James Cook wrote:
>
> ...but as a vendor we can not distribute our results...
>
> why not?
>
> This sue-happy society is ridiculus. It's getting so bad that a non-vendor
> recently was forced to pull their app server benchmarks off the web. This is
> BS. Post the test, the conditions, and the results. If any vendor complains,
> they can tweak their own container and post the original test, the
> conditions, and the new results.
>
> Inprise recently released their TPC-W results. It has since disappeared from
> their website because the results were not "audited". What a bunch of crap.
>
> Consider that Microsoft can take out full page ads proclaiming the record
> performance of their unreleased SQL Server product on an unreleased
> operating system. Where's the justice? Get some balls and release some
> numbers. Give me a copy of your product and I'll run some tests on it
> myself. I've got lots of EJB benchmarks ready. I'll post the results and see
> how long it takes for the first injunction to arrive. I'll identify the
> vendors and their original letters.
I would at least love to find benchmark code so that I could run
benchmark tests myself. I have had to create some on my own but I feel
it is inadequate to properly stress an app server and because it is
inadequate, something I overlooked will come to bite me in the rear once
I deploy to production and start to see real load.
Right now, we have been brainstorming on what kind of tests would
represent a real world app and how best to package it to touch all the
areas of an app server such as servlets, JSPs and EJBs. Right now, it
is the standard walk of:
1) Login (this could be a true login or just the creation/issue of a
cookie)
2) Present Product Listing (randomly select 20 rows from a table and
present the id/descriptions like a product listing)
3) Select One (from the twenty, select one and pull across the network
the full product description)
4) Add to Shopping Cart (self explanatory)
5) Checkout (purchase product that was put in the cart)
It would have been nice to have something pre-build that I could have
loaded into an app server, run my test scripts against it, and compare
the results against the other app servers I was evaluating. One idea
was using the Pet Store demo that comes bundled with the J2EE reference
implementation. Has anybody tried the Pet Store as the basis of their
benchmark testing?
--
Perry Hoekstra
Talent Software Services
[EMAIL PROTECTED]
"I don't see much sense in that," said Rabbit.
"No," said Pooh humbly, "there isn't. But there was going to be when I
began it. It's just that something happened to it along the way."
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
