Message from the Program Chair

Creating a new symposium is a major effort that in OSDI's case required a year and a half of attention, commitment, and negotiation from many in the community. Starting from observing the overlap of the three USENIX symposia on Mach, Microkernels, and SEDMS, we have come a long way. From now on, OSDI should occur every two years, sponsored by USENIX, with co-sponsorship by ACM SIGOPS and IEEE TCOS. My baby is born; from now on it's up to others to raise it well.

The birthing process was arduous. The review process was intended to encourage submissions of timely work, minimizing the time from submission to publication, while retaining high quality. It was all of: short, intense, stressful, massive --- but despite and because of that, high quality. I would not repeat it! An intense five week process culminated in choosing the 21 titles you find here. The original extended abstracts averaged about 10 pages in length, with a total of 178 submitted. This was considerably more than expected, so we scrambled, recruiting a large number of outside referees, and squeezing in a second one-week round of reviews for the top 80 papers. My gratitude to all of the referees and committee members who responded to our calls for last minute reviews. In the short review time, people contributed 736 formal ratings and reviews. More often than not, the reviews were lengthy and thorough. Besides aiding the selection process, these were major factors in helping the authors improve their papers. The median number of reviews for the second round papers was 5.5. Every paper was read by at least two program committee members, and usually three. Nearly every paper of the entire 178 had at least three reviews.

We accepted three papers co-authored by program committee members, out of 12 submitted. Committee authors did not see reviews or rankings of their own papers until the mass mailing after the meeting, did not know the fate of their papers until nearly the whole program had been selected, and of course were not present when their papers were discussed. In other cases of obvious conflict of interest, the affected members were to remain silent unless I asked them to speak.

Finally, every accepted paper was shepherded by a member of the program committee, with this interaction often resulting in substantially improved presentations or more extensive results.

This conference could not have been created without the efforts of a vast number of people. Even without the detail each deserves, listing them all approaches Academy Award level excitement for the reader. Above all, the small 10-person program committee did an outstanding job, generously giving their expertise, their time, and their prestige. Great thanks also to the referees, some of whom reviewed a large number of papers in a very short time. The authors of all submitted papers were the most important contributors of all, and took a greater chance than usual, on a new conference.

On the committee, Karin Petersen deserves special thanks for help in recruiting referees, as does Willy Zwaenepoel for being a source of excellent advice and help. I thank Dave Black, Ellie Young, Lori Grob, Steve Johnson, and the entire USENIX board for their confidence in me, and their support in merging the three small symposia and creating OSDI. Thanks to Lori for her excellent help throughout most of the process, as board liaison. Hank Levy was a valuable mentor to me throughout the process, especially regarding conflict of interest procedures. My thanks also go to Hank and the SIGOPS board for co-sponsoring OSDI, and, with USENIX's help, publishing the proceedings for the SIGOPS membership. I am grateful to Darrell Long, chair of TCOS, for their co-sponsorship and other aid, and to Margo Seltzer, for general advice.

Being program chair for a USENIX-sponsored event is normally easier than for most other conferences, because the USENIX professional staff are so excellent. First and foremost is Ellie Young, who was wonderful, and wonderfully effective, in every way. Judy DesHarnais is the other key person, whose huge experience ensures the conference mechanics go smoothly. I much appreciate their remarkable good humor in dealing with me and the stress I brought. Carolyn Carr deserves a medal for patience in dealing with the proceedings and my last minute ways. Toni Veglia and others at USENIX helped in a variety of important ways, as always. Dan Klein and the tutorial presenters are crucial in making this conference of broad appeal. Rob Kolstad and Jeff Mogul contributed the software we used to run the review process. I am sure there are several others to thank that elude me this moment.

Finally, special thanks go to some great people at Utah: above all, my kitchen cabinet of Doug Orr, Bryan Ford, and Mike Hibler, who digested vast numbers of papers. I am grateful to Sean O'Neill for support of the review software and mechanics, to Joe Hitchens for a long night designing a snazzy cover (that unfortunately, USENIX did not use), and to Loretta Cruse, Steve Clawson, and Linus Kamb for administrative support. Finally, my best care, thanks, and apologies to all who lived and worked with me while I spent too much of my life on this. You know who you are.

Jay Lepreau
October, 1994