Question about incremental backups [Devs?]

[Mattias]

Hi !

I did read this article about the "new way" to do backups inv eeam 8

http://www.virtualtothecore.com/en/new- ... cation-v8/

That made me think. Why (or is it already) not an option where i can chose this for reversed incremental ? That first write out the increment to the repository and then and close snapshoot . After that inject the increment into the full backup that is the last backup.

As i understand the article the second oldest increment is injected into the fullbackup file witch is the oldest according to retention time. I want my last backup to be my full backup (as in reverse incremental) but benefit from shortest possible "Snapshoot_open_time" .

I hope i managed to explain what i mean else feel free to ask !

Thansk for the help !

Regards Mattias

[foggy]

Mattias, the difference here is that with reverse incremental, data is written directly into VBK file and substituted blocks are immediately saved into rollback file. So injection is performed on-the-fly. What you're suggesting will require even more I/O on the target storage, while reverse incremental is already more I/O intensive.

[Mattias]

Yes true. The I/O will be more on the repository. On the other hand it will cut snapshot_open_time with quite a bit, If i can write out the increment only and then close the snapshot and after that inject the increment in to the full backup file ?

Has this or will it be considered ? I kinda often see problems with the time window snapshoots have to be "open" for a long time, specially on I/O intensive systems like SQL or Mail.

So in the end we would get:
+full backup is the latest backup
+Shortest possible snapshot window
-Most I/O on repository

I can my self see customers i have that would use this.

Thanks for the help !

Regards Mattias

[veremin]

I want my last backup to be my full backup (as in reverse incremental)

Just out of curiosity - is that because you're archiving to tapes or it's related to something different?

[Mattias]

Just out of curiosity - is that because you're archiving to tapes or it's related to something different?

Its more that if i have a long retention time , lets say 30 days i get a very long increment chain that might get "broken". If i have reversed then i know that latest backup "always" works and i "risk" that older files in the chain get broken. Also i get fastest restore times for latest files (important in a total restore scenario). Then one can always discuss back and forth how high the risk is that the the chain breaks but.

Also , yes , it may be more useful for tape rotation also.

Regards Mattias

[veremin]

If i have reversed then i know that latest backup "always" works and i "risk" that older files in the chain get broken.

Regardless of backup mode, you can't be sure without SureBackup jobs. Storage failure may result in data corruption that then will be built into the vbk file and could be carried forward into each subsequent restore point in the chain.

[Gostev]

Its more that if i have a long retention time , lets say 30 days i get a very long increment chain that might get "broken". If i have reversed then i know that latest backup "always" works and i "risk" that older files in the chain get broken. Also i get fastest restore times for latest files (important in a total restore scenario).

Your perception is not correct. The chance of the latest backup being corrupted with reverse incremental backup is exactly the same as the chance of something in the long incremental chain of forward incremental backup being corrupted. Also, reverse incremental backup does not improve the restore speed for latest files.

[Mattias]

Regardless of backup mode, you can't be sure without SureBackup jobs. Storage failure may result in data corruption that then will be built into the vbk file and could be carried forward into each subsequent restore point in the chain.

Yes also true. Should run sure backup on regular basis.

[Gostev]

Why (or is it already) not an option where i can chose this for reversed incremental ? That first write out the increment to the repository and then and close snapshoot . After that inject the increment into the full backup that is the last backup.

You can achieve this today with the following settings:
1. Incremental backup.
2. "Create synthetic fulls periodically" option enabled with every day of the week checked.
3. "Transform previous backup chains into rollbacks" option enabled.
http://helpcenter.veeam.com/backup/80/v ... up_vm.html

[Mattias]

But with these options, don't i need double diskspace to create the synthetic full ? I dont have double diskspace so i need to run revereincremental.

[Gostev]

No. You will need the same disk space as with reverse incremental backup.

[Mattias]

Ok, thanks for clearing that up.