topic Re: FC60 Performance Problem in Disk Enclosures

FC60 Performance Problem

Joe Redman — Mon, 18 Nov 2002 15:57:59 GMT

I have 2 FC60's with 2 controllers in each. One FC60 (name=FC60) has 6 SC10's associated with it, fully loaded (10 36GB disk in each SC10). The other FC60 (name=LFC60) has 2 SC10's associated with it(10 72GB disk in each SC10). I set up filesystems on each running RAID1/0. Everything I set up is consistent between the 2 FC60's as far as filesystems are concerned. Revision levels are the same and hardware connections are the same. When I started backing up the filesystems on LFC60 my backup time increased dramatically (7GB took a little over an hour to backup). Can anyone give me any areas to look at.

Thanks in advance.

Joe

Re: FC60 Performance Problem

Eugeny Brychkov — Mon, 18 Nov 2002 16:07:40 GMT

Joe, more info needed.
- which OS do you use for backup (hpux, windows)?
- how is LFC60 connected to host - direct, hubs, switches? I you use switches check port mode (N-port, NL-port, error rates etc). You can post supportshow here;
- use amlog utility to read FC60's logs. Maybe there'are some hardware problems?
Eugeny

Re: FC60 Performance Problem

Joe Redman — Mon, 18 Nov 2002 16:43:05 GMT

Eugeny,
Using Veritas NetBackup. System is at HPUX-11.0. Both FC60's are connected direct to host. Nothing shows up in the AMMLOGS. I also have to add, these are Oracle filesystems.

Thanks, Joe

Re: FC60 Performance Problem

Eugeny Brychkov — Mon, 18 Nov 2002 16:47:48 GMT

Joe, please do the following outputs:
- tdutil /dev/tdX for each FC HBA (fcmsutil /dev/fcmsX for old Tachyon cards);
- amlog -a output for LFC60 for the backup period;
- amdsp -a for LFC60;
I think if you have both FC60s connected to one machine and one performs good but another do not then the problem not in the host/software, but in FC60/cables/dedicated FC adapters (but... there are device files for each device... do you think they can be a problem?)
Eugeny

Re: FC60 Performance Problem

Joe Redman — Mon, 18 Nov 2002 18:26:40 GMT

Eugeny,

Come to find out that according to the Oracle DBA there was many new redo logs that were backed up this past backup. We will monitor and see if we have same problem this upcoming backup as things should be back to normal then.

Appreciate the assistance and sorry to have you on a wild goose chase.

Joe