strange acceleator behavior ?

[StefanosM]

Hello,

I'm working with a new netbackup installation, and I'm having difficulties understanding how the accelerator works in the specific case I have.

The client system is an AIX 7.1 server, and we are backing up only a file system that contains Oracle dump files. Oracle is dumping the database every day at 6 a.m., and before that, all previous dump files are deleted.

In my eyes, every day the files are new, and I can see that from the timestamp. The [dump files are 5TB total

The initial backup took 15 hours. The second backup took 3 minutes with 99% acceleration. I know that the files are almost the same as the previous ones, but I cannot understand how Accelerator knows that.
Info bpbkar (pid=27330030) accelerator sent 1024 bytes out of 5421600012288 bytes to server, optimization 100.0%

Does anyone can explain this behavior ?

PS: We are using media server deduplication. Client-side deduplication cannot be used because the deduplication ports are closed. We will open them later.
We are using a flex appliance with master, media, and worm storage roles. The Netbackup version is 10.1.1.

[davidmoline]

Hi @StefanosM 

I agreee it is strange - I assume that the file dumped by Oracle each day (in question) all have the same size and file name - do you know if the contents would be similar?

Anyway if you want to determine the cause I would log a support case as the behaviour seems wrong. Otherwise, if the backup is just the Oracle dumps, modify the policy to remove the accelerator flag.

David

[StefanosM]

Thanks @davidmoline
The majority of the export files are the same, as the customer informed me. But all agree that there are changes as it is a live Oracle database.

I will open a case to investigate. I hope that the engineer will understand...