sharing indirect values for parallel processing?


I am going to start some large renderings during the next weeks, and I would like to have a little park of rendering machines available by the time. Now, parallel processing has been a topic here for some time, and I am wondering about the changes in the recent developments.

I know there is the new -N switch of rad, which will distribute renderings to multiple processes, right? So this seams to be the frontend to use, maybe better than starting all the rpict & Co.s manually. What I would really like to know what file sharing mechanism should I use for indirect values? It used to be nfs, and the documentation in the radiance distribution still reflects how to use nfs and lcking to share data between processes. Would the same be possible e.g. with samba? I am asking because I have a working samba 3 server available, offering enough space and lots of bandwidth. If samba is not supported, I could still install a nfs server, however, I remember some trouble with different nfs-locking mechanisms and the os-dependance of nfs implementations (we use OS X, Linux and Solaris here...;-).

So if there is any news about this topic, maybe some reports from people who use radiance in parallel environments, it would be of great help.

TIA+CU, Lars.