Version Retention with Full+Diff

[jgraul]

We've recently started using differential backups along with our full backups on some of our database. I'm trying to grasp how retention works now that we've added differentials. Our backup policy (SQL_MC) is as follows:

verexists: 7
verdelete: 1
retextra: 60
retonly: 60

The include statement in dsm.opt is simply:
INCLUDE * SQL_MC

On one particular database server, we run full backups twice a week (Sunday and Wednesday). We also run three differentials per day.

Why do the full backups erase all prior differential backups? I'm not sure what we're doing wrong. We'd like to be able to restore any of the backups within the past 7 days.

 

[GregE]

A Differential becomes inactive when the next Differential runs. It's not being erased. It exists. You just need to view both active and inactive files in your TSM SQL client. You will see all of your Differentials, but previous ones will be marked "inactive."

Differentials contain all data that has changed since the most recent full backup, so when a Differential backup runs, it contains all data that also exists in the prior Differential.

You don't need to use "verexist" and "verdelete." Just use "retextra" and "retonly" and let your retention be handled by timeframe, not versions. You can use versioning but it's going to be more complicated.

This will give you what you want:
verexists: nolimit
verdelete: nolimit
retextra: 7
retonly: 7

 

[jgraul]

Actually, when I look at active and inactive I see 7 full backups and only differentials since the most recent full backup.

But we'll check out the recommendation. Thanks!

 

[tvbe39]

tdp sql

We have a similar situation to that of post #1:
verexists: 14
verdelete: 1
retextra: 60
retonly: 60

and actually i see only one DIFF backup available for restore (even if i visualize both active and inactive backups),
but full backups available for restore are 14.

Is there an explaination of this fact (in ibm documentation or somewhere else) ??

thanks in advance.

andrea

 

[domino]

This work as designed

If you run a full backup, it will set all previous full, log, and differential backups to expired - according to your retention settings.

 

[blew]

so if i need to retain a db for a month (poin in time restore) doing daily diff, weekly full and hourly log i can set my copygroup like this (and no need others mgmtclass for diff and log)?

verexists: 5
verdelete: 5
retextra: 35
retonly: 35

Thanks!