, my file server use DFS-R to keep files in sync. I know there is a proper method of backing up at the DFS-R level in the SCC selection, but I understand performance is greatly reduced, but what I read was based on 11d - has anything changed in 2010 R3?