cancel
Showing results for 
Search instead for 
Did you mean: 

Netbackup Policy : Calendar OR Frequency

posie80
Level 4

Hi all,

Need your expertise here :)

I have 10 policies, and each is supposed to run according to this schedule :

 

1st and 3rd week of the month

============================

Policy 1 -  Monday 12:00

Policy 2 - Tue 12:00

Policy 3 - Wed 12:00

Policy 4 - Thu 12:00

Policy 5 - Fri 12:00

 

2nd and 4th week of the month

============================

Policy 6 -  Monday 12:00

Policy 7 - Tue 12:00

Policy 8 - Wed 12:00

Policy 9 - Thu 12:00

Policy 10 - Fri 12:00

 

At first, I was using the Calendar as a schedule type, and setting 1st and 3rd week for the first bunch of policies, and 2nd and 4th for the remaining policies (Policy 6 - 10).

However, there was some concern raised due to the last week of the month (for months that have 5 mondays, for example, instead of just 4), we may be missing backups, or have way too many backups running (if we set all policies to backup on the last week of the month).

So now, I've changed my schedule type to Frequency, set to 2 weeks.

Now, the problem I have is that, how does this work exactly? Last Friday, so Policy 1 to 5, I set the "Go into effect at:" to 11/20/2011, 6:44:13 AM to ensure that these policies will run first.

I set the Policy 6 to 10, I set the "Go into effect at:" to 11/27/2011 6:44:13 AM , so that I could ensure that these bunch of policies more or less, run later (I am trying to simulate the 2nd and 4th week run as I described above).

My questions :

 

1) Now, yesterday on Monday, right on time the Policy 1 has run as expected. But today, the Policy 2 which should be running, did not run.

What could be the problem? It should be activated the same time as Policy 1 and should be running by Tue 12:00...

 

2) If, for example, Policy 1 failed and I take 2 days to fix it, and once fixed, I re-run it using the Manual Backup way (instead of from the Activity Monitor), will the 2 weeks frequency be reset so that it counts from the time I re-run, instead of the original activation date?

 

I just would like to prevent, in the future, that the 2 weeks should always be counted from when the Policy was originally activated (either 11/20/2011 or 11/27/2011) to prevent that policies end up running together at the same time or day..(because we may not have enough backup window to run more than 1 policy at one time, which is why we spread the policies to run on different days).

 

I hope that is clear enough! Any other ideas how to do this once-every-2-weeks backups in a better way..?

 

Thanks and appreciate any feedbacks you may have.

 

Regards

Posie

 

 

6 REPLIES 6

J_H_Is_gone
Level 6

Why the two sets of polices - what is unique about them that they only run every other week?

Mark_Solutions
Level 6
Partner Accredited Certified

Have to agree with JH above - i terms of data protection backing up servers once every 2 weeks is not good practice so worrying about missing a week hardly seems the real issue here - and more confusingly when you do finally back it up you do it every day for a week but thenm leave it for a week before backing it up again - very unusual.....

To actually answer your question using every 2 weeks is bound to end up not doing what you wanted due to failures and re-runs and it is bound to go out of synch.

The only way you could pin it down with frequency based scheduling is to set a start window of say 12:00 until 16:00 (so a duration of 4 hours) and set the frequency to every 5 hours and then exclude every other day of the week when you dont want it to run.

As that would be a nightmare i would suggest that you need a full review of what you are doing - of course you may have really good reasons for this so let us know what the thinking behind it is and lets see what we can come up with.

Hope this helps

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

I am battling to understand this statement:
"because we may not have enough backup window to run more than 1 policy at one time"

Maybe have another look at your RECOVERY requirements and design a backup solution around that.

Look at the amount of data that needs to be backed up on each server, network bandwidth, backup devices.

How many tape drives do you have? Make/model? How many servers to be backed up via LAN as NBU Clients? How much data on each? What type of data?

You should be able to at least perform a Full backup of all your servers over a weekend. Multiplex and multistream multiple clients simultaneously to ensure maximum throughput.

Do Incremental backups everyday  - differential incrementals will just backup each day's changes.

posie80
Level 4

I'm sorry guys - forgot to add one detail into this discussion :

The once-every-2-weeks policies are to backup Exchange mailbox databases. Each policies contain one database, and we have a total of 320 policies all in all to backup. Because of this, we stagger the policies to be backed up on different days. The reason for this is because

a) they dont want to spend money on a weekly backup's tapes (I suggested weekly full backups). So its a cost restriction..

b) because DAG is used for the exchange, the nodes within the DAG and database copies will be their first recovery point, so backups are the last point of recovery and so, they are willing to take the risk of having a database being backed up only once  every 2 weeks.

Marianne, you are right - the environment should be properly sized so that I wouldnt even need to make that statement! - however the backup requirements and sizing were submitted somehow, without proper consultations from engineers (long story) so we have to make do with what we have....

Mark_Solutions, you are right..it is very messy, specially once we receive backup failures and need to rerun, eventually we will go out of sync.

So I guess the best way is to perform once-every-2-weeks backups using Calendar instead of Frequency, since its more predictable?

The only thing is that I need to make sure all policies are set to run on the Last Week of the month as well (or else nothing will be running on that week since we set policies to run either on 1st and 3rd week of the month, or 2nd and 4th week of the month). Better have a bit more coverage just incase, rather than less!

I appreciate your help in trying to understand my scenario and your suggestions!

 

Andy_Welburn
Level 6

just set the actual frequency to 8 or 9 days instead of 2 weeks - this will allow plenty of time for a re-run of failed jobs without impacting on the time that the next scheduled job will run.

I'm certainly glad that I don't have to look after this set up tho'!!

Mark_Solutions
Level 6
Partner Accredited Certified

OK - difficult to select the last week of the month as it can clash with the 4th / 5th week of the month.

The only way to be sure is to use the 1st week of the month or the a specific day (1st / last day of month)

Let us know if we can advise further, otherwise if you are happy please close the thread by marking a solution