Notes from the breakout session on graduate course

1
Notes from the breakout session on graduate course

• Summary
• Not much agreement on anything
• A lot of discussion about a few things with bent
  on performance evaluation, protocol design,
  experiments, case studies, etc

2

• What are foundational courses (concepts?) that
  prepare students for research or industry career
  in networking?
• (Graph theory, queuing theory, random processes)
  vs (Distributed systems, operating systems)
• Experimental methodology (incl simulation,
  designing experiments), ability to design an
  experiment, comparing expected behavior vs
  simulated behavior
• experience building a system incl design
  tradeoffs and perf characterization
• Theory of protocol design
• Modeling a system (abstract vs concrete?)

3

• Case studies
• Study RFP, RFP responses, evaluation
• E.g. state of KS chooses telecom switches
• what if scenarios and trying them out in
  simulation or experimental systems
• Cant do all of perf evaluation thing in a
  networking course?
• Choose one problem (e.g., routing algorithm) and
  then walk students through a complete exercise of
  experiment design and evaluation to prepare them
  for subsequent evaluations
• Case study of a large system could also achieve
  it
• Two course sequence vs one course
• Not an agreement
• If we assume an undergrad course, one course
  should cover a set of important concepts
  otherwise becomes a breadth-oriented course