[BBLISA] shared network disks - vs gfs - vs distributed filesystem - vs ...

[Edward Ned Harvey]

> Similar to what John Hanks suggested...export each node's scratch via
> NFS and on each node combine all the remote scratch disks with the
> local
> scratch using unionfs, with of course the local disk getting priority
> for writes.

That sounds cool ... never heard of unionfs before ... Will have to check it
out.  :-)