Notices for Capacity Planning


This page contains general notices for Capacity Planning.


12/08/01 - The Final Exam will be on Tuesday, December 11th at 10:30 to 12:30 in the classroom. The instructions that will appear on the final exam are:
11/12/01 - The Final Exam will be on Tuesday, December 11th at 10:30 to 12:30 in the classroom.
11/02/01 - The instructions as they will appear on Exam #2 are....
10/18/01 - Zane Reynolds modified clktod.c to handle roll-over. This tools converts cumulative time stamps to delta. Here is the new and improved... clktod1.c.
10/11/01 - For HW #4 you are to determine if a time series of interarrival times (the interarrival times between ARP packets) is exponentially distributed (i.e., that the arrival process is Poisson). Some key properties of a Poisson process are... the CoV of interarrival times is 1.0. The autocorrelation of interarrival times is 0 for all lags. I suggest looking for these properties. Also, I suggest plotting a histogram of the interarrival times and seeing how close to an exponential distribution it appears to be. As always, the tools page is a good place to look for useful tools to help you with this assignment. Other good resources are efunda... here for Poisson distribution and here for exponential distribution.
09/26/01 - Here are the instuctions as they will appear on exam #1. Pay attention to the rules for the "formula sheet".
09/16/01 - Here is a description of what is autocorrelation. This should help you with the HW #2. Correlation is a way of measuring the dependence between two datasets. A high correlation means that the "Y" can be predicted from the "X". For example, age and height are correlate strongly for young children. The older the child, the taller they will be. Give the age of a child, I can fairly closely predict their height. Here is what Excel Help says about correlation... So, what is autocorrelation? Autocorrelation measure the dependence between values in a time series separated by a lag. For example, if lag is 10, then the autocorrelation determines the dependence between every value in the time series and the value 10 steps ahead of it. For example, the following time series would have a very high autocorrelation for lag 3 (and 6 and 9 and...): 
4, 5, 6, 4, 5, 6, 4, 5, 6, 4, 5, 6, 4, 5, 6, 4, 5, 6, ...
09/16/01 - I received a good question on "what should be the format of our HW #2". Here is the answer I gave... 
Like with all things... use common sense.  You have done many lab reports by
now... you should know how to present experiments and results.  "Make it nice"
sounds like you are going to give me a fancy cover.  This is NOT what should
be done.

You have some experiments to do.  Some of the results belong in tables (e.g.,
mean, variance, etc. for the datasets).  Other results belong in graphs (e.g.,
histograms and autocorrelation).  Everything needs to be presented so that
comparisons can be made between the I1 and I2 and between am and pm (comparison
is the ultimate purpose of this assignment, right?!).  All results should be
"explained" and insights made.  By insights, you need to state what meaning
the results have.

You should know how to present graphs by now.  All axes must be labeled.  Units
must be used.  The graphs must show something "interesting".  A graph showing
all white space (and any graphs with gray backgrounds will be deducted 15 pts...
you tell me why!) is not interesting.  Sometimes a log scale is called for.
Other times a "blow-up" of a critical portion of a graph is needed.
08/23/01 - This site is now "up".
Last updated by Ken Christensen on DECEMBER 8, 2001