Python App Server Advice

[Graeme Matthew]

Hi all

I have written a multithreaded app server that accepts incoming XML
requests. My aim is to create apool of database connections and load all
xslt template files when the server is started. I am using pyApache so
inerpreter startup costs are nil. The area that is concerning me is for
every web request a socket connection will need to be made to the appserver,
the cgi args formatted into xml then send to the model layer (appserver) and
wait while it processes it and then recieve html page and then print back to
the client. I am looking at it this way, there will be no cost in creating
database connections and reading templates form disk so I figured that
equals out the socket connection cost on each request ?

any advice would be most welcome or even ugly criticism as I would like to
get the architecture right before writing lots of code ....


many thanks