The first request can take some time ~5 secs because you are starting up
several services. You can see which ones take the most time in
turbine.log. Make sure you are only starting those that you are using.
But that should not extend to any further requests. The default
template service performs some string parsing and filesystem io matching
up layouts and screens and java classes. The module.cache flags the
service to cache the information, so the performance is not an issue in
production, but I do not think it should be that noticeable without the
cache with a single developer hitting pages. Something like 300ms,
iirc. As far as db connection pooling, Do you have the #connections
property set higher than the concurrent requests? One should be
adequate in most apps, with one developer.
john mcnally
Berin Loritsch wrote:
>
> Kasper Nielsen wrote:
>
> Does anyone know why I am getting such bad performance? I am running
> on my local system with the following parameters:
>
> SUN JDK 1.3.0_01 (HotSpot Server)
> Tomcat 4.0b5
> Turbine 2.2-CVS as of today
> Turbine Flux/TDK CVS as of today
> Oracle 8i (8.1.6 on a remote host)
> Win2000 Athlon 750 w/256MB RAM
>
> After moving to Turbine 2.2-dev, the problem actually got a bit worse.
> My processor is pegged out for 3-5 seconds before a response. When
> cache is enabled, the same phenomenon happens for the first few requests,
> then it starts getting faster. I am using Netscape 4.75, with the priority
> below normal--a trick I use when testing complex pages with the servlet
> container on my machine.
>
> The only thing I can think of is that the JDBC Connection pooling is
> creating a new Connection object for each request. This is clearly not
> necessary, and does not seem right.
>
> >>From what I can glean from the comments is that my experience is not
> normal, and I would like to know what is going on. Any advice is
> welcome.
>
> >
> > ----- Original Message -----
> > From: "Jon Stevens" <[EMAIL PROTECTED]>
> > To: "Turbine-user" <[EMAIL PROTECTED]>
> > Sent: Tuesday, July 03, 2001 8:56 AM
> > Subject: Re: Performance of sample app
> >
> > > on 7/2/01 11:26 PM, "Kasper Nielsen" <[EMAIL PROTECTED]> wrote:
> > >
> > > > ECS has.
> > >
> > > Sorry. Not significantly. Pages still return in the ms range with ECS.
> > > >> People have been using it for years now and the performance of Turbine
> > > >> itself has never been a major issue.
> > > >
> > > > Well, can it ever get fast enough...
> > >
> > > I don't believe that...I believe that currently, a Turbine application can
> > > more than saturate a T1 line which is what most people are using. Even if
> > > you go higher, I'm sure that with the right hardware you can saturate a
> > > 10mbit connection. Regardless, few people on the other end will even be on
> > > that high of a connection so the speed will all go to waste anyway...
> > >
> >
> > I don't believe that... ;-)
> > We really don't have any numbers, so this is a pretty useless discussion
> > anyway.
> >
> > - Kasper
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
