User talk:Mirelatabra
Revision as of 17:15, 26 September 2008 by Mirelatabra (talk | contribs)
Mirela: (Answer) I think the node the node failure problem can be solved with checkpointing, by rolling back the execution of the failed node. I had read a paper that demonstrated a roll back recovery solution, in which each node will save its execution state, but it will not have the checkpoint of the shared memory.