presentation_planning.txt

1) Theoretical Paxos
	a) What problem Paxos solves
	b) Why that matters
	c) How it works
		i) "Live" Demo
	d) Questions?
2) Paxos Simulator
	a) Our Goals Starting Out
	b) Demo Simulator
	c) Explanation of code structure
3) Difficulties with the Simulator
	a) Overly ambitious goals for time given
	b) Understanding the Paxos protocol
		i) Classes like Decree were redefined several times as our understanding became clearer
		ii) Understanding time and clocks within the situation
		iii) Understanding three phase commit
	c) Amnesia between class periods (poor documentation habits)
	d) No ability to test until most of the code was written
	e) Bottleneck at the Leader
		i) Message explosions
		ii) REQUEST_LOG messages
4) Future improvements
	a) Implementing Clocks in Participants
	b) Adding the leader expirations and SET_LEADER decrees
	c) Allow for REJECT messages ("Dealing with Rejection")
	d) Spreading the load of log requests to avoid leader overload
	e) Limiting the number of duplicate accept messages
	f) Allow for user intervention during simulation runs
		i) Modifying queue data
		ii) Modifying the probability of different events happening
		iii) Additional Participant directions (Initiate Proposal, Set Leader, etc)
	g) Display Improvements
		i) Allow viewing multiple participants at once
		ii) Concurrency issues with Swing
	h) Allow user scripts detailing instructions for a simulation
	i) Allow the simulator to run over several machines
	j) Suggestions for improvements?
5) Questions?
6) A Word From Dr. Gray