SSO makes tape drives go more slowly!

[Enness]

Without SSO, I get 12,000 MB/min.

With SSO, I get 7,500 MB/min.

No other changes!  What's the difference?  SSO option installed.  That's it.  Tried both with and without the recommended tape drive properties 'write single block mode' and 'write SCSI pass-through mode'.  No difference.  SSO remains slow. 

Anyone seen this, and any ideas how to 'fix' SSO?

Environment: BE server.  Fibre-attached tape library, 4 x LTO3 drives.  Shared between 3 Backup Exec servers.  All tests performed on same server.  No other server using library.  No other jobs running on any server.  Backing up very large files from Backup Exec local drives directly to tape.

Cheers!

[CraigV]

Are those other 3 backup servers configured correctly with SAN SSO?

What sort of data are they backing up, and do you get these speeds (with/without SAN SSO) on the other servers too? Or is just the primary server?

[JoaoMatos]

Hi,

How does a backup direct to tape (SSO), is slower than a backup over a network? Something is misconfigured...

Check, if zoning is well configured.

Regards,

JoaoMatos

[CraigV]

...not quite sure how zoning can affect this. The FC HBAs of each server will log into the switch port at a specific speed, along with the FC connections on the tape library.

Zoning would ensure that the servers are abole to see the library. The only time I can see this being an issue is if you have 1 massive zone with all the servers HBAs and the library in 1, which can cause an issue, or if you are backing up across an ISL with not enough connections.

If it is slower, chances are it is a misconfiguration of SAN SSO instead. But please report back if you don't come right.

[Enness]

Hi all, thanks for responses.

Q. Which of the 4 block mode & path-through mode options do you have selected in your SSO environment?  (I have just the 2 write options).

In response to your questions:

The source of the data is local HDDs on the Backup Exec servers.  So, data path hasn't changed at all.  Has always been from local HDDs, direct to fibre tape drives.

Zoning correct (only one host per zone & only one zone per host).

Speeds quoted are on secondary SSO servers, we don't run any jobs on the primary SSO server (which is also the CAS).

Weird, eh?

I have asked Symantec for technical details on the effects of the tape drive options, but their front-line staff don't understand me, nor do they understand Backup Exec.  Interestingly, this article has the warning 'selecting any of the above options can adversely affect the drive's performance if the device is attached via a SCSI or an IDE channel' - makes me wonder effects of doing this on fibre-channel too.  http://www.symantec.com/docs/TECH91959

[CraigV]

...take those 2 options off. I rarely use them. See if this makes any sort of change.

Mmm...the other thing...what hardware are you using?

I have experienced an issue with high polling rates on HP libraries with FC attachments, which causes constant disconnects as a result.

If you have HP hardware, then check the link below...

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=110&prodSeriesId=...

It's going to give you some sort of idea of what happens, and how to fix it.