+1 on profiling.

I would not want to see any performance related changes to the
Velocity code base until there is a very good understanding of the
performance issues.

I wonder if we could get some multi-threaded unit testings going which
perform web app like rendering operations, this would get rid of a lot
of noise from the rest of the web application when profiling.

regards Malcolm Edgar

On 5/4/07, Ahmed Mohombe <[EMAIL PROTECTED]> wrote:
> I haven't seen any data as to specifically how fast
> Velocity is vs. JSP or other tools - Ahmed do you have benchmarks you might
> share?
I don't have a benchmark application that could be publicly used, otherwise I 
would have already
submitted it to Apache :).

At my previous job there were several applications combinations that ran on 
tests systems with
various framework combinations.
e.g. Few average values relevant for Velocity that I've saved:
- Struts+Velocity vs Struts+JSP webapplication ~2.5x slower
- CMS with Velocity vs CMS with JSP ~ 5x slower (lot of dynamic content with 
frequent page changes,
Velocity cache shows it's limits).
- SiteMesh+Velocity vs SiteMesh+JSP webapplication ~3.5x slower (here is the 
fault of the SiteMesh
velocity decorator too, since the FM one is faster).

Applications were implemented with several technologies for performance, 
learnability, reliability
and other parameter measurements important in a commercial context.

Basically Velocity is at least 2 times slower than JSP for all tests I've seen 
last year,(however it
was not getting slower - but JSP was constantly getting faster :) ). Most of 
the results from
2002/2003 have reversed values (at that time Velocity was faster with the same 
magnitude, hence the
reason it was included in so many benchmarks :) ).

Ahmed.


---------------------------------------------------------------------
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]

Reply via email to