Saturday, November 9, 2013

My notes from SOSP13 welcome and awards

The ACM Symposium on Operating Systems Principles (SOSP) is arguably the top conference in the computer systems area. The SOSP conference took a start with a welcome talk from the General Chair, Michael Kaminsky (Intel Labs), and PC Chair, Mike Dahlin (Google and UT Austin).

The chairs started by thanking to the many sponsors (platinum, gold, silver, bronze level) for the conference.

This year SOSP had 628 registrations, which made it the biggest SOSP as of yet, with a 16% increase over 2011 SOSP (which was yet biggest till then). Attendance distribution to SOSP is 76% from North America, 15% Europe, and 11% Asia. Among those attending 42% is faculty, 42% students, and 15% industry. There were 7 workshops on the weekend preceding the SOSP (LADIS was one of them), and 40% of attendants also attended workshops.

This year, for the first time, SOSP had full open access conference proceedings (the cost, $1100 per paper has been paid by SIGOPS), and this announcement got a huge applause from the audience.

Among the 150+ papers submitted, 30 papers are accepted to SOSP. There was a 3 round review process, with a total of 740 reviews, and on average 5 reviews per paper. 70 papers that fell in the middle were discussed in the depth PC meeting.

Three papers have been awarded with the best paper awards:

  1. The Scalable Commutativity Rule: Designing Scalable Software for Multicore Processors, Austin T. Clements, M. Frans Kaashoek, Nickolai Zeldovich, Robert Morris (MIT CSAIL), Eddie Kohler (Harvard)
  2. Towards Optimization-Safe Systems: Analyzing the Impact of Undefined Behavior, Xi Wang, Nickolai Zeldovich, M. Frans Kaashoek, Armando Solar-Lezama (MIT CSAIL)
  3. Naiad: A Timely Dataflow System, Derek G. Murray, Frank McSherry, Rebecca Isaacs, Michael Isard, Paul Barham, Martin Abadi (Microsoft Research)

The accepted 30 papers were presented within the duration of 2.5 days as single track. Each paper got a 30 minute time slot for which 22-25 minutes is for presentation, and 5-8 minutes are dedicated to question-answering session. This year Google-moderator tool has also been employed for taking questions from the audience, in addition to walk-to-microphone questions. The tool enables the audience to vote on the questions and get the questions with most votes rise to the top.

A majority of the presentations were very good (which is not the case in a typical conference where a big majority of the presentations are dull). It was clear that presenters had practiced extensively before the presentations, which helped them to deliver polished 22-25 minutes talks. Question-answering sessions were lively and there were several insightful questions asked.

The papers and  presentation slides (and soon talk videos) are available from   http://sigops.org/sosp/sosp13/program.html

Some of my top picks (reflecting my prejudicies and research interests) from the conference are as follows:

  • The scalable commutativity rule: Designing scalable software for multicore processors
  • Dandelion: a compiler and runtime for heterogenous systems
  • Sparrow: distributed, low latency scheduling
  • From L3 to seL4 what have we learnt in 20 years of L4 microkernels?
  • An analysis of facebook photo caching
  • Towards Optimization-Safe Systems: Analyzing the Impact of Undefined Behavior
  • Transactions chains: achieving serializability with low-latency in geo-distributed storage systems
  • Consistency-Based Service Level Agreements for Cloud Storage
  • Tango: Distributed Data Structures over a Shared Log
  • There Is More Consensus In Egalitarian Parliaments
  • Naiad: A Timely Dataflow System

I will try to edit/organize my notes about some of these talks and share them soon. Especially the last five papers on this list appeal a lot to me since they are more relevant and related to my research interests large-scale distributed systems.

Tuesday night there was a banquet and award ceremony. Stefan Savage has been awarded with the 2013 Mark Weiser award. He gave a humble yet powerful talk, and shared personal stories about how his career has benefited a lot from interacting with the late Mark Weiser. Two recent PhD thesis were awarded with the Dennis Ritchie award.

Finally, the following five papers have been added to SIGOPS Hall of Fame:

  1. “Tenex, A Paged Time Sharing System for the PDP-10”, Daniel G. Bobrow, Jerry D. Burchfiel, Daniel L. Murphy and Raymond S. Tomlinson, Communications of the ACM 15(3), March 1972. 
  2. “A NonStop Kernel”, Joel Bartlett,  in Proceedings of the Eighth ACM Symposium on Operating Systems Principles (SOSP’81), Pacific Grove, California, December 1981 
  3. K. Mani Chandy and Leslie Lamport, “Distributed Snapshots: Determining Global States of a Distributed System”, ACM Transactions on Computer Systems 3(1), February 1985. 
  4. Kenneth P. Birman and Thomas A. Joseph, “Exploiting Virtual Synchrony in Distributed Systems”, in Proceedings of the Eleventh ACM Symposium on Operating Systems Principles (SOSP’87), Austin, Texas, November 1987. 
  5. Eddie Kohler, Robert Morris, Benjie Chen, John Janotti and Frans Kaashoek, “The Click Modular Router”, ACM Transactions on Computer Systems (TOCS), 18(3), August 2000.

No comments: