On Thu, Feb 24, 2011 at 5:52 PM, Benoit Chesneau <[email protected]> wrote:
> Reading your doc, I'm was saying myself erlang would be appropriate > for the node. You already have the crash recovery pattern + it's > multiplatform. But i'm curious what could be other solutions. Erlang would be a good solution if the volumes aren't too high. For high speed work it has to be C/C++ and the best crash recovery patterns IMO are (a) don't crash, (b) restart quickly, and (c) live-live redundant nodes producing redundant streams if you really can't afford a gap in the data. -Pieter _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
