I saw on the nginx faq that “10,000 inactive HTTP keep-alive connections
take about 2.5M memory”
Sounds incredibly good! I was wondering if nginx would scale to 300k
comet connections on a single server. From the faq entry, it sounds like
that would require 75 megs of memory, which sounds like it’d be barely
raising a sweat.
Our application involves holding open comet connections to about 300k
clients, at very low per-connection messaging rates, using NHPM
(pushmodule.slact.net) or whatever is the best way to use nginx with
Would nginx be ideal for this?
Thanks so much
Posted at Nginx Forum: