views:

35

answers:

2

I have a lot of trivially parallelizable computations and a lot (100s) of cores distributed overs SSH + NFS network. What is the simplest way of parallelization. The problem is that I don't know how long each task will take so I need some kind of queue. Is there something that is very easy to use?

+1  A: 

Mincemeat.py is a lightweight Python implementation of MapReduce designed just for this purpose. There's not an easy way to launch all of the processes (unless you have something like Condor on the cluster), but otherwise it should fit your needs.

Michael Fairley
+1  A: 

You may also have a look at mpssh & autossh.

http://www.mynitor.com/2010/08/16/top-50-ssh-helper-tools-omg/

telld