How to do research in Experimental Computer Systems Lab ?

Reference

Outline

• Goal: Put Stony Brook on the map of experimental systems research
  
• Approach: Build the "real stuff"
• My responsibilities as advisor: sort of like an NBA coach
  • Teach you how to play the game as a whole, assuming you know how to dribble
  • Create a stimulating research environment
  • Come up with good topics for you to work on
  • Show with example how to do research
  • Lead you through the adventure
• Your responsibilities as graduate student:
  • Educate your professors, including me
  • Occasionally drop by my office and say ``what if we do X...''
  • Learn from one another
  • Come up with possible solutions when stuck
  • Perform experiments and analysis
• Essential ingredients that make a good graduate student: (With my personal bias)
  • Absolute enthusiam in computing technology
  • Common sense
  • Independent problem solving skill
  • Responsible for assigned work
  • Responsive to feedbacks and criticism
  • Clear mind
• Preparation:
  • Motivation
    Curiosity > Prove yourself > Impress your advisor > Get a degree
    You should be in this for yourself, not for anyone else.
    
  • Ambition
    Do the impossible > Build a cool widget > Just get by
    A sense of purpose and urgency
    It's YOUR project, i.e., your job counts on it!!
    
  • Hard Work
    Must be prepared to sustain a period of 12-18 months of >= 12 hrs/day, 6 days/week work
  • Strategy
    How to implement a system on schedule? Prioritization
    What is the structure and tone of a paper?
    Good writing and verbal communications skills are CRITICAL
    Build up your connections whenever possible
    Cooperate by exploiting local talents
    To walk on top of waters, you have to know every stone underneath
    TRANSLATION: every impossible-looking thing has tricks!
    
  • Broad
    Read BYTE, EE Times, and other trade rags
    Computer science is too young a field to be divided into subfields
    Must have the capability to learn all the related math and algorithms and work on a new area QUICKLY
    
  • Smart
    It's not given, i.e., it can be learned
    Experience and taste are what really counts in our business
    The nature of this group's research dictates that the speed with which we built a prototype is at least as important as the speed of the prototype itself
    
  • Solid
    Excellence is expensive
    The last 10% to perfection typically consumes 80% of the effort, but is also what distinguishes between mediocrity and GREATNESS
    The devil is ALWAYS in the details
    
• When you are done, you should master the skills of
  • Designing and implementing novel systems,
  • Analyzing the corretness and performance of working systems and improving them,
  • Putting in clear writing how a complicated system works, and
  • Coming up with interesting project ideas with excellent taste.
• General Advice:
  • Read widely
  • Learn to communicate well
  • Be a team player
  • The name of the game for being a graduate student is efficient time management
  • Always make plans ahead
  • Talk to knowledgeable people regularly
  • Practice solving problems and building systems whenever you have a chance