Moving serving of all public images to a separate non-OpenACS web
server process seems like a common engineering approach for busy
sites. However, how busy is "busy"?
It would be nice if the OpenACS toolkit can Do The Right Thing, such
that the point where the user of OpenACS needs to dedicate engineering
man-hours to special purpose scalability solutions is delayed as late
in their site growth as is reasonably feasible. Ideally, a medium
sized (but what does that mean exactly?) public site should be able to
just use vanilla OpenACS (and thus AOLserver) without any special
image-only web server, any special hacking of the OpenACS request
processor, etc. etc.
Don, it sounds like OpenACS 5.0 already is doing this Right Thing with
the new resources directory filter? How far does this seem to take
stock OpenACS up the "scalability without extra man hours" curve? Is
there anything else along these lines that OpenACS can or should
consider doing in the future?
