RE: Java Benchmarks - AS/400 Kicks - A#$

I believe the benchmark can be run at a different number of connections.
The larger the number of connections means the message throughput will be
lower.  The results show that as the number of connections increases the
AS/400's performance isn't affected as much as the Sun's, thus the AS/400
scales better especially considering that the second benchmarck used a
smaller 400 although it makes you wonder why the 24-way wasn't also used in
the second comparison.  Nevertheless, I agree with you that you have to be
careful when someone publishes numbers like this.  Whoever does the
publishing will pick the particular tests that show their machines in the
best light and 'forget' those that don't.  You really need to see all the
results of various test configurations and results to try and draw a
conclusion and even then it can be diffucult unless the configurations are
very similar.

Scott Mildenberger

> -----Original Message-----
> From: Leif Svalgaard [SMTP:leif@leif.org]
> Sent: Tuesday, June 06, 2000 8:15 AM
> To:   MIDRANGE-L@midrange.com
> Subject:      Re: Java Benchmarks - AS/400 Kicks - A#$
> 
> > IBM's AS/400e Model 840 server is the first to achieve a six-digit score
> on
> > Volano LLC's VolanoMarkTM 2.1.2 local performance benchmark. In testing
> > performed by IBM following the VolanoMark run rules, a 24-way AS/400e
> Model
> > 840 server with 4GB of memory handled a record 108,153 messages per
> second
> > using 200 concurrent connections -- more than four times faster than
> Sun's
> > E6500 22-way server with 30GB of memory running the Solaris operating
> > environment.
> 
> the next paragraph reads:
> 
> In additional testing using VolanoMark's network scalability test, a
> 12-way
> AS/400e Model 840-2418 using a single Gigabit Ethernet connection
> surpassed
> all previous results. The AS/400e server with 8GB of memory was able to
> process 39,529 messages per second at 9,000 connections -- more than eight
> times the results of the Sun E6500 22-way server.
> 
> 
> So, 108,153 messages/second was more than FOUR times Sun's E6500 22-way
> server
> and   39,529 messages/second was more than EIGHT times Sun's E6500 22-way
> server.
> 
> Clearly someone got some numbers screwed up here casting doubt on the
> whole
> thing. Also, the first benchmark used 200 connections, the next 9000
> connections.
> Something is not right here. Don't just be hornswoggled by marketing hype.
> Think
> a bit yourself...
> 
> Actually, the two paragraphs are not strictly speaking contradictory.
> Assume
> that
> Sun's server could only handle 10 messages/seconds, then both paragraphs
> would
> be true, since 108,153 is MORE than 4*10 and 39,529 is MORE than 8*10.
> :-)
> 
> 
> 
+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---