📑 Papers
Responses and Discussions
As part of the technical reading and communication goal of this course, you will read one technical systems paper each week. Your job is to:
- Read the paper. See how to read. While you should aim for a general understanding of its key contributions and impact, understanding of every detail is not expected.
- Lead: For 1 paper, with a group of up to 4 students:
- Prepare a short joint introductory presentation (about 3 minutes) about the broader systems area in which this paper arose. There are many ways to frame this background; check in with me ahead for ideas.
- Develop a set of discussion questions that will be used for breakout group discussion of the paper, introduce those questions before we begin breakout discussion.
- The group may divide this work as they wish; each team member should contribute in roughly similar measure.
- Sign up for a spot on this sheet. You may pre-arrange a group or just put your name down and see who joins you.
- I am available for advice and feedback while you prepare these parts. My intro and questions for the first paper discussion aim to set the tone.
- Response: For at least 4 papers (the plan is to read 6), write
a short paper response (200-400 words) and submit on
gradescope by the start
of class the day the paper discussion is scheduled. The response
should:
- Briefly summarize the key contributions or findings of the paper.
- Discuss at least one strength that makes you enthusiastic about the value of the paper’s contribution.
- Discuss at least one limitation that makes you skeptical about the value of the paper’s contribution.
- Describe how this paper changes your thinking about building systems.
- Participate in paper discussion on the scheduled day.
Evaluation
Grading for the Lead and Response elements is on a checkmark basis with the intent to reward critical engagement with the ideas in the paper. (Notice fully understanding the paper is not required; evidence of thinking about it is.)
- check-plus: especially effective
- check: successful (expected to be typical)
- check-minus: incomplete attempt that does not address all required elements
- zero: no submission / participation
“Check” or above yields full credit for the relevant course grade component. When averaging paper responses, a check-plus cancels out a check-minus. Two check-pluses cancel out a zero.
Paper List
See also: leader signups, response submissions.
- The UNIX Time-sharing System.
Dennis Ritchie and Ken Thompson. Communications of the ACM, 17 (7), July 1974.
Discussion date: Thursday, 29 Oct - Hints for Computer System Design.
Butler Lampson. Symposium on Operating Systems Principles, 1983.
Discussion date: Thursday, 5 Nov - Managing Contention for Shared Resources on Multicore Processors.
Alexandra Fedorova, Sergey Blagodurov, Sergey Zhuravlev. Communications of the ACM, 53 (2), February 2010.
Discussion date: Thursday, 12 Nov - Eraser: A Dynamic Data Race Detector for Multithreaded Programs.
Stefan Savage, Michael Burrows, Greg Nelson, Patrick Sobalvarro, Thomas Anderson. ACM Transactions on Computer Systems, 15 (4), November 1997.
Discussion date: Thursday, 19 Nov - The Design and Implementation of a Log-Structured File System.
Medel Rosenblum and Jon K. Ousterhout. ACM Transactions on Computer Systems, 10, February 1992.
Discussion date: Tuesday, 1 Dec - Xen and the Art of Virtualization.
Paul Barham, Boris Dragovic, Keir Fraser, Steven Hand, Tim Harris, Alex Ho, Rolf Neugebauer, Ian Pratt, Andrew Warfield. Symposium on Operating Systems Principles, 2003.
Discussion date: Monday, 7 Dec