🚀 go-pugleaf

CS4753 and 5743 is a combined undergraduate & graduate class on
computer performance modeling. The cll numbers are 63362 for 4753
(undergraduate) and 63364 for 5753 (graduate). Undergraduates will be
graded on a different scale than graduate students, and will have a
lighter work-load.

In this class, you'll learn the tools of the trade of computer
performance analysis.  For example:

	+  if you have a large computer installation & are being paid to
	   improve the overall performance, what should you upgrade?
	   Memory?  Disk? CPU power? How do you decide what to
	   upgrade?

	+  Your CORBA-based appplication performance is slow.
	   Is it the network performance; should you upgrade
	   to 100Base-T networks? Buy a faster computer? Switch
	   CORBA implementations?
	
Answering each of questions, and the zillions like them faced by
computer scienctists & engineers, requires a variety of performance
analysis tools. You need to collect data, have confidence you've
collected enough and the right data and then draw inferences from that
data.

In this class, we'll be using a variety of "live data" sources,
including information from network monitoring tools (e.g.,
"traceroute"), hardware computer performance monitoring tools (i.e.,
hardware measured cache misses, etc) and possibly from O/S
instrumentation.

You'll learn how to do "capacity planning" and "bottleneck analysis"
to determine what is limiting performance in a system. You'll learn
enough about statistical analysis to be able to lie in public and get
paid for it.  You'll learn about computer-based simulation of
networks, software and hardware, and how to draw lessons from such
simulations.  You'll learn how to avoid doing simulation at all, using
techniques like mean-value analysis and queuing theory.

These skills are useful for anyone pursuing a graduate degree in
"systems", and historically, students will good performance analysis
skills are highly valued by employers.