Buffer Simulation

##Antoine GARGOT - CS 555 - Project

Buffer simulator for Analytic Models and Simulation of Computer Systems

Simulation of a M/M/2/2+5 system (i.e. 2 servers with the common buffer of capacity 5) by means of discrete event simulation. Investigate the following two cases:

• System A: arrival rate lambda=2, service rate of each server mu=1
• System B: lambda=10, mu=1

Test your random number generator (RNG):

• Does your RNG generate random numbers?
• How do you initialize the seed of your RNG?
• Generate two sequences of 1000000 numbers each, for every sequence use a different seed. Are these two sequences different? How do you know this?

Eliminate the warm-up period separately for each of the following cases by applying the Welch graphical procedure: * System A with the initial condition x(t=0)=0, i.e. the system is empty at t=0. * System A with the initial condition x(t=0)=7, i.e. the system is full at t=0. * System B with the initial condition x(t=0)=0, i.e. the system is empty at t=0. * System B with the initial condition x(t=0)=4, i.e. there are 4 customers in the system at t=0.

Change the seed of your RNG if you run multiple simulations to investigate a nonstationary state.

For the stationary regions of the Systems A and B identified in item (2) find:

• Blocking probability.
• Mean time a customer spends in the system.
• Mean number of customers in the system.

Find the confidence intervals for (3.1), (3.2) and (3.3). Assume the 90% confidence level.