You might also try a ping from the client to the server with the route
option (what ever that will be under your version of tcp/ip) and see
if there are any 10BaseT legs in there.
MTU size is a BIGGIE...
ADSM Server Data Base location is a BIGGIE... many smaller bits across
disks and controllers seems much better than ONE BIG MONSTER. Toss
together a script to do "iostat" against your log & DB volumes and
check that during backups.
Are you running client compression ?
Do you have accounting turned on and have you looked at the accounting
records to see if they indicate anything in the blah wait fields?
Just some thoughts...
Archive processing of 'D:\BOOKS\TEMP\*' finished with 0 failures.
Total number of objects inspected: 26
Total number of objects archived: 26
Total number of objects updated: 0
Total number of objects rebound: 0
Total number of objects deleted: 0
Total number of objects failed: 0
Total number of bytes transferred: 30.6 MB
Data transfer time: 2.84 sec
Data transfer rate: 11,024.18 KB/sec
Average file size: 1,206.5 KB
Elapsed processing time: 0:00:42
dsmc>
and this IS over 10BaseT
______________________________ Reply Separator _________________________________
Subject: Re: Boosting performance of the Solaris server?
Author: ADSM-L (ADSM-L AT VM.MARIST DOT EDU) at unix,mime
Date: 4/28/97 11:33 AM
Hi Russell,
Russell Street wrote:
>
> Hello...
>
> Any ideas how to make ADSM work faster than I am seeing it at the moment?
>
> For example, doing a back up to a disk storage pool across FDDI I only
> see about 500-600 Kb/sec as reported by ADSM --- even lower with
> smaller files or more data.
>
> e.g., 28 files:
>
> Total number of objects inspected: 28
> Total number of objects backed up: 28
> Total number of objects updated: 0
> Total number of objects rebound: 0
> Total number of objects deleted: 0
> Total number of objects failed: 0
> Total number of bytes transferred: 193.5 MB
> Data transfer time: 350.40 sec
> Data transfer rate: 565.74 KB/sec
> Average file size: 7,079.0 KB
> Elapsed processing time: 00:08:57
>
> I am using a Sun Solaris (2.5.1) system as a server and variety of
> clients (Suns, Win95, Novell). (Oddly enough, the Novell system
> managed 1200Kbytes/sec...) Neither the server or the clients were
> doing much else than the ADSM work.
>
> For a comparision, using rcp to move the file around it can do at >
> 1Mb/sec. For writing direct to a DLT drive across the network, I
> could manage 3.5Mb/sec at one point.
>
> Any suggestions are welcome. [1]
>
> Russell
>
> [1] except for moving to AIX ;)
Well, I am using AIX 3.2.5 and I don't mean to say that AIX is better
or worse than any other UNIX or O/S, I have used Solaris, Sun O/S,
HP-UX, OSF/1, etc and I find that somewhere along the line all of them
can do what you want them to do.
In comparison to you machine, I have four machines and here are the
results that I got for them this weekend:
M1 M2 M3
------- ------- -------
# inspected: 204,211 299,832 234,553 295,859
# backed up: 161 332 1,092 236
# updated: 0 1 0 0
# rebound: 0 0 0 0
# deleted: 44 76 612 49
# failed: 2 0 0 0
# bytes transferred: 177.9 MB 72.3 MB 133.3 MB 3421.1
KB
Data transfer time: 186.53 sec 96.80 sec 180.95 sec
4.93 sec
Data transfer rate: 976.75 KB/sec 765.16 KB/sec 754.83 KB/sec
693.55 KB/sec
Average file size: 1,571.5 KB 383.8 KB 193.3 KB 50.7
KB
Compression %: 39% 50% 50% 72%
Elapsed time: 01:00:24 00:35:02 00:30:03 00:26:46
If you want, we can try to compare the different opt, sys, etc files and
see if any of them help you. I guess that the other thing to do if for
you
to try and discover if anything outside of the machine is hampering your
machine, like overnight batch jobs, other backups, loaded network
traffic,
etc.
Francisco
x9477
--
Francisco Franco e-mail: francisco.franco AT hydro.on DOT ca
Francisco Franco e-mail: francisco.franco AT hydro.on DOT ca
Ontario Hydro tel: (416) 592-9477
700 University Ave fax: (416)
Toronto, Ont.
--MimeMultipartBoundary--
=========================================================================
|
