cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew C. Oliver" <acoli...@apache.org>
Subject Re: benchmarking Cocoon?
Date Sat, 19 Apr 2003 13:30:27 GMT
A recent professional benchmark done against .NET versus a similar Java
application proved that .NET is actually faster than Java.

Following this a noted Java/AOP expert replied showing flaws in the
benchmark and said "No it isn't"

It turned out that the benchmark was funded (BTW) by Microsoft, not that it
Influenced the result of course.  :-)

Another recent benchmark funded by IBM showed that WebSphere was the fastest
and most scalable J2EE AppServer.  This was refuted by BEA's funded
benchmarking which proved that the fastest was indeed Weblogic.

Cocoon can be WICKED fast or WICKED slow depending on how you use it.  It
has a number of architectural advantages and flaws.  It can solve ANY
problem but many problems it solves poorly or expensively.

The best benchmark is two identical applications implemented TheRightWay in
the two comparable frameworks.  Then go use Cactus or apbench depending on
whether its a one trip operation or not.  Record and average out the times
and look for any configuration issues or data that doesn't make sense.
Attempt to flatten the curve by optimizing the configuration of the slower
value.  Then make sure its still comparable.  Get experts from both sides to
give you a hand and make sure you're doing the best with each app.

In the end you're still benchmarking the people more than the technology,
but that¹s probably just as important if its all OpenSource.

The question isn't whether something is the right technology, but is it the
right technology for you and your problem.  This isn't a car.  Its not
really analogous.  The laws of physics do not apply to software at this
level.  There is no silver bullet.

But what do I know?

-Andy

On 4/19/03 5:30 PM, "Argyn" <akuketayev@cox.net> wrote:

> Look, when I buy a car I know that 100hp for a family sedan with auto
> transmission is "not enough for me". There are many other things involved,
> handling, steering, etc. But this horsepower thing (whatever it means) is a
> strong marketing argument. Do you think I know how they measure it? I've no
> clue. I even suspect that it has nothing to do with real horses. And even if
> it had, I've no clue how much power has a horse. Normal horse, besides there
> are hundreds of breeds of horses... The point is that people ask this
> question "Is Cocoon fast". I've been asked it many times. The must be an
> answer. I should say "Here's the benchmark. Run it on your hardware, and
> give me a break, please".
> 
>> -----Original Message-----
>> From: John Morrison [mailto:john.r.morrison@ntlworld.com]
>> Sent: Friday, April 18, 2003 3:50 PM
>> To: cocoon-dev@xml.apache.org
>> Subject: RE: benchmarking Cocoon?
>> 
>> 
>>> From: Argyn [mailto:akuketayev@cox.net]
>>> 
>>> Hi
>>> 
>>> Do you think there's a need for benchmarking Cocoon?
>> 
>> Nice thought, but you know the saying; there's lies
>> dam lies and benchmarks ;)
>> 
>> What would you measure and what would you compare it
>> with?
>> 
>> J.
>> 
> 
> 


Mime
View raw message