The SIGOPS Business Meeting

(Download spreadsheet (MS Excel 97, 11 KB) with member data or view the data in html format.)

ACM SIGOPS Business Meeting 12/14/99, Kiawah Island Resort, SC

The meeting was called to order at 8:40pm.

The old and new boards were introduced. Old: Carla Ellis, Marc Shapiro, Paul Leach, Mary Baker. New: Bill Weihl, Valerie Issarny, David Kotz, ?. (both: chair, vice-chair, sec/treas, info director).

Satya was introduced as the program chair of the next SOSP.

John Wilkes proposed changing how the SOSP best paper awards are handled. Instead of forwarding the best papers to TOCS for inclusion in a special issue, simply denote those papers that deserve special merit. The idea is that a good conference paper is not necessarily a good journal paper. Authors can always choose to submit their papers to TOCS.

Larry Peterson (TOCS editor): SIGOPS currently provides financial support to TOCS in return for the special issue. One advantage of TOCS over SOSP is that it has more liberal page count. Perhaps SIGOPS support could be decoupled from the special issue?

Hank Levy (previous SOSP chair, associate editor of TOCS): The special issue is a benefit to the community, as some people get TOCS but not the SOSP proceedings.

Site for the next SOSP

Carla Ellis proposed Lake Louise in the Canadian Rockies. It's just big enough for us, and we would have the place to ourselves. It's about two hours from Calgary, and shuttles are available. It's lower than Copper Mountain, so the altitude shouldn't be as big a problem.

Andy Tanenbaum: One of the problems with Copper Mountain was the obvious distractions. Won't Lake Louise provide too many distractions?

Carla: In October there won't be much skiing. Even if there is, it isn't as convenient as Copper Mountain.

David Steere proposed Sun River, Oregon. It's a golf resort community about 30 minutes from the nearest airport (Redmond OR), and in October the temperatures are 46-62. It's a high desert (4000') and will be dry, although there could be snow in December. It is 20 minutes to Mt. Batchelor (sp?). Everyone would be housed within 5 minutes of the conference room. The conference facilities have recently been upgraded. He has a local committee lined up. http://www.sunriverresort.com

Andrew Black: The choice of date might be more important than the location. OOPSLA should be avoided.

Jeff Chase: I've been to both locations, and both are nice. Lake Louise, however, is probably the most beautiful place on the continent. It may be too expensive, though.

David Kotz: Choosing a location is a difficult process. The time of year, location, and price must all be considered. The general chair should be given latitude to make the decision.

A show of hands revealed a preference for Lake Louise.

Asilomar was suggested, but Mike Schroeder pointed out that it SOSP has grown in size and will no longer fit. Besides, the food isn't that great.

John Bennett: I chose Copper Mountain, but considered several resorts in the Rockies. It was very helpful to negotiate with several locations to get the best deal. The general chair should be given the flexibility to negotiate the place, time, and other details.

Guerney Hunt discussed the scholarship program. There was considerable debate in the scholarship committee as to what the criteria for receiving one should be. The committee should be given written guidelines to streamline the process, e.g., how much weight should be given to need versus merit. The current committee will produce a report, but in summary four industrial sponsors provided $10K each, and SIGOPS provided $20K. The committee consisted of 1 person from each industrial sponsor, plus Marc Shapiro representing SIGOPS, plus Ken Birman and Garth Gibson.

Marc Shapiro spoke about the European Workshop. We should decide the status of the workshop. One possibility is to evolve it into a conference and produce real proceedings that would be distributed outside the workshop. Another possibility is to keep it the way it is, but distribute the proceedings to the SIGOPS membership.

Andrew Tanenbaum: It's a question of format, not the number of people. We could make it more of a conference and less of a workshop. That way there would be an OS-related conference every year.

Hank Levy: Keep it a workshop. OSDI serves as the "off year" OS conference. Providing the proceedings to the SIGOPS members provides value to those people who can't attend the workshop.

Satya: Workshops and conferences serve different purposes. Workshops are a forum for provocative and less-finished ideas. Notes from the workshop are useful to have in the digest of proceedings, but then the proceedings cannot be distributed in advance.

Carla Ellis: Why not follow the HotOS model? All proceedings should be in the digital library, so that SIGOPS members have access.

Marc will investigate the cost of distributing proceedings to all members.

Jean Bacon spoke about the status of IEEE Concurrency. There is significant overlap with the SIGOPS community. Concurrency could provide a magazine format for SIGOPS, providing an additional publication forum without competing with TOCS. ACM and IEEE should be forced to collaborate.

Greg Minshall: IEEE and ACM serve different communities. Don't try to blur them. TONS (SIGCOMM) tried this and it's not working.

Andrew Black: Perhaps ACM members could be given access to the IEEE digital library. (Much approval for cross-licensing and discounts, people figured there was no harm in this sort of deal.)

Michael Scott: IEEE Concurrency is not only about OS, so there is some content that doesn't overlap.

Dejan Milojicic proposed a Workshop on Industrial Systems. OSDI and SOSP are mostly academic; the proposed workshop would address the needs of industrial folks. It would be held every year before these conferences, merging the academic and industrial communities. Would be organized by TCOS in cooperation with SIGOPS and USENIX.

Satya: It is valuable for academics to get input from the "trenches". The problem is that these meetings are not much benefit to the industrial people. For it to work, we have to make it easy for people to attend. Every year is too often.

(Note: Ken Birman made a comment here but Jay was talking to me and I didn't catch it. I apologize. -- jhh). (dfk: My notes say that "What should SOSP pair with, if anything? Birman prefers distributed systems conference... some support." But I don't know exactly what those notes mean.)

Werner Vogels: The workshop should be held in a convenient location, while SOSP typically isn't. It doesn't make sense to co-locate them.

It was proposed that we videotape future conferences. Jay Lepreau suggested that we audio record the invited talks, and anything else for which there isn't a written record.

Carla Ellis reported on the ACM Digital Library. It currently has 42,000 entries: 34,000 proceedings plus 8,000 journals. 30,000 entries are the complete text (23,000 proceedings and 7,000 journals), totalling 250,000 pages. Subscribers include 19,000 professional and 14,000 student members. Monthly activity includes 96,000 downloads, 1 million hits, 79,000 searches, 111,000 unique visitors, and 95 GB of traffic. Revenue sharing would be 70% to the general ACM and 30% to the SIGs.

The SIGs have already agreed to cover the costs of putting their conference proceedings from 1985-1990 into the digital library (est. $2K). An additional $6K will be required to go back as far as 1967. The Portal project is putting together an index of CS-related publications. SIGOPS would be responsible for OS-related material. SIGOPS members get access to SIGOPS content in the digital library even if you don't subscribe to the ACM digital library.

Carla: if you're a member of SIGOPS you should have access to all of our (SIGOPS) content in the DL, even if you don't subscribe to the DL.

Marc Shapiro: European connectivity to the ACM DL is poor. A European mirror is needed.

Carla: it's coming.

Andrew Tanenbaum: In addition to being on the Web, content should also be published on paper. Libraries are hesitant to subscribe to electronic publications such as the ACM DL because if the ACM goes away, so does the content. Paper doesn't go anywhere and never becomes obsolete.

SIGOPS budget (details attached in spreadsheet).

Hank Levy: One way to handle the large surplus is to lose money on everything. Unfortunately, the ACM requires conservative budgets for conferences such as SOSP, with the result that the conference often makes money. Keep in mind that SOSP costs about $200K, and we have to commit in advance without knowing how many people will actually show up. Our surplus really isn't that large; it's akin to having only enough money in your bank account to cover your next party.

Satya: If the balance is large enough, perhaps we could create an endowment.

(dfk: We're endowing the Mark Weiser award.)

Andrew Black: We should consider creating student fellowships like USENIX does.

Marvin Theimer: Use the money to put everything in the DL.

(dfk note: we've committed $8k to put all SIGOPS stuff into DL, back to 1967.)

Andrew Tanenbaum: Using the money to reduce the conference costs is not as important as increasing the DL content.

Hank Levy: The decline in SIGOPS student members is a real concern. Perhaps we could offer 200 students a free first year of SIGOPS with their ACM membership.

The meeting adjourned at 10pm.

John Hartman (jhh), with additions by David Kotz (dfk)