On 9 Aug 2006 at 10:35, Also Sprach Matt Sergeant: > On 9-Aug-06, at 9:47 AM, [EMAIL PROTECTED] wrote: > > > The server sits there listening for a request. This request > > has to go to a db, retrieve lots of data, generate the xml etc > > and eventually pass back html. Let's say that the retrieval of data > > takes about 5 mins. > > If 100 users put in a request at the same time, the server blocks to > > service to first user, how does it service the other users without > > forking or passing them on to other daemons ot threads? > > (I have never used threads and have little idea about them) > > So rather than answer this, I'll throw the question back to you - how > do you currently cope with a system that expects 100 concurrent users > requesting pages that take 5 minutes to generate?
The only similar thing I have is a daemon that sits and blocks on a pipe waiting for requests from a process that monitors directories. So I don't lose requests (assuming I could, not sure if it's possible) as soon as I get a request I fork and let the child handle the request without waiting for it and the parent goes back to listening on the pipe. When there are a lot of requests it slows the system down somewhat :) I wrote it about 3 years ago when learnin perl so I am more than happy to explore better methods than perlIO teaches. I presume the way to go would be to have the Danga stuff handle the request in a listener and have it pass on the request to 1 of N other daemons, like apache does now. Probably. John --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]