The purpose of this message is for GemStone to comment on the "EJB
Comparison Project" report recently released by a group at Charles
University in the Czech Republic. This report was the subject of a posting
on January 25th by Petr Tuma, one of the report's authors.
While we are pleased that organizations are taking a systematic approach to
evaluating EJB implementations and their performance, we have grave concerns
about the accuracy and relevancy of much of the information in this report.
First, there are numerous tests that reflect the author's interpretation of
the EJB 1.0 specification in ways that are not consistent with Sun's
interpretation, nor with ours, as a member of the J2EE expert group. We
will highlight examples of this on our web site.
Second, these tests were run with GemStone/J version 3.0. GemStone released
version 3.1 in October, 1999 and is just releasing version 3.2 this month.
Gemstone/J version 3.1 increased the performance and scalability of the EJB
implementation by over an order of magnitude across the system, and realized
even larger performance increases in the transaction service. It also made
numerous changes to EJB behavior that eliminated minor inconsistencies with
the EJB 1.0 specification.
Third, the tests were run on a very underpowered desktop Intel PC. This is
not a realistic environment for evaluating the performance nor scalability
of an application server for Internet commerce environments. Further, their
tests stressed performance in a lightweight environment, but did nothing to
determine scalability with realistic Internet loads. Our rapidly growing
customer base require both high performance and scalability for their
Internet sites, something that typically results in a very robust hardware
configuration, something that leverages the unique multi-VM, shared memory
GemStone/J architecture.
So while GemStone is pleased with the intent behind the work at Charles
University, we are disappointed that they didn't open a dialog with our
company or with the industry in order to ensure relevancy and accuracy in
their report, and to abide by the terms of our license agreement. We have
contacted the authors to request the source code from their tests in order
to run them on a GemStone/J 3.2 system and report on the updated results.
We would be pleased to provide these results back to the authors in order
for them to update their report to reflect current reality.
Lougie Anderson, Ph.D
Vice President, Engineering
GemStone Systems
> -----Original Message-----
> From: Petr Tuma [SMTP:[EMAIL PROTECTED]]
> Sent: Thursday, January 27, 2000 5:01 PM
> To: Lougie Anderson
> Subject: EJB Server Comparison (WebLogic, WebSphere, NetDynamics,
> GemStone)
>
> Hello,
>
> for those interested, our research group has recently finished a
> comparison study of four EJB servers (WebLogic 4.5.1, WebSphere 3.0,
> GemStone/J 3.0, NetDynamics 5.0.40). The results of the study are
> available for free download at http://nenya.ms.mff.cuni.cz/thegroup
> under the EJB Comparison Project heading.
>
> Sincerely, Radek Pospisil & Marek Prochazka
> Distributed Systems Research Group
> Charles University, Prague, Czech Republic
>
> ==========================================================================
> =
> 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".
===========================================================================
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".
