Items for 6-10 meeting (besides Chapter 4 feedback from Rick)



-	Related Work (3.4)

	Can break down by fault tolerance structuring paradigm!!!!
	(Hilights the fact that FT-Linda is the only one based on
	the state machine approach...)

	-	Object/Action 

		- Transactional Linda: Plinda, MOM
			[ need to reread MOM; Plinda doesn't seem to
			replicate checkpoints/TSs, and assumes all
			processes should be resilient.]
			



	-	Primary/Backup

		-	[Xu/Liskov only provided stable storage, not
			any recovery for processes as with primary/backup.
			Ditto for Patterson et. al. and Chiba et al]

			==> Say insufficient as in Segall, and seem
				to be more expensive than FT-Linda


		-	OGI (unimplemented one) IS SUFFICIENT!
			[ Dave needs to reread/rescan ]

			==> Say seems very expensive
			==> Say assumes processes must be  resilient.
			    This is a high cost for a bag-of-tasks
			    worker that does not need to be resilient
			==> No provisions for atomic execution.  User
			    can construct with semaphore tuple, but
			    this would be a hot spot and seems
			    a lot more expensive than the AGS