help with a multi process http server.

[Richard Gaskin]

Andre Garzia wrote:
> Thats why I (and others) recommend using plain old CGI for real world  
> scenarios. Spawning one rev instance per request is not that bad, the  
> interpreter is small enough and starts very fast.

I enjoyed that blog post as well (thanks for posting it, David), and 
among other things it offers one of the clearest explanations of the 
various CGI alternatives I've come across.

That said, I think you raise a good point here, Andre.  There's a lot of 
concern about "scalability", but perhaps not enough about "right-sizing".

Not every system will need to handle the traffic load Google Maps gets. 
And the relative few that do have budget concerns far bigger than 
picking a CGI alternative; if they can address those, the interface is a 
no-brainer.

Where I see Andre's work as being very useful is the other 99% of web 
services that serve a smaller audience.

Look at Base Camp:  most teams using a Base Camp installation number in 
the dozens, maybe the low hundreds, rarely much larger.   The number of 
concurrent users in any given group is of course a fraction of that.

Being able to snap together instant web services for small and medium 
businesses with an ROI greater than the alternatives would seem a worthy 
goal, and it seems Andre's work moves us all a bit closer to that goal.

-- 
  Richard Gaskin Managing Editor, revJournal
  _______________________________________________________
  Rev tips, tutorials and more: http://www.revJournal.com