Assignment #5 for Computer Simulation (Spring 1999)


Situation:

From the previous assignment you discovered that the simpler TurnTaker was better than the more expensive ShortQ method. You thus decided to purchase a TurnTaker. Now, the vendor for the ShortQ (who is a golfing buddy of your manager) is questioning your results, did you run your simulation long enough? Are your results accurate?

Requirement:

You are to change your previous simulation (or use the posted solution if your solution was incorrect) to have as its stopping criterion convergence on queue length for one of the servers to a 95% confidence interval with a 1% accuracy. Re-run your simulation and conclusively prove (or dis-prove!) that TurnTaker is better than ShortQ.

Guidelines and hints:

You can use mm1a.c as a guideline of what needs to be changed to as3_99_4sol.c. You will need to change one line of code in bmeans.c to achieve 1% accuracy at stopping. It is good experience for you to read through the ugly and obtuse code in the SMPL source, if for no other reason than to convince you to write cleaner code yourself!

Extra Credit (10 points):

Make an argument (and show it with real simulation results!) that ShortQ is better than TurnTaker for a "real" Web server.


Last updated by Ken Christensen on MARCH 7, 1999