Dynamic server distributed on static servers (with Squid?)

Hello everyone -

We are trying to distribute a "semi-dynamic" website (we use JSP pages w/
a MySQL DB, and our content doesn't change very often) across a pool of
"semi-static" servers.

THE PROPOSED SOLUTION
We want to distribute the content across a bank of static, redundant,
"front-end" servers, which essentially act as "active" mirrors of what is
on the dynamic server.

On each user request, the request will get distributed to one of the
redundant "front-end" machines.

This machine will serve its copy of what it thinks is currently on the
server.

To make sure the "front-end"'s content is fresh, the "front-end" machine
could either: query the main "dynamic" server on each client request to
make sure it still has a fresh copy (compare timestamps)... or it could
time-out its own copy every half hour or so. If it gets a request it
doesn't have content for, it can get it from the central server.

BTW... we are rewriting all our URLS to be in "static" servable form: all
URLs ending in either .htm / .gif / etc. In reality the dynamic server
will dynamically be serving these requests.

For example /foo/bar.jsp?one=15&two=25
could be rewritten: /foo/bar/15-25.htm

or something like that.

SO NOW FOR THE QUESTION:
Is this a good idea? Should we just be distributing JSP across dynamic
servers via some expensive Enterprise servers?

Are there any relevant info. sources on this somewhere on the net?

What should be run on those "front-end" machines? Can this be done with
Apache, or will we need Squid? Or something else?

Thank you-
Received on Fri May 26 2000 - 09:03:11 MDT