Dennis Clarke wrote:
Comparison numbers are only valid of the tests run are the same.

So here is the UFS test once more without the compression and
with -no-fifo :

jupiter-sparc-SunOS5.10 # ptime /opt/schily/bin/star -x -xdir -xdot
-no-fifo -U file=../linux-3.5.1.tar
star: 46849 blocks + 0 bytes (total of 479733760 bytes = 468490.00k).

real 27:44.237
user 2.031
sys 42.419

Not a good result.
So try to think about the reasons..... star is definitely not the reason.
The fact that you spend 10x the amount of expectes SYS CPU seems to
lead to a
problem on your system.

Also the USER CPU time is 8x the expected amount. Did you run this
test in
_very_ old hardware?

It would be reasonable to think of a Sun Fire V480 as old hardware yes, but
not *very* old. I do have *very* old if you would like me to test there?
Well, this machine is 11 years old now.

This explains the large amount of CPU time.
The server runs fine, is patched up to date. The UFS filesystem that was
used is actually the root filesystem and it is a metadevice mirror of
the two internal disks.
A simple mirror is slow.

However, a 2300 GB FCAL drive should be faster.

But... did you turn on logging?


EMail:joerg at schily.isdn.cs.tu-berlin.de (home) J?rg Schilling D-13353 Berlin
js at cs.tu-berlin.de (uni)
joerg.schilling at fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
URL: http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily

Search Discussions

Discussion Posts


Follow ups

Related Discussions



site design / logo © 2017 Grokbase