This page contains assignment #3 for
Performance Evaluation. This assignment is due on Monday,
October 11 class. Any solutions that you find "on the web"
must be referenced as such. Please start each problem on a new
sheet of paper and put your name on all sheets of paper.
Download and compare ttcp and netperf. Try to find
two interconnected machines that you can make measurements between. you
can also make measurements "talking to yourself" (e.g., open two console
windows in one PC), but this is less desirable. Do ttcp and
netperf give similar throughput results for TCP/IP? Why or why
not? Try to configure the two tools to have the same parameters for your
experiments. For your solution please include screen shots of the execution
of the two tools.
Compare TCP and UDP throughput using either ttcp or
netperf. If there is a significant difference in throughput,
hypothesize on what might be the reasons. Discuss how you could test
your hypothesis.
Sketch a D/D/3/10 queue. For steady-state performance, how would
increasing the queue size affect loss and delay?
Kleinrock defines two regions of operation for a network - latency
limited and bandwidth limited. In words, what is the dividing line between
these two regions?
If a single server queue has a service rate of 10 customers per
minute and customers arrive at a rate of 12 per minute for 10 minutes,
what is the queue length at the end of 10 minutes? Assume that the queue
was empty at the start of the 10 minute period and that fluid flow assumptions
hold.
Consider an M/M/1 queue with service rate 10 customers per minute.
Plot the mean queueing delay (Wq) as a function of customer arrival rate.
What are the numerical values for mean queueing delay for the arrival
rates of 6, 7, 8, and 9 customers per minute?
