(RADIATOR) threading

[Andrew]

Hi,

I'm experiencing a performance bottleneck in Radiator due to "slow"
requests of the type that Fork is recommended for. They aren't really that
slow at about 200ms, but the bottom line is they are coming in faster than
they can be read from the socket. The 200ms is due to network latency, not
load on the Radiator host itself.

I didn't expect Fork to work well, but tried it anyway. Forking on each
and every request made that 200ms into 800ms (and wreaked havoc with DBI).
It led me to wonder though if Radiator could support a predefined pool of
children. If I had a way of running two Radiator processes receiving
requests to a shared IP/port (I don't have a load balancer available),
then my problem would be solved. It would also scale quite well, I'm sure
I could run 10 without load issues, providing they are preforked. Is this
practical in anyway?

I'm aware the obvious solution is to remove the latent network component.
I'm also aware that another Radiator host could do the load balancing for
me. I'm really just curious as to whether Hugh and Mike have any plans for
anything like this.


Regards,
Andrew

--
Archive at http://www.open.com.au/archives/radiator/
Announcements on radiator-announce at open.com.au
To unsubscribe, email 'majordomo at open.com.au' with
'unsubscribe radiator' in the body of the message.