acm sigcomm

SIGCOMM FY’18 Annual Report July 2017 - June 2018

Submitted by: R. Guerin, Chair

SIGCOMM is ACM's professional forum for the discussion of topics in the field of communications and computer networks, including technical design and engineering, regulation and operations, and the social implications of computer networking. SIG members are particularly interested in the systems engineering and architectural questions of communications.

SIGCOMM continues to be a vibrant organization serving a broad community of researchers from both academia and industry interested in all aspects of computer networking. We sponsor several successful, single-track, high-impact conferences, several of them in cooperation with other SIGs. There are a number of highlights to report from the past year.

Conferences

The SIG sponsors an eponymous flagship conference and is the sole sponsor of the following conferences: CoNEXT,  eEnergy, Information-Centric Networking (ICN), and the HotNets Workshop. The eEnergy conference will, however, migrate to the recently formed Energy EIG in 2019.  The SIG also jointly sponsors the following conferences: Internet Measurement Conference (IMC), SenSys, ACM/IEEE Symposium on Architectures for Networking and Communications Systems (ANCS), Symposium on SDN Research (SOSR), and ANRW, the joint ACM, ISOC, IRTF Applied Networking Research Workshop.

Our flagship conference, continuing our policy of rotation among regions on a 3-year cycle, was held in Los Angeles, CA, in August 2017. The conference had an attendance of 785, which,  after the lower attendance of 2016 held in Brazil where the Zika scare had affected attendance, was back in keeping with participation levels of previous years. This fact, together with strong support from industry sponsors produced a healthy surplus for the conference while allowing us to at the same time award over $60k in travel grants to attend the conference.  

This relatively large surplus together with more modest surpluses from other conferences we sponsor meant that even if the SIG continues to seek to down-spend its large financial reserve, it continues to remain financially extremely strong.  This strength will allow the SIG considerable freedom in continuing to expand its support for the community.

In particular, one of the SIG’s main focus area is on outreach and diversity, with a new Director of Diversity and Outreach, Marinho Barcellos, having been added to the EC.  In addition to Marinho’s responsibilities on exploring new initiatives to foster greater diversity in the SIG and its conferences, the SIG has also been expanding its support of several activities aimed at increasing diversity.  They are further detailed in the section titled “Support for the community and new projects” below. The 2017 SIGCOMM conference also provided the opportunity to recognize a long-time SIG volunteer who recently passed away, Chris Edmondson-Yurkanan.  A travel grant named in Chris’ honor was introduced and will be awarded each year to a young woman, preferably but not necessarily a student, who has demonstrated the kind of enthusiasm and eagerness to become involved in the SIG and its activities that was Chris' trademark.  The travel grant will be awarded by the SIG’s liaisons to the N2Women organization.

As in previous years, the SIG continued to financially support several regional conferences in computer networking. The current set of regional conferences we support financially includes COMSNETS, a major networking conference in India, and the Asian Internet Engineering Conference (AINTEC).  COMSNETS has been quite successful and become a strong regional event in its own right, while AINTEC has remained somewhat confined in realizing its original goals of broadening participation in the Asia-Pacific region. This may have been in part because the conference has remained anchored at the same location.  As part of the feedback given to the organizers, the SIG recommended exploring alternative locations in the future. Subsequent support from the SIG may, therefore, hinge on whether or not this advice is followed. The SIG had previously been supporting the Latin American Networking Conference (LANC), but decided to withhold its support this year, as feedback from our community indicated that the conference had drifted away from its original goals and was not viewed as particularly relevant anymore.  We are in the process of exploring starting an new event in Latin America that will better fulfill our goal of better serving and growing our community in that region.

We also continued to support a new conference focussed in Asia-Pacific, APNET, whose first edition was held in Fall 2017. We continue to foster the success of these conferences by means such as invited speaker travel funds and student travel grants. In addition to supporting regional conferences, the SIG provides generous general student travel support to all of its sponsored conferences.

We are in-cooperation with a number of events, including the International Conference on Network and Service Management (CNSM), the International Teletraffic Congress (ITC), Multimedia Systems (MMSys), the USENIX Symposium on Networked Systems Design and Implementation (NSDI), the Workshop on Network and Systems Support for Games (NetGames), the International conference on Networked Systems (NetSys) besides the aforementioned COMSNETS, AINTEC, and APNET.

The SIG also supports a handful of summer schools with grants of up to $25k, including the PhD School on Traffic Monitoring and Analysis (TMA) that was held in Vienna in June 2018, and the ...MeetGreen has continued to provide administrative support to our volunteers, but we are exploring possible alternatives that may allow us to offer additional services to our volunteers, including a menu of options that conference organizers could chose from.

SIGCOMM Survey and Its Summary

Towards better understanding a number of questions that had been raised regarding exploring possible changes to the SIGCOMM conference process, we conducted a survey shortly after the 2017 edition of the conference.  

Specifically, the submission deadline for SIGCOMM papers is currently near the end of January. Reviewing starts soon after, and takes place in multiple rounds. Papers that have positive reviews at the end of the first round are promoted to round 2; others are rejected. Papers in round 2 get additional reviews. There may optionally be a third round. Finally, papers with sufficient support at the end of the final round are discussed in a 1.5-day long in-person PC meeting, where the final program is determined. Notifications of acceptance are sent around the end of April. The program typically has roughly 40 papers presented in a single track fashion. The conference takes place in August.  The survey focused in part on understanding better what aspects of this process work for both authors and reviewers, and what aspects, if any, don't. The broad goal was to continue to improve the conference and the review process.

The survey was released in October 2017, and stayed open for two weeks. Roughly 170 responses were received. To put this in context, this is about 10% of the overall SIG membership.

The results from the survey are publicly available at http://pages.cs.wisc.edu/~akella/sigcommsurvey/, and have been shared broadly both within the SIGCOMM and elsewhere (e.g., the NSDI community). The SIGCOMM Technical Steering Committee and 2018 PC chairs were also provided access to the survey results.

Key conclusions from the survey are:

• 60% of the respondents were supportive of SIGCOMM adopting a revision process.
• 64% of the respondents were supportive of SIGCOMM employing multiple rolling deadlines as opposed to a single annual one.
• 54% of the respondents were supportive of SIGCOMM adopting a journal model similar to PACM.

We also received several detailed comments from respondents about what aspects of the current process they like or dislike the most. Many respondents like the fact that SIGCOMM reviews are detailed, and that SIGCOMM program committee meetings provide a high-bandwidth avenue to ironing out major differences of opinion among reviewers. On the flip side, the fact that rejection means waiting a whole year before resubmission was brought up as a major source of concern with the current process, especially for young/untenured researchers in the community.

Newsletter

The SIG’s newsletter, Computer Communications Review (CCR), continues to publish four issues per year. Since 2016, CCR is published entirely on both the ACM Digital Library and at https://ccronline.sigcomm.org. We publish two types of articles: technical papers and editorial contributions. The editorial contributions range from meeting reports to reflections on the evolution of the field. Technical papers are peer-reviewed by members of the editorial board and external reviews. In order to encourage the authors to release their artifacts (software, datasets, measurements, ...) the papers that release their artifacts can be longer than the regular six-pages limit for technical papers. We observe that a growing fraction of the accepted technical papers provide artifacts that enable readers to extend and reproduce the results describe in the paper. Every year, the CCR editorial board selects the two best papers for oral presentation during the SIGCOMM conference. This year, the two best papers have released their artifacts.

A recent survey conducted among the authors of the accepted papers at SIGCOMM’17, CoNext’17, IMC’17, and ICN’17 revealed that one third of the accepted papers contained artifacts. Despite ACM's reproducibility initiate, these artifacts have not been reviewed. SIGCOMM has launched an artifact evaluation committee that will evaluate artifacts of accepted papers and assign the ACM reproducibility badges. A first evaluation of SIGCOMM paper artifacts will be organised this fall.

Awards

SIGCOMM Lifetime Achievement Award:  This year, SIGCOMM recognized Prof. Jennifer Rexford from Princeton University with the SIGCOMM Award for Lifetime Achievement; She will receive the award and present a keynote talk at the annual SIGCOMM conference in August 2018 in Budapest, Hungary. She was recognized "For her fundamental and practical contributions to making the Internet more reliable and predictable, and for her outstanding mentoring and community service.”

The award committee consisted of Bruce Davie (VMware), Albert Greenberg (Microsoft), Nick McKeown (Stanford University, chair), Nina Taft (Google), Don Towsley (University of Massachusetts Amherst)

SIGCOMM Doctoral Dissertation Award for Outstanding PhD Thesis in Computer Networking and Data Communication: The award for the best doctoral dissertation submitted in 2017 went to Anirudh Sivaraman Kaushalram for his thesis titled “Designing Fast and Programmable Routers.” His thesis was cited as follows: “Sivaraman's dissertation makes pioneering and impactful contributions to the design and implementation of programmable routers that run at hardware line rates.”

The committee consisted of Marco Mellia, T.S. Eugene Ng (chair), Xiaowei Yang, and Haifeng Yu

SIGCOMM Rising Star Award: The recipient of the 2017 SIGCOMM Rising Star Award was Mohammadreza Alizadeh Attar from MIT.   The award was in recognition of outstanding research contributions, early in his career, in the area of large scale datacenter network architectures and protocols.

The 2017 rising star award committee consisted of Sujata Banerjee (VMware Research), Sylvia Ratnasamy (UC Berkeley), Peter Steenkiste (CMU), and Balachander Krishnamurthy (AT&T Research).

The criteria for subsequent awards were also modified to specify a relative period of time since obtaining a Ph.D. rather than an absolute age limit.  Specifically, the modified description of the award mow reads as follows:

Each year, ACM SIGCOMM presents a "Rising Star" Award, recognizing a young researcher - an individual within 10 years of receiving their Ph.D. or equivalent degree  - who has made outstanding research contributions to the field of communication networks during this early part of her or his career.  Depth, impact, and novelty of the researcher's contributions will be key criteria upon which the Rising Star award committee evaluate the nominees. Also of particular interest are strong research contributions made independently from the nominee's PhD advisor.

SIGCOMM Test of Time Paper Award: Two papers were selected for the 2018 award by a committee composed of John Byers (Boston University, chair), Krishna Gummadi (MPI), Brad Karp (UCL), and Lili Qiu (University of Texas).  The two papers were:

• “A Scalable, Commodity Data Center Network Architecture” by Mohammad Al-Fares, Alexander Loukissas, and Amin Vahdat. SIGCOMM 2008.          This paper lucidly articulates a vision for what is today the standard structure of a data center network: commodity packet switches interconnected in a fat-tree topology. By posing and addressing practical challenges in data center networking, the authors drew attention to this then-emerging area, and propelled the community to consider the design of new networking techniques for the relatively ‘green field’ of the data center -- a research area that has flourished since.

• “XORs in the air: practical wireless network coding” by Sachin Katti, Hariharan Rahul, Wenjun Hu, Dina Katabi, Muriel Médard, and Jon Crowcroft. SIGCOMM 2006. This paper's interdisciplinary team brought information-theoretic research on network coding to bear on the domain of wireless networks, with an ingenious new scheme for achieving capacity gains by jointly coding information from multiple flows into individual packets. This paper's design and implementation upended the networking community’s understanding of limits on wireless capacity and ushered in diverse work on capacity improvement in the decade that followed.

SIGCOMM Networking Systems Award: This award was awarded for the first time in 2018 by a committee comprised of: Edouard Bugnion (EPFL), Ratul Mahajan (Intentionet), Jeff Mogul (Google, chair), and Ellen Zegura (Georgia Tech).  The committee selected the following contribution:

The Akamai Content Delivery Network (CDN):  The Akamai CDN pioneered the concept of a content distribution network, combining numerous technical innovations with an equally innovative business model that simultaneously met the needs of multiple stakeholders (site owners, ISPs, and users).  Akamai’s technical contributions include a system for mapping clients to the best CDN server, active probing to create a latency model of the Internet, and a dynamic control system that provides load balancing and fault tolerance. In particular, the paper "Consistent Hashing and Random Trees: Distributed Caching Protocols for Relieving Hot Spots on the World Wide Web" (STOC ‘97) provided a deep algorithmic basis, introducing random cache trees for load-balancing, and consistent hashing to minimize churn. With its enormous worldwide scale, the Akamai CDN is an exemplary study in translating research results into a successful operational system.

Contributors: Mike Afergan, Andy Berkheimer (YouTube), Bobby Blumofe (Akamai), Bill Bogstad, Chad Brown, Tim Canfield (Akamai), Alex Caro (Akamai), Rizwan Dhanidina (Akamai), John Dilley (Rafay Systems), Hilla Dishon, Ken Iwamoto (Akamai), Chris Joerg (Akamai), Vinay Kanitkar (Akamai), David Karger (MIT), Brian Kim (Alpine Global), Robert Kleinberg (Cornell University), Sef Kloninger (YouTube), Will Koffel (Google), Leonidas Kontothanassis (Google), Bradley Kuszmaul (Oracle), Tom Leighton (Akamai/MIT), Charles Leiserson (MIT), Danny Lewin (Akamai, died 9/11/2001) , Matthew Levine, Philip Lisiecki (Akamai), Bruce Maggs (Duke University/Akamai), Luke Matkins (LifeStreet), Sean McDermott (Akamai), Gary Miller (Carnegie Mellon University), Erik Nygren (Akamai), Andrew Parker (Netflix), Roberto de Prisco (University of Salerno), Harald Prokop (LevelUp), Hariharan Rahul (MIT), Satish Rao (U. C. Berkeley), Kyle Rose (Akamai), David Shaw (Nasuni), Alex Sherman (Google), Ramesh Sitaraman (UMASS Amherst/Akamai), Scott Smith (Pure Storage), Bin Song (Google), Daniel Stodolsky (YouTube), Ravi Sundaram (Northeastern University), Joel Wein (Google), Chen Welinder, Yoav Yerushalmi (Google)

ACM Fellows: At the ACM level, two SIGCOMM members have been selected as ACM Fellows this year: K.K. Ramakrishnan and Geoffrey M. Voelker, One other member of the community was recognized as an ACM Distinguished Member, Sharad Agarwal.

Industry Liaison Board

Three years ago, an industrial liaison board (ILB) was established with the goal of coming up with ideas and suggestions to increase industry participation at SIG-sponsored conferences. The ILB is currently chaired by Venkat Padmanabhan.  A summary of its recent activities follows:

• Industrial demos:  In the initial years, the ILB owned and drove the industrial demos process. However, more recently, including in 2018, the process has been driven by the designated industrial demos co-chairs in the conference organizing team, with the ILB playing an advisory role. The initial call for demos in 2018 received a rather tepid response, following which the ILB stepped in and had the chairs extend the deadline by a few weeks and leaned on them to be more proactive in getting the word out, including in the local region (Europe). The ILB members also used their own networks to spread the word. In the end, there was a healthy response and the conference program has ~13 industrial demos, which is as good or better than it has been in the past years. 
• Based on the experience from 2017, where the industrial demos were segregated in a separate room, the ILB also urged the industrial demos co-chairs and the general co-chairs for 2018 to have these demos be part of the regular demo program in terms of the venue and timing. Hopefully, this suggestion will be acted on at the 2018 conference.

• Additionally, based on the experience from 2018 and the past years, the ILB had the following suggestions going forward:
  • The industrial demos chairs should be proactive in seeking demo proposals. Assuming that these will come anyway is unlikely to work since the industrial demos track isn’t quite as established as the paper or regular demos track. Besides, the industrial audience we seek to attract may not have SIGCOMM on their radar.
  • Particular emphasis should be placed on reaching out to companies in the local region, who are more likely than others to be willing to travel to the conference.
  • The chairs should feel free to leverage the ILB’s network and, in any case, should keep the ILB informed of their progress, so that lacunae, if any, can be detected and plugged sooner rather than later.
  • At the conference itself, the industrial demos should be co-located with the regular demos to maximize opportunities for engagement between the SIGCOMM attendees and the industry folks.
• Student survey done after SIGCOMM 2017:  After SIGCOMM 2017, the ILB surveyed the student attendees on behalf of the EC and the ILB, and received 79 responses.  Major questions in the survey were as follows:
  • “What did you hope to achieve at SIGCOMM?”: most people answered “learn about cutting edge networking research” and “network with attendees.”   42% wanted to meet with potential recruiters, and 80% wanted to get to know senior researchers.
  • “Which event did you find most useful?”: 38% said “hallway conversations”, 30% said “attending talks”, 11% said “student dinner”, 6% said “mentoring moments”, 5% said “banquet”, and the rest were below 5%.
  • “Least useful?”: 42% answered “none”.  “Student dinner” and “reception” each got 17%, banquet got 9%, “attending talks” got 5%.

92% of the respondents did attend the student dinner.  85% rated it 3 or better on a 0..5 scale, with 5 as the best rating.  Reasons for at least 50% of attendees were “hang out with students”, “free food”, “meet industry attendees”, and “meet organizers and senior researchers” — in particular, 55% were specifically interested in meeting industry people.   However, only 38% actually made a connection with someone looking for an intern or employee. (But 55% had interesting technical conversations with industry people, which might mean that a conversation doesn’t count as a a “connection.”)  43% said they wanted more industry people at the student dinner, 37% “about the same”, and 10% didn’t realized industry people were invited. 56% liked receiving a list in advance of industry people and their interests, 36% did not, and a few didn’t realize that the list existed.

51% of students stated that they want a dinner format that encourages more mingling, 17% want a sit-down dinner, and the rest don’t care.  52% would like us to keep the student dinner while 25% would be happier with two banquets.

Support for the community and projects aimed at broadening participation and/or diversity

On the heel of his appointment our new director for diversity and inclusion, Marinho Barcellos, undertook a review of the SIG’s standing using data from membership reports, conference attendance, program chairs, and participation in program committees.  The results confirmed our expectation that women are generally under-represented, even if our situation is comparatively better than in other SIGs. With respect to geo-diversity, the data confirmed that conference attendance and representation are predominantly US-based. This provides a baseline we can use to evaluate progress.

A specific initiative currently underway is the creation of a network of researchers in Latin America, a currently under-represented region but with substantial potential to grow. We hope to inform research groups there about the benefits of getting involved with the SIG, and also understand their reality to become more efficient in how we support them to gradually increase engagement in conference participation and organization.

The SIG has also been using its strong financial position to help initiate and provide financial support to a number of activities.  In particular, it awarded a total of about $280k in various grants to support either travel to conferences (mostly though not solely limited to students) or a variety of events, e.g., summer schools or regional conferences, of interest to the SIGCOMM community.

In particular, our geo-diversity program continues to provide support of up to $10k to events such as AINTEC, APNET, and COMSNET.  In addition we provide up to $40K in geo-diversity grants that support faculty and students from under-represented regions in attending our sponsored conferences.  Additional support is also available for attendance of TPC meetings.

We continue to support the N2Women organizations with travel grants to the SIGCOMM Conference ($20k) as well as other smaller conferences, and supported the organization of an N2Women dinner at SIGCOMM 2017 (~$12k).  We are currently in the process of discussing broadening of our support with the N2Women leadership to help organize a more permanent participation by N2Women across our conferences.

We provide $15k in support for the CRA-W Grad Cohort Workshop, which is used to allow SIGCOMM students to attend the Workshop. We also support attendance of the workshop by a couple of senior SIG members to represent the SIG at the workshop.

We have been continuing the practice of preview talks at the SIGCOMM conference.  The talks seek to provide background on the papers to be presented during the technical sessions. Preview talks have been popular among students who feel they allow them to get more out of the conference by helping them come up to speed on ‘hot’ topic areas.

At SIGCOMM 2017 we shifted our approach to childcare support from trying to organize childcare support on site to offering childcare grants.  This only had limited success in 2017, possibly due to limited advertising of this option, but we are trying to expand it for SIGCOMM 2018 and have also provided similar support (up to $5k) at SOSR’18.

We have continued the practice of waiving the SIGCOMM contingency share for our fully sponsored conferences to give the organizers more flexibility and allow them to keep registration fees as low as possible.  

Issues facing the SIG

Besides the fact that data communications has matured as a research area, the SIG (like other SIGs) continues to face a challenge when it comes to increasing its diversity.  We are focused on addressing this issue, but an incident earlier this year that related to a previous situation involving possible sexual harassment drove the point home that there is still much room for progress.  The new ACM policy against discrimination and harassment that strengthened the SIG’s own policy and is now prominently displayed in all our conferences represents a step in the right direction, but it is only one step and much work remains to be done in this area.