[erlang-questions] More nodes than machines
Sun Oct 21 08:09:48 CEST 2007
For the most part we use a lot of nodes because we start one node per
component of the system. It has more to do with ease of deployment and
management than anything else.
As far as mochiweb goes we definitely aren't bound by it's
limitations. It's as least as good as any other pure erlang http
server that takes advantage of the VM's header parsing. Ours all sit
behind nginx so there are not massive simultaneous connections for our
typical uses, though it should be good at that too.
On 10/17/07, Jay Nelson <jay@REDACTED> wrote:
> This question is directed to Bob Ippolito because he said the
> following in the thread on "OCaml vs Erlang":
> > A year later we have about 16 machines running 80 Erlang nodes
> powering about 16
> > different "components" of our infrastructure
> 1) Why did you choose to use multiple nodes on each physical machine?
> - Are your machines multicore?
> - Are you using SMP for your erlang nodes?
> - Are you using virtualized machines (via Xen, VMWare or other
> - What kind of machines are you running (CPU and memory)?
> 2) Any performance numbers on simultaneous connections or connection
> rate per minute sustainable by your mochiweb http server?
> erlang-questions mailing list
More information about the erlang-questions