[Twisted-Python] twisted, SMP and clusters
dfarrell at mypublisher.com
Mon Dec 17 14:44:55 EST 2007
I'm considering using Twisted to build a soft realtime system to control a workflow for a physical production facility. Although there would be some socket work for it to do (XML-RPC, logging, etc) I also like it because of the nice way it can be daemonized. One of my goals for this project is to take better advantage of our multiple Linux servers, each of which has multiple CPU's with dual cores. I'd like to take our workflow objects that are fairly long running state machines and cut them loose on other cores/CPUs/servers if I can. So here's my question: should I be looking at twisted.spread for this kind of work? Can the Parallel Python or Pyro projects be incorporated into Twisted, and if so, what kinds of things should I watch out for, or is it even possible? I'm thinking if I use Parallel Python or Pyro I could make remote calls to our objects in a DeferToThread object and get back the results/errors via callbacks. I'm still really pretty new to Twisted, so forgive my naïve questions if they are so.
Thanks in advance,
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Twisted-Python