A rough speed guide would be that for a cross network save from one particular 
server we would previously got save speeds of around 2.5MB per save stream (we 
always run two concurrent streams in this example). On the newer server each 
stream will frequently run at less than 1MB.

If the local /nsr save is the only thing running, then it tends to run at 
around 7MB. If additional streams come in (to different tape devices) then the 
speed of the /nsr save commonly drops to below 4MB.

On the patch issue raised by someone else, the additional thing we have tried 
that I forgot to mention was to apply the latest Quality Pack (standard patch 
bundle) from HP.

Thanks,
Michael.

-----Original Message-----
From: Matthew Temple [mailto:mht AT RESEARCH.DFCI.HARVARD DOT EDU]
Sent: 20 August 2003 19:28
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Subject: Re: [Networker] Strange performance issues since NetWorker
upgrade


On Wed, 20 Aug 2003, Saunderson,Michael Robert wrote:

> Hello all,
>
> I'm hoping there's a brain or two that can throw some ideas or light on the 
> following scenario (sorry if it's a bit long winded):
>
> Up until just over a month ago we were running NetWorker 5.1 on HP-UX 10.20. 
> This system ran backups at good speeds, but was highly prone to falling over 
> and NetWorker itself would often suffer from freezes.
>
> To attempt to solve this (and other support issues), we moved NetWorker 
> across to newer hardware with the /nsr directory stored on striped disks, 
> faster processors, more memory, etc. The OS this newer system runs is HP-UX 
> 11.0. Following the move to the new system, we have noticed a significant 
> reduction in the speed of data streaming down to the tape drives and as a 
> result many of our larger backups are failing to complete due to lack of 
> time. We have since upgraded to NetWorker 6.1.3, but this has made no change 
> to the situation.
>
> The new server has identical numbers of SCSI channels and NICs. There are no 
> apparent network configuration issues as the neither the cards nor the switch 
> show any errors. Test FTPs also run at good speeds. The system also isn't 
> logging any SCSI errors. A local backup of /nsr runs at very good speed until 
> additional backup streams come in across the network.
>
> Our vendor went through all the usual settings such as Parallelism, Drive 
> Sessions, etc., but all to no avail.
>
> Does anyone have any further ideas where we could start looking?

It would help if you could give some notion of what Networker reports
about how fast (what qunatity) before and after.   (KB/Sec)


                                                Matt Temple
>

--
=============================================================
Matthew Temple                Tel:    617/632-2597
Director, Research Computing  Fax:    617/582-7820
Dana-Farber Cancer Institute  mht AT research.dfci.harvard DOT edu
44 Binney Street,  ML105      http://research.dfci.harvard.edu
Boston, MA 02115              Choice is the Choice!

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=

+

+

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=