Revolution and the Web, feedback wanted, Part 1 of 3

[Brian Yennie]

Actually, scalable PHP really isn't that hard. It's easy to *screw  
up*, but for all of the theoretically objections it just happens to  
run some of the largest websites on the net. With a well-written PHP  
application, how long does it take to stick a load balancer in front  
of it? Assuming your database is your persistent storage (and those  
can of course be scaled separately), scaling PHP is beyond simple.  
Same with any stateless scripting language. (Yes, I've done it - not  
massively, but over several servers - and you can hum pretty well  
with 3 dedicated PHP boxes)

Perl/Python/PHP/Ruby are certainly flawed, but I've never understood  
why there are such tall claims about their unsuitability with all of  
the evidence to the contrary. And yes, I've load balanced PHP  
applications over multiple servers - and it was actually quite  
straightforward. If anything, MySQL is trickier to scale out, but  
there's also no requirement to use that =).

Here's one reference which I think is a pretty fair treatment of  
scaling PHP or similar applications (based on Flickr):

http://www.oreilly.com/catalog/web2apps/

Perhaps more to the point - if we CANT use Rev CGI (which I'd like  
to, but I don't know if it's wise for both technical and practical  
reasons), what we would use other than PHP/Perl/Python? We certainly  
wouldn't want to lock into something more proprietary like .NET ....?

> I think this argument is sort of like the current generation of AJAX
> programmers who think that Perl/Python/PHP/Ruby is an acceptable and
> reasonable long-term implementation plan.  It isn't because the
> learning curve is too steep and the deployment tasks too complex.