Oracle backup green but cannot be used for recovery !?

[cxwsrei]

I issued a case (2635113) regarding oracle backup.
Finally I realized that I made a big configuration fault : one disk of the VM has been excluded due to being defined as a independent disk.
Meanwhile I changed it and included the formerly excluded disk.
BUT: all doracle database backups until that time has been ratd as green/ok.
How can a database backup be rated green/ok when it cannot be used for database recovery ?
I asked this question to the analyst that worked on the ticket mentioned above.
His response was : "The tool works as designed, all is ok".
I don´t agree !
@veeam: It´s from my point of view not a good policy. I have to repeat my question form above : How can a database backup be rated green/ok when it cannot be used for database recovery ?

[Andreas Neufert]

Hi,

thanks for the valuable feedback. Let`s discuss the situation. Usually B&R checks whatever you have configured is correct. It will not check what you have set is correct or not (this is another topic to discuss).

I guess that our Guest processing worked as expected and brought Oracle in a consistent state before backup (Alter DATABASE begin backup), then it had created a snapshot to preserve this stage and then backup all disks that are possible. You excluded one/many volumes, so we did not back them up. As we process VMs and physical Server on Image level the backup process itself has no overview about Oracle itself and the structure on disk by design. Many customer exclude in the way you had configured some of the databases or disk areas that they do not want to include in the backup (as we operate on image level, there is no other way to exclude DBs on a linux Server for example). That said it would be hard to warn you and not the one that did this on purpose.

For that and many other reasons we have implemented 2010 a feature called SureBackup where you can check the recover-ability of VMs and you can implement a test there that shows if the DB will start after a restore.
Actually it was exactly added to the product to cover your described scenario. Backup so it´s job as configured and restore failed because of missconfiguration of Veeam or within the VM.

[cxwsrei]

Dear Andreas

thank you for the fast response - but I still do not agree (at all).

First- Regarding the backups of a VM - I agree

Second - But when I include "guest processing - in this case oracle database backup - then I make an oracle database backup.
So - when the VM backup is confirmed green it´s ok from my point of view - but there is a second task/layer (whatever synonym) - here it´s an oracle database backup.
And when a database backup cannot be used for a database recoivery it cannot be rated green.
This is not really discussable from my point of view.
Each customer must able to rely on a "database backup" - but it cannot be the case that in an emergency case I have to realize I cannot use a so called database backup for recovery and a analyst of veeam would tell me it behaves by design respectively it has been my choice that I cannot recover because I designed respectively configured it that way.
The only thing that is to be discussed at that point whether the final status is yellow (a warning has to be thrown) or red.
But green ? Never. You´re much too much focussed on the backup of a VM but not to backup of a database.

[Andreas Neufert]

Hi,

I would agree if the backup was not successful because of processing reasons, but you have specifically selected a disk to be excluded. So how should we assume if it was a mistake or on purpose. We have to assume that it is done on purpose.
That`s why SureBackup is there to detect all this.

We announced a while ago that we will have RMAN based Oracle Backup soon, so maybe this is a better way for you to process the oracle databases.

[cxwsrei]

Andreas

thatswhy I closed the ticket I mentioned. Because I thought I see a discussion coming up finally for nothing.
But on the other hand it still went through my mind.
The problem from my point of view has been mentioned by you. Veeam BR -- ASSUMES -- that I excluded the disc intentionally.
And exactly that´s the point.
A backup software should never continue processing without a warning or a process stop when the evaluation of a backup is that it cannot be used for a recovery.
What philosophie do you represent ? Strange. You don´t warn your customers when they might run into a trap.
A backup that is for nothing is really a big trap.
Simply change the software that you warn your users that they might doing something wrong (when a warning is raised it does not mean that the rpocess is aborted).
But to ASSUME that something has been done intentional and to IGNORE that a backup is useless in a case of recovery is negligent from my point of view.
Let the user decidde if something is intentional but never ASSUME those things.
Really strange to me to hear such a statement from a representative of a backup soluion provider.

Las but not least : I haven´t known the tool SureBackup until you mentioned it. Is it part of the BR installation ?
And why in this case simply a warning is not raised like "Your backup might be useless for recovery purposes. Please verify by SureBackup" ?
I hope you got my point.

Kind regards
Stefan Reitinger

[tsightler]

Actually, we don't just 100% assume an independent disk is configured correctly. The first time we backup a system that has an independent disk we issue a warning status for the backup if any disks on a VM are independent and specifically call out those disk. If this warning is ignored, then we assume that this was intentional. Here's an example of a VM with an independent disk being backed up and the warning that is presented:

[cxwsrei]

First of all : my issue is only a customer feedback, nothing else.
As already mentioned this assumption is negligent from my point of view.
Example : this warning raises when the substitute of the reponsible administrator is in charge, due to vacation for example.
After his return he does not get aware of that warning.

But my complain is in context to a database backup.
I cannot rate a database backup green when it cannot be used for a recovery.

When you exclude a disc and consequently any application data is inconsistent it´s the fault of the customer.
veeam cannot know if anything gets inconsistent when a customer does configure it that way.

But when you configure guest processing -> oracle this is done for only one dedicated purpose : to make a database backup.
That is the difference. A backup software that makes EXPLICIT database backups cannot rate a database backup green/ok when it´s definitely inconsistent.
As already mentoned : THIS IS NEGLIGENT FROM MY POINT OF VIEW.

Anyway - as I mentioned above - this is only a customer feedback and I only like to give a hint to improve the product.
But maybe this is only the my way of thinking about that constellation respectively I represent a minority.

[mweissen13]

Hi! The warning is good and i wasn't aware of it. But i think i also see a problem here.
I don't think it's a good design that it only appears the first time and then disappears automatically.
IMO the user should have to ACTIVELY confirm that the disk was set to independent mode on purpose in order to set the state go green for the consecutive runs.