[erlang-questions] Big distributed project architecture
Tue Jul 15 22:30:25 CEST 2008
Just looking for some input from people that may have worked on really big,
scaled out applications in the past.
The project is a stereotypical wanna-be game/MMO server.
I've got a decent prototype working on a single box, but now I'm starting to
put thought into distribution, scalability, etc.
In particular I'm wondering how big scaled out applications handle their
tons of user connections. Is it typical for applications like this to have
their clients connected to a single box/node whose job is just to read data
off the socket, decode it, and forward data/messages/commands/whatever to
other boxes that handle specific things like logging in, hosting "shards" of
the game, etc?
Or is there some other technique that's commonly used?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the erlang-questions