You are viewing a plain text version of this content. The canonical link for it is here.
Posted to modperl-cvs@perl.apache.org by Ask Bjoern Hansen <as...@valueclick.com> on 1999/07/14 20:37:22 UTC

Re: http://perl.apache.org/bench.txt

On Wed, 14 Jul 1999, Matt Sergeant wrote:

...
> > The benches I'm maintaining are "Hello World"... I think
> > Matt was going to work on "Real World", but perhaps you'd
> > like that, or some more clever title?
> 
> What I was working on was a spec for a "Real World" app that would then
> have to be implemented in the different languages. And also a spec for
> what results were required, and some definitions of how we were going to
> make the benchmarks more realistic. Let me know if I'm wasting my time.

[ CC'ing modperl-cvs so Doug and others can see ]

It sure sounds interesting, although I could be afraid that it easily
could end up giving certain languages with certain features or effective
ways to implement certain applications great advantages. The test will be
a lot more advanced and might be more "real" than the hello world test,
but it's still just a benchmark of a very specific application in a very
specific environment.

(quoting the rest of your mail for modperl-cvs to see if you want to)
 
> What I had also intended to do was provide a web form on sergeant.org
> that you entered the benchmark data into and it saved it as XML. I could
> then leave the final format till later, and besides - I like playing
> with XML.
> 
> A preview to what's coming is that It's a web based calendar, you have
> to implement a database interface, authentication, user-configuration,
> updates, selects, and outputting a HTML table containing the calendar,
> and user-info. There only has to be one screen. I'll supply the table
> spec (although you're free to index it however you like in an effort to
> improve your benchmarks), and the data for the tables (I'll try and make
> the contents larger than most DBMS's page sizes), and a picture of the
> resulting output (you only have to make it look the same - not produce
> identical output). You (or whoever) provides benchmarks of hits per
> second for both static HTML of a certain file-size, and for the
> script/app. The result is the percentage difference.
> 
> There's a lot more detail in the document I'm writing - although because
> the doc is so long I don't know if I'll have time to implement an
> example...
> 
> I'll post it to modperl@apache.org when I'm done for a peer review. Then
> I think it would be good to post it to lots of different lists (e.g. ASP
> lists, cold fusion list, velocigen list, PerlEx list, etc). I also think
> it might be good to post it to slashdot, although I don't know if my
> server could stand the effect... :)
> 
> -- 
> <Matt/>
> 
> Details: FastNet Software Ltd - XML, Perl, Databases.
> Tagline: High Performance Web Solutions
> Web Sites: http://come.to/fastnet http://sergeant.org
> Available for Consultancy, Contracts and Training.
> 

-- 
ask bjoern hansen - <http://www.netcetera.dk/~ask/>
more than 14M impressions per day, <http://valueclick.com>


Re: http://perl.apache.org/bench.txt

Posted by Joshua Chamas <jo...@chamas.com>.
Matt,

My only concern when reading about the bench you are constructing
is its complexity.  It would be great if your final codebase
to test were very streamlined algorithmically.  Calendars
can be ugly logical beasts, especially with a grid layout, 
and a complex bench might thwarte the commmon folk 
from trying it.

Otherwise, sounds great !

--Joshua

Matt Sergeant wrote:
> 
> Ask Bjoern Hansen wrote:
> >
> > On Wed, 14 Jul 1999, Matt Sergeant wrote:
> >
> > ...
> > > > The benches I'm maintaining are "Hello World"... I think
> > > > Matt was going to work on "Real World", but perhaps you'd
> > > > like that, or some more clever title?
> > >
> > > What I was working on was a spec for a "Real World" app that would then
> > > have to be implemented in the different languages. And also a spec for
> > > what results were required, and some definitions of how we were going to
> > > make the benchmarks more realistic. Let me know if I'm wasting my time.
> >
> > [ CC'ing modperl-cvs so Doug and others can see ]
> >
> > It sure sounds interesting, although I could be afraid that it easily
> > could end up giving certain languages with certain features or effective
> > ways to implement certain applications great advantages. The test will be
> > a lot more advanced and might be more "real" than the hello world test,
> > but it's still just a benchmark of a very specific application in a very
> > specific environment.
> 
> This is an absolutely fair assesment. However I figure this could be
> just the start, so once we've evaluated which is the best performing
> database access script environment we can go on to evaluate other stuff
> at a later date. I figure this would be the sort of app most people are
> doing (besides extremely simple stuff). Help me out if there's a way I
> can improve things.
> 
> --
> <Matt/>
> 
> Details: FastNet Software Ltd - XML, Perl, Databases.
> Tagline: High Performance Web Solutions
> Web Sites: http://come.to/fastnet http://sergeant.org
> Available for Consultancy, Contracts and Training.

Re: http://perl.apache.org/bench.txt

Posted by Matt Sergeant <ma...@sergeant.org>.
Ask Bjoern Hansen wrote:
> 
> On Wed, 14 Jul 1999, Matt Sergeant wrote:
> 
> ...
> > > The benches I'm maintaining are "Hello World"... I think
> > > Matt was going to work on "Real World", but perhaps you'd
> > > like that, or some more clever title?
> >
> > What I was working on was a spec for a "Real World" app that would then
> > have to be implemented in the different languages. And also a spec for
> > what results were required, and some definitions of how we were going to
> > make the benchmarks more realistic. Let me know if I'm wasting my time.
> 
> [ CC'ing modperl-cvs so Doug and others can see ]
> 
> It sure sounds interesting, although I could be afraid that it easily
> could end up giving certain languages with certain features or effective
> ways to implement certain applications great advantages. The test will be
> a lot more advanced and might be more "real" than the hello world test,
> but it's still just a benchmark of a very specific application in a very
> specific environment.

This is an absolutely fair assesment. However I figure this could be
just the start, so once we've evaluated which is the best performing
database access script environment we can go on to evaluate other stuff
at a later date. I figure this would be the sort of app most people are
doing (besides extremely simple stuff). Help me out if there's a way I
can improve things.

-- 
<Matt/>

Details: FastNet Software Ltd - XML, Perl, Databases.
Tagline: High Performance Web Solutions
Web Sites: http://come.to/fastnet http://sergeant.org
Available for Consultancy, Contracts and Training.