G'day experts,
As I'm not a B&R specialist, please allow me to ask some questions here.
We're starting to implement Accelerator for NDMP with our NBU 5220 Appliance. The goal is to speed up weekly/monthly full backups. Let's assume the following schedule:
1st Tuesday of month (monthly FULL) - Retention on Disk (Appliance MSDP) = 1 month, Retention on Tape (Duplication) = 1 year
2nd Tuesday of month (weekly FULL) - Retention on Disk (Appliance MSDP) = 1 month, Retention on Tape (Duplication) = 1 month
3rd Tuesday of month ( weekly FULL) - Retention on Disk (Appliance MSDP) = 1 month, Retention on Tape (Duplication) = 1 month
4th Tuesday of month ( weekly FULL) - Retention on Disk (Appliance MSDP) = 1 month, Retention on Tape (Duplication) = 1 month
last Tuesday of month ( weekly FULL) - Retention on Disk (Appliance MSDP) = 1 month, Retention on Tape (Duplication) = 1 month
The first FULL Backup with Accelerator sends all Blocks from NetappFiler to NBU Appliance MSDP ("Initial Full Backup) and creates the TrackLog.
Subsequent FULLs will send only changed blocks and update the TrackLog.
Conclusion so far: Subsequent FULLs depend on the blocks of previous backups, if they are not found, the need to be transferred another time.
Questions:
I'm concerned about the "retention period on disk" (in our case defined in a Storage LifeCycle Policy).
What happens if the Retention on Disk (in example: after 1 month) expires ?
I believe, the "Weekly Full Backup Image" will disappear from the catalog, but what about the all Data Blocks from the initial backup? Are they removed from the appliance disks as well? In this case the next FULL Backup will be slow again, because ALL Blocks are sent again? What happens if the expiration takes place in the middle of a running backup ?
If the data blocks are NOT removed and data in the volume to be backed up changes frequently, many "changed" blocks will be sent with subsequent FULLs and disk space will fill up fast. How to avoid such a situation?
Another set of questions in this context: The docs recommend an additional schedule with "Accelerator forced Rescan" set to "enable". Would it be sufficiant to have this in the monthly schedule (in the example: First tuesday of month)? Is this just a refresh of the TrackLog or are all Blocks transferred again?
Thank you !
