I'm wondering wether someone is working on mechanisms that would help
build a HURD cluster, that supports network memory and load balancing.
The kind of functionality I have in mind is somewhat similar to Tanenbaum's
Amoeba System.
Mach has some unique features that were already used in research of load
balancing and if they were enabled/enhanced in gnumach, some nice translators
could be added to the Hurd, like a node scheduler for a task, which may
always choose the less busy node, probably even migrating an already
running task to another host etc... Another application of those features
may be redundancy algorithms, that have the same tasks running on many nodes,
checkpointing them, so that even if one(some) node crashed, the task would
still continue running on other nodes.
The Hurd seems naturally suited to research in distributed computing,
since the main mechanisms can be implemented in user-space.
--
Farid Hajji -- Unix Systems and Network Administrator | Phone: +49-2131-67-555
Broicherdorfstr. 83, D-41564 Kaarst, Germany | [EMAIL PROTECTED]
Fermat: ...I've found a remarkable proof for this: Let x,y @#$!@$!2@ NO CARRIER
