Schedule and Retention Ideas for NetBackup
These documents are for entertainment purposes only. Use at your own risk. This site is in no way affiliated with Veritas or NetBackup, and any trademarks are those of their respective companies.
If you need assistance with anything NetBackup and/or Disaster Recovery related, You can find me on linkedIn.
Schedules determine when backups start, the type of backup that is run, how long to keep the backed up data, and can even over-write certain Policy level settings (especially needed when using SLPs).
Only a single word is really needed for a Schedule name, if you are very standardized on your frequency, type, and retention (i.e. Hour, Daily, Weekly, Monthly, Quarterly, Yearly, Manuel, User, Archive, Vault, etc). It also allows you to use your storage targets more efficiently (especially tape). Often this can be accomplished by bumping up lower retentions requirements to the next highest retention to combine them into only a few basic standard retentions.
I.E.: Daily incremental backups required for 1week and 2weeks can both be kept for 2 weeks Weekly Synthetic Full backups required for 1month and 3months could both be kept for 3 months Monthly Full backups required for 7years and 10 years could both be kept for 10 years
A bad example would be to keep every full backup for 10years. The worse case I've seen is that every single backup was kept on tape for "infinity", which lead to an unmanageable tape inventory, confusion over what's truly important, a fast growing NetBackup catalog, legal and liability issues, and skyrocketing storage and data recovery costs.
Standardizing on a few basic schedules will greatly improve the cleanliness and clarity of your backup environment. However; In some cases were special retentions, or different schedule types are needed on the same frequency, more fields may be needed to differentiate different schedules. Even then, using a standardized naming convention will allow you to maintain cleanliness and clarity.
Using two or possibly even three (if required) fields in your schedule names (separated by an underscore "_") can be done similarly to this example:
On-site (usually disk targets) retentions often differ from the above listed maximum retention depending on a number of factors:
- How much Dep-duped disk space do you have, and what is your de-dup ratio.
- Keeping on-site copies of “Archive” backups may require a tape storage cabinet or large volume tape library.
- Catalog backups are usually only needed for the short term both on and off-site
- Anything not replicated or duplicated to an off-site tape should keep its default retention as defined above.
- Compressed and/or Encrypted data should not be retained on disk for very long due to poor de-duplication rates.
- Try to standardize on either Calendar based or Frequency based schedules. Mixing (especially within the same policy) can have unexpected results.
- Start Windows determine when a backup can start (not necessarily when they go active, and certainly not when they finish), and should be wide enough to allow all clients to start their backups.
- If backups don't finish within less then 24 hours from the opening of the start window, this is usually a good indication of insufficient capacity or insufficient performance.
- Stagger your Start Windows (even by as little as 5 minutes) so that not all backups from all policies and schedules start simultaneously. This will prevent unnecessarily high loads on the Master Server.
If needed for databases or other special situations, an additional field can be appended to the schedule name using an underscore (_) as a separator.
The only time I've seen a need for this was for Oracle Databases Automatic Backup schedules that needed a SID specified to insure easy identification:
Hourly_Logs_1week_SID, Daily_Incr_2weeks_SID, Weekly_Full_SID, Monthly_SID, etc.
Back to NetBackup