Workshop on Community Oriented Network Measurement Infrastructure

The workshop was held on March 30, 2005 in Boston, Massachusetts from 9am-5pm, one day before the Passive and Active Measurement (PAM 2005) Workshop. The workshop was by invitation-only. NSF Sponsorship pending.

Organizers: Mark Crovella <crovella@cs.bu.edu>, Joerg Micheel <joerg@nlanr.net> and k claffy <kc@caida.org>

Workshop Background

Internet research critically depends on measurement, but effective Internet measurement raises some big issues. There is increasing awareness that the size and scope of the Internet calls for large scale distributed network measurement. However to be viable, large scale measurement must overcome a number of challenges. Passive measurement raises significant privacy issues. And active measurement raises serious concerns about network impact.

This workshop will bring together key members of the Internet measurement community in an attempt to explore a community-oriented approach to these problems. Our inspiration comes from the astronomy and high-energy physics communities which have self-organized to build, operate, and allocate the use of large, unique, and expensive measurement platforms. We hope to explore whether the Internet measurement community is ready for this challenge.

Our vision is that a community oriented approach may be sufficient to address issues in large scale measurement such as privacy and network impact. With respect to privacy, we hope to explore the issues surrounding a facility that would "accept experiments" to be run over full-packet capture systems. Each experiment would be examined in advance through community mechanisms (eg, a review panel) to ensure that the results, if released, would not violate privacy concerns. We seek to explore whether a fundamentally new model such as this for passive measurement could enable a dramatically more powerful set of measurement experiments.

With respect to network impact, we hope to explore community mechanisms for overseeing large-scale distributed active measurement. For example, experiments involving many thousands of active probing hosts are now being proposed. In the absence of community oversight these could easily go awry with undesirable results for the Internet as a whole. Since there are no barriers to performing such experiments, we hope to find ways to facilitate them in a safe way, both via community review in advance and ongoing monitoring during execution.

This workshop will be held March 30 in Boston, the day before the 2005 Passive and Active Measurement Workshop. It is by invitation only to a select group of around 20 researchers. The output of the worshop will be a workshop summary with suggested next steps for investigation.

Agenda

8:30-9:30 Mark Crovella: Organization: welcome, and goals for the day
- * overall goals: addressing challenges in both passive and active measurement
- * projected approach: community oriented infrastructure (preview, not a full discussion yet)
9:30 - 10:30 Round table: 3 minutes per participant
- who you are, where you work and have worked, background
- your present research portfolio and interests
- single-line expectations toward measurement infrastructure requirements
10:30-12:00 Passive measurement infrastructures, activities, and needs
- 20 mins: Joerg Micheel: Passive Infrastructure: Opportunities and Challenges
- 20 mins: Colleen: current problems in trace collection and analysis
- 20 mins: Gianluca: Intel CoMo project
- 30 mins: Rob Beverly: The Spoofer Project
  open discussion
12:15 - 1:15 Lunch (brought in)
1:15 - 2:15 Active measurement infrastructures
- 30 mins: David/kc: caida topology measurement review and future directions
- 15 mins: Tony McGregor: NLANR AMP: Lessons Learned
- 15 mins: George Riley: NETI@Home; open discussion
2:15 - 2:45 Rick Summerhill: The Abilene Observatory and Measurement Opportunities
2:45 - 3:15 Break
3:15 - 4:30 Round table discussion (please see mark if you want a 10-minute slot)
4:30 - 5:00 Outline of recommendations, signups for report writing