Network Performance: Determining Throughput and Call Set Up Time in ATM Networks, Assignments of Communication

Instructions for determining the throughput performance and call set up time in an atm network using the atm forum performance benchmark suite. It includes details on conducting experiments for two traffic scenarios, analyzing results from a 2k factorial experimental design matrix, and simulating a printer system.

Typology: Assignments

Pre 2010

Uploaded on 09/02/2009

koofers-user-yk5-1
koofers-user-yk5-1 🇺🇸

10 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
TELCOM 2120 – Network Performance
Homework 4 Spring 04
1. You are asked to determine the throughput performance of an ATM switch using two of the
ATM Forum Performance Benchmark Suite. Specifically you are asked to find a 90% confidence
interval with 5% relative precision on the Peak throughput and associated mean cell loss rate of
the switch for the following traffic scenarios.
1) Each input port has a multicast stream destined for the other output (i.e., (N-1) output)
ports
2) Each of the n input ports has a traffic source creating traffic destined for the same
output port (N to 1 stress test).
In each case the traffic source is a Variable Bit Rate connection with a maximum burst size of
50 cells. Describe a set of experiments you would conduct to achieve this. This is an open ended
problem with no one set answer as is the case in most real world experimental studies. Follow
the concepts of experimental design.
2. The 2k Factorial experimental design matrix below shows the results of a measurement study
on switched virtual circuit setup times in an operational multi-switch MAN ATM network. The
factors are
A: the number of hops between source and destination node (measured by number of
switches.)
B: the number of peer groups the switches organized into.
C: the number of destinations per source connection (amount of multicasting.)
For the factors the following level values were used
A: -1 level (1 hop) +1 level (5 hops)
B: -1 level (1 peer group) +1 level (2 peer groups)
C: -1 level (1 destination - unicast call) +1 level (4 multicast call)
The response variable is the call set up time - measured over an average of 10 experiments where
each experiment consisted of setting up and tearing down 400 switched virtual circuits.
A B C Call Set Up Time
(milliseconds)
-1 -1 -1 46.5
+1 -1 -1 240.7
-1 +1 -1 50.2
+1 +1 -1 590.0
-1 -1 +1 201.6
+1 -1 +1 1076.5
-1 +1 +1 880.4
+1 +1 +1 4449.0
(a) Determine the main effects and the interaction effects.
(b) Explain what you have observed based on values obtained in (a).
(c) Construct a regression model to predict a call set up time.
pf2

Partial preview of the text

Download Network Performance: Determining Throughput and Call Set Up Time in ATM Networks and more Assignments Communication in PDF only on Docsity!

TELCOM 2120 – Network Performance Homework 4 Spring 04

  1. You are asked to determine the throughput performance of an ATM switch using two of the ATM Forum Performance Benchmark Suite. Specifically you are asked to find a 90% confidence interval with 5% relative precision on the Peak throughput and associated mean cell loss rate of the switch for the following traffic scenarios.
    1. Each input port has a multicast stream destined for the other output (i.e., ( N -1) output) ports
    2. Each of the n input ports has a traffic source creating traffic destined for the same output port ( N to 1 stress test). In each case the traffic source is a Variable Bit Rate connection with a maximum burst size of 50 cells. Describe a set of experiments you would conduct to achieve this. This is an open ended problem with no one set answer as is the case in most real world experimental studies. Follow the concepts of experimental design.
  2. The 2 k Factorial experimental design matrix below shows the results of a measurement study on switched virtual circuit setup times in an operational multi-switch MAN ATM network. The factors are A : the number of hops between source and destination node (measured by number of switches.) B : the number of peer groups the switches organized into. C : the number of destinations per source connection (amount of multicasting.) For the factors the following level values were used A : -1 level (1 hop) +1 level (5 hops) B : -1 level (1 peer group) +1 level (2 peer groups) C : -1 level (1 destination - unicast call) +1 level (4 multicast call) The response variable is the call set up time - measured over an average of 10 experiments where each experiment consisted of setting up and tearing down 400 switched virtual circuits. A B C Call Set Up Time (milliseconds) -1 -1 -1 (^) 46. +1 -1 -1 (^) 240. -1 +1 -1 (^) 50. +1 +1 -1 590. -1 -1 +1 (^) 201. +1 -1 +1 (^) 1076. -1 +1 +1 880. +1 +1 +1 (^) 4449. (a) Determine the main effects and the interaction effects. (b) Explain what you have observed based on values obtained in (a). (c) Construct a regression model to predict a call set up time.
  1. Jobs arrive to a computer printer with a time between arrivals of either .5, 1, 1.5, 2, 2.5, 3 minutes at equal chances of occurrence. The time to print the jobs is either 1 or 2.4 minutes. Using a die and a coin, generate job arrival and service times and simulate the system by hand for 15 minutes of operation. Show the event list and plot the number in the system vs. time. Find the average time in the system for jobs and the average printer utilization. Note that, every student should get slightly different answers as it is unlikely that everyone will generate the same sequence of random arrivals and service times.
  2. Consider a simulation of a system with two identical servers labeled server 1 and server 2 respectively. A customer arriving to an empty system will always select server 1. Defining the states of the system as: n the number of customers in the system, b1: busy or idle status of server 1, and b2: busy or idle status of server 2. The events are the job arrivals and departures. Let b1 = 1 busy, b1 =0 idle, b2 = 1 busy, b2 =0 idle, e_i = type of i th event, e_i = 1 arrival,

e_i = 2 departure.

(a) Given the portion of an event list and the state variables values shown below determine the interarrival time of jobs 1-5 and a set of possible service times

clock event type n b2 b

0 - 0 0 0 3 e_1 = 1 1 0 1 5 e_2 = 1 2 1 1 6.2 e_3 = 1 3 1 1 7.4 e_4 = 2 2 1 1 9 e_5 = 2 1 0 1 10.5 e_6 = 1 2 1 1 12 e_7 = 2 1 1 0 14 e_8 = 1 2 1 1 15 e_9 = 2 1 0 1 17 e_10 =2 0 0 0

(b) Determine the mean utilization of each server and the mean number in the system over the

course of the simulation. (c) Determine the mean delay in the system.