According to these tests, it seems to me that for purely servlet
web-applications, it is significantly faster to use any standalone version
of Tomcat, than to use an apache connector.
Is anyone doing this for production environt, or are there other concerns
that prevent that?
(Sorry for drifting off topic ... )
Magnus
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Jon Stevens
> Sent: 20. februar 2001 21:44
> To: Turbine
> Subject: Re: Turbine and JServ
>
>
> on 2/20/01 6:38 AM, "Rafal Krzewski" <[EMAIL PROTECTED]> wrote:
>
> > Shamil Yahin wrote:
> >> JServ is dead, but it still runs faster then present Tomcat...
> >> Therefore people can have interest in JServ.
> >>
> >> What is the prognosis - will Tomcat be as fast as JServ in the
> near future?
> >
> > Tomcat 3.x probably won't.
> > Tomcat 4.x (aka Catalina project) definetely will.
> >
> > Rafal
>
> Actually recent benchmarks (what are they worth anyway) posted to the
> tomcat-dev list showed 3.x being significantly faster...
>
> ---------------------------------------------------------------------
> Hi all.
>
> Here are the results of some measurements made with ab on Tomcat
> versions 3.2.1, 3.3.m1 and 4.0.b1 with different JDKs.
>
> System characteristics:
> - Pentium III, 450 MHz / 128 MB RAM.
> - Redhat Linux 6.2
> - hdparm -T gives 170 MB/s
>
> Software versions:
> - 3.2.1
> - 3.3 milestone 1
> - 4.0 beta 1
> - Apache 1.3.14 ( -DEAPI )
> - mod_jk.so (EAPI) from 3.3.m1
>
> JDKs (all fresh dowloads from 17 feb 2001):
> - IBM SDK 1.3
> - Sun SDK 1.3
> - Blackdown FCS 1.3
>
> Conditions:
> - ab is running on the same system
> - the test consists of 4 sets of 10000 requests at each condition
> - Concurrency levels are 1, 10 and 100
> - test targets are a servlet (examples/Hello.class from TC3.3.m1) and a
> file containing only the string "<html></html>".
> - both, standalone and with apache have been tested.
> - TC 3.2.1 and 3.3.m1 where used out of the box. TC 4.0.b1 was set up
> without file logging (SystemOutLogger), without AccessLogValve and with
> maxProcessors set to 120.
>
> This table summarizes the results. The values shown are requests per
> second for the last set of 10000 requests at c=10. The first value of
> each x/y set is for the static target (test.html). Apache was only
> tested with the JDK from IBM.
>
> Tomcat version (IBM-JDK | Sun-JDK | Blackdown-JDK )
> 3.2.1 standalone (668/994 | 491/783 | 463/749)
> 3.3.m1 standalone (759/1294 | 556/1158 | 544/1107)
> 4.0.b1 standalone (400/841 | 215/719 | 207/701)
> 3.2.1 Ajp12 (940/407)
> 3.3.m1 Ajp12 (960/421)
> 3.3.m1 Ajp13 (952/488)
>
> The static part of the last 3 rows shows the response of Apache, not
> Tomcat, about 950 req/sec.
>
> The attached zip file contains the test script and the result files.
> Some notes on the results:
>
> - Tomcat 4.0.b1 shows long maximum times at c=100: 21 seconds.
> - For c=1, TC 3.3.b1 gives 50 req/sec with Ajp13 and 196 to 344 req/sec
> with Ajp12 (Hello servlet).
> - No errors where reported by ab.
>
> Suggestions for improving the tests are welcome.
>
> Kind regards.
> Rolf.
>
>
>
> ------------------------------------------------------------
> To subscribe: [EMAIL PROTECTED]
> To unsubscribe: [EMAIL PROTECTED]
> Search: <http://www.mail-archive.com/turbine%40list.working-dogs.com/>
> Problems?: [EMAIL PROTECTED]
>
------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Search: <http://www.mail-archive.com/turbine%40list.working-dogs.com/>
Problems?: [EMAIL PROTECTED]
