Video Screencast Help

Policy Based Backups Need Automatic Feature

Created: 11 May 2011 • Updated: 24 May 2011 | 2 comments
BernardJL's picture
0 Agree
0 Disagree
0 0 Votes
Login to vote
Status: Alternative Solution

We have implemented and used your "policy based" backups.  They seem to be cumbersome and have at least one major issue that reduces their value as a feature.

Perhaps we are missing something, but here is a sample scenario we run across:

  • There are ten machines being backed up by a policy called "Monday Cycle". 
  • This policy does a complete backup on Monday using media set "Complete Backups."  Other days it does a differential backup using media set "Differential Backups".
  • Now two of the machines were off-line for some time time and miss a few "Complete Backups".
  • The media for those two machines gets recycled since the retention period has passed.
  • At this point we have no backup for those machines.

To avoid this, we can create a media set for each machine.  That way any machines off-line will still have their latest backup and it would not be re-used. 

But using a "policy" determines the media pool to use, that means we cannot use policy based backup unless we are positive that all machines are on-line and ready to be backed up.

Because of this limitation to your "policy" mechanism, we have folks checking backups everyday and re-running backups or finding the media to change to another pool to keep the media for missed machines.

You should be able to create a better way for this work.

For example, the "policies" could have "keywords/variables" that contain information about the current resource being processed.  The variables could be use to select a media set unique to that machine. 

Or maybe make changes to how media is managed during a re-cycle operation.  If the resource in the media references a policy based backup and there is no newer backup then delay the recycle.

But the bottom line we see is the "policy" based backups have a high overhead and the value of this feature is limited.

Thanks, and if you have any suggestions on resolving this let us know.


Comments 2 CommentsJump to latest comment

Peter Imming's picture

Hello Bernard,

I am glad to hear you are looking at using Backup Exec Policies in your environment to help automate your backups. I have read your post a couple of times and I think I understand what you are looking to do.

Rather than creating a Media Set for each server. I would create a Selection List instead. BE can automate this for you after you change the setting to Per Computer in the Tools|Options|Selection Lists setting in Backup Exec. This will allow you to have a single Policy that covers your Full and Differential backups at the schedule of your choice. Each system you select will then generate a new Selection List that has only that system in it and will be named by the name of that system. You can then choose which Selection Lists should be included in the Policy itself when you edit it. This allows you to have a smaller number of media sets and use them for what Media Sets were intended for, data rentention intervals.

You can then target your Media Sets with what Selection Lists (per computer) you want.

This will essentially create a job instance for each computer that you are protecting to better track the status of each system (as opposed to a single Selection List with multiple computers in it that could result in an overall failed job if the backup failed for any single computer in the Selection List).

Try this out and let us know how it works out for you. Policies are undergoing a pretty major overhaul so we are interested in your feedback.


Peter Imming, Sr. Product Manager, NetBackup, Symantec

Login to vote
BernardJL's picture

Peter, thanks for taking the time to respond to our comment.

Sorry I was not clear on the steps outlined in my original post.

In the original post where is says:

  • There are ten machines being backed up by a policy called "Monday Cycle".  

It would be more accurate to have said:

  • There are ten machines, each with their own selection list, being backed up by a policy called "Monday Cycle".

With that you can see your suggestion is exactly what we are doing and which is problematic for us.

You point to the core problem with policy based backups in your statement:

This allows you to have a smaller number of media sets and use them for what Media Sets were intended for, data retention intervals. 

So what happens in our example is, one of the ten machines will be off-line for a few days.  The "selection list" for that machine will fail since the resource is not available. 

However, because that machine shares a media set with the other nine selection lists using the same policy, the media for the off-line machine will eventually be re-cycled and the backup for that particular machine (selection list) is now gone.

There is no way in your current implementation (without a major effort) to protect machines that may not always be on at the time a backup run takes place.

So as noted before, the only solution in your current implementation is to create media set for each machine.  Which has to be done by hand for every machine.  Then we can be sure other machines will not re-use the media, or we can change the retention period for that machine's media.

We all hope your team is working on a better implementation of how "policies" work.  It is a great idea but as implemented, requires a lot of oversight and cost to insure we have good backups.

Thanks again for quick response.

Login to vote