From: Tim Chipman (chipman@ecopiabio.com)
Date: Mon Jul 07 2003 - 18:09:43 EDT
I've dug through google, sunmanagers archive to no avail on this. Any
thoughts are greatly appreciated. Summary will follow. Hopefully this
isn't totally stupid obvious but I'm going batty.
Situation: DBA has done a dump of large dataset from Oracle to the
(logging, UFS) filesystem, generated a 130gig data file. We need to
back this up to tape, and ideally also put a copy of this data onto
another box for redundancy purposes.
The file resides on a JetStor III Disk array (HW raid5, total capacity
approx 1.4 Tb split into three volumes of ~500g/500g/400g). The host
system is solaris.8.x86 on an athlon.2000+ with 1gig ram / Ultra160 scsi
/ GigEther.
Problem: Netbackup client starts to backup this file, we get "blazing"
performance of 0.6 megs/second ; IOwaits go to 100%, and system
performance is horrible / laggy. If anything else starts to load this
box at all, it starts getting non-responsive.
Similar vein,
-> attempt FTP transfer of the file and we observe slow xfer / high
IOwait as per above
-> Simply attempt to copy the file between slices, we see the same
behaviour.
Normally, with this system, we can backup "moderately big" files (5-10
gig) at ~12megs/second via netbackup, and IOwaits *never* goes anywhere
remotely near 100%.
Is there some "Known system-level limitation" (buffer, cache,
otherwise?) which results in incredible performance drop when you are
pushing around a 100+gig file ? Possibly any workarounds or other ways
to move this data without spending 2+days waiting for netbackup to suck
the data off to tape?
Thanks VERY much for your help,
Tim Chipman
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:26:43 EDT