Sybase Performance on HP vs. Sun 
Author Message
 Sybase Performance on HP vs. Sun

I have heard it through an associate that Sybase runs 6 times faster on HP's
vs. Suns.  Now before you start flaming I was not able to establish where
this information originated from or what sort of performance benchmarks were
being used for reference.  I would imagine that Sybase should run significantly
better on a latest and greatest HP 9000 ( or whatever the number as I am not
that familiar with the current HP Risc | Sparc platforms ) than the Sun III
but I would like to know if there was some sort of publications using Suns
latest offerings or even the Sparc 490/690 series vs. the HP by some third
party independent evaluation team.  I also heard through the rumor mill that
Sybase has different teams which are responsible for porting their product
to the different platforms and that the HP team just did a better job than
the Sun team.  I don't know why I find it hard to believe that any secrets
that one team would have to optimize performance wouldn't be available if
not utilized by the other team if this was the scenario.  Any info would be
greatly appreciated and please reply via email.  If there is enough interest
I'll post a summary of all responses.  Many thanks in advance.

Mark Gross



Sat, 05 Aug 1995 23:41:21 GMT
 Sybase Performance on HP vs. Sun

Hey!  We have a sybase 4.9.1 server on a HP9000/835 running HP-UX 9.0.

It is SLOOOOW!  I have heard that it should probably not be this slow,
but we cannot figure out what needs to be tweaked to up the performance.
We have the Async drivers in there, but they actually seeem to slow it down!

Here is some output from '/usr/bin/top'

Load averages: 1.35, 1.33, 1.26
81 processes: 79 sleeping, 2 waiting
Cpu states:
CPU   LOAD   USER   NICE    SYS   IDLE   UNK5   UNK6   INTR   SSYS
 1    1.35  90.7%   0.0%   6.4%   2.9%   0.0%   0.0%   0.0%   0.0%

Memory: 5860K (1372K) real, 5184K (1268K) virtual, 2900K free     Screen # 1/6

CPU  TTY   PID USERNAME  PRI NI   SIZE   RESD STATE   TIME %WCPU  %CPU COMMAND
 0     ? 25312 sybase    228 20  1476K   228K wait  557:36 92.50 92.34 /users/s
 0    sa  3854 dealscan  179 20   196K   236K wait    0:00  4.62  2.64 top
 0     ?     3 root      128 20     0K     0K sleep   4:12  0.21  0.21 statdaem
 0    sa  2812 root      154 20    52K   140K sleep   0:09  0.16  0.16 rlogind
 0    sa  2813 dealscan  158 20   216K    72K sleep   0:00  0.14  0.14 -ksh
 0     ?     0 root      128 20     0K     0K sleep   2:32  0.09  0.09 swapper
 0    r1  2880 root      154 20    52K   140K sleep   0:13  0.09  0.09 rlogind
[more stuff]

Now, the sybase dataserver (the first one '/users/s') is just aobut the only
thing running (well, 'isql' but that doesn't really "run")  No idle time,
used to be more idle time before the async.  disk stuff.  But it never goes
very fast.

Anybody got any ideas?  Anyway, what does this screen mean, I can
never quite figure out what 5860K (1372K) real is supposed to mean.
Should we give sybase more/less memory?
--
--------------------------------------------------------------------------------
Some people call me "Mister."   Some people call me "E."
Some people call me "Mister E." Some people call me "Mister Mystery." -- Sun Ra



Mon, 07 Aug 1995 05:01:01 GMT
 Sybase Performance on HP vs. Sun

Quote:
>Newsgroups: comp.databases.sybase
>Path: comserv.itri.org.tw!news!news.edu.tw!newsserver.jvnc.net!darwin.sura.net!wupost!eclnews!wucs1!mig

>Subject: Sybase Performance on HP vs. Sun


>Nntp-Posting-Host: wucs1
>Organization: Washington University, St. Louis MO
>Distribution: na
>Date: Tue, 16 Feb 1993 15:41:21 GMT
>Lines: 19
>I have heard it through an associate that Sybase runs 6 times faster on HP's
>vs. Suns.  Now before you start flaming I was not able to establish where
>this information originated from or what sort of performance benchmarks were
>being used for reference.  I would imagine that Sybase should run significantly
>better on a latest and greatest HP 9000 ( or whatever the number as I am not
>that familiar with the current HP Risc | Sparc platforms ) than the Sun III
>but I would like to know if there was some sort of publications using Suns
>latest offerings or even the Sparc 490/690 series vs. the HP by some third
>party independent evaluation team.  I also heard through the rumor mill that
>Sybase has different teams which are responsible for porting their product
>to the different platforms and that the HP team just did a better job than
>the Sun team.  I don't know why I find it hard to believe that any secrets
>that one team would have to optimize performance wouldn't be available if
>not utilized by the other team if this was the scenario.  Any info would be
>greatly appreciated and please reply via email.  If there is enough interest
>I'll post a summary of all responses.  Many thanks in advance.

>Mark Gross


Name    Robert Ju-Ching Chen

Smail   P.O. 9-65 Chutung, Hsinchu, Taiwan 310, R.O.C.
Tel     886-35-917558
Fax     886-35-820026


Sat, 12 Aug 1995 13:05:47 GMT
 
 [ 3 post ] 

 Relevant Pages 

1. SUN vs HP Performance!

2. Informix on SUN vs HP

3. HP or Sun: Which is best for Sybase?

4. Ingress vs Sybase vs Informix Performance?

5. NT Sybase vs Unix Sybase performance

6. Raw vs. Unix Device Performance on Sun

7. Performance Tuning Guide for Sybase SQL Server on Sun Solaris by Anand Atre

8. SUN vs Sybase RPC Calls

9. Sybase on SGI vs. Suns

10. Sybase vs. Sun RPCs

11. Sybase performance under hp-ux

12. sybase versus HP UX sorting performances


 
Powered by phpBB® Forum Software