tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "GOMEZ Henri" <>
Subject RE: Proposal draft for Tomcat 5.0
Date Mon, 24 Jun 2002 08:59:14 GMT
>Craig... One question: have you ever worked on a site that 
>REQUESTS does something like 10 MILLIONS requests a day (and 
>you have to be
>called out at 3 AM on a Saturday morning because things don't work?)

Pier, I also works in production environment and make my best to
avoid being called in the nigth. So I understand pretty well your concerns,
that's are mine each days.

>> But that doesn't mean I'm going to suggest that the 
>connectors be kicked
>> out just because *I* don't care about them.
>Connectors are already out... That's what 
>jakarta-tomcat-connector is for,
>and maybe one day someone will get a clue and ask to the HTTPD 
>PMC to make a
>subproject for them and remove them from the Jakarta scope at all...

Let me recall that when I asked for jtc project it was to decouple
connection part from servlet-engine core, thats' why jk and webapp
fly to there and that's coyote found it a natural home.

>Craig, be real... Tomcat, despite the beautiful design we have 
>for Catalina,
>and we have to give you KUDOS for that, is not _yet_ ready to 
>be used in
>production... It simply doesn't work, it doesn't deliver the same
>performance and reliability that (for example) ServletExec does (just
>because we use it on our main production server).

We're using Tomcat in production, many tomcat instance by OS/400,
and they works great. We didn't have 1 Million pages by days but
have about 5000/10000 servlets requests by days, all the static
contents being handled by Apache 1.3/2.0 servers (which are perfectly
tuned and stable for that task), and that's why I'd like so much mod_jk
since you could have all /APP handled by Apache HTTPD and /APP/*.jsp
and /APP/servlet/* by tomcat via mod_jk, I've the best of both world,
Apache HTTPD for static/php and Tomcat for Servlets/JSP.

>Tomcat might meet some needs of this community, and please don't preach
>about the "Apache culture", because facts might contradict 
>you. That's why
>I'm spending much more time in the APR and HTTPD projects, 
>where features
>are not the key, reliability, scalability, performance are. I 
>just know that
>HTTPD works for 58% of the internet. I'm not sure that Tomcat 
>can even be
>suited for a 10% of those...

Tomcat SHOULD be suited for reliability, scalabitity and performance,
and that's why I found Tomcat 5.0 proposal so important.

1) A smallest core, will make it faster, easiest to understand and
   more reliable.

2) That's really a share of experience from TC 3.3 and 4.x teams,
   and we should avoid here mistake from the past and use the best
   of both tomcat implementations

3) If we could have TC 5.0 avoid duplicate decoding and handling of
   HTTP stuff, which is allready done by Apache 1.3/2.0/IIS, we'll save
   cycles, profits from HTTPD speed, quality and stability and may be 
   we could hope to reach the quality standard awaited from an Apache 


To unsubscribe, e-mail:   <>
For additional commands, e-mail: <>

View raw message