Salt Lake City Joint Techs (JT15) Summary

By Ben Chinowsky

The 15th Joint Techs Workshop (JT15), organized by ESCC and Internet2, took place February 13-16 at the University of Utah. Some major themes were:

• International scientific collaboration. Large international scientific projects, particularly in the areas of astronomy and high-energy physics, are emerging as major drivers of growth in international R&E networking. JT15 attendees got an overview of one of the most ambitious of such projects -- ITER (PDF), which will likely become the first fusion reactor to produce more energy than it consumes -- and the network capabilities that will be needed to support it. Cees de Laat noted several others in his update on the Global Lambda Integrated Facility (GLIF), which is principally driven by the data-transport needs of Big Science. JT15 included updates on networking in Asia and the Pacific, Latin America, and Europe, as well as the NSF International Research Network Connections (IRNC) program which is funding network development in these and other parts of the world.
   
• IPv6 deployment. Growing international collaboration also means growing need for IPv6. For ITER in particular, both security issues and the participation of Asia, which is well ahead of the US in IPv6 deployment, make IPv6 connectivity crucial. National and international research collaborations are also driving IPv6 deployment in ESnet; Kevin Oberman presented lessons drawn from their experience. There was also a comprehensive update on IPv6 in DREN; DREN is the lead network in DoD's drive to move all military networks to IPv6 by 2008. IPv6 multicast is receiving growing attention. Bill Wing noted recent speculation that IPv6 multicast over the Access Grid may prove to be the killer app for IPv6, and a multicast BoF discussed ways to get IPv6 multicast deployment moving. A joint meeting of ARIN and the North American IPv6 Task Force is coming up April 17-21 -- see http://www.arin.net/ARIN-XV/.
   
• Moving beyond 10 Gigabit Ethernet. With 10GigE increasingly well-established in metropolitan-area and local-area networks, precedent would seem to dictate that it's time to get started on 100GigE. This time around, however, a variety of technical and economic issues suggest otherwise. While providing a wealth of detail on these issues, a panel discussion showed that no consensus has yet emerged on the central question of what the next standard should be. At the same time, there was agreement that much network growth over the next few years will be in metropolitan-area rather than long-haul networking, further raising the stakes involved in getting this next step right.
   
• New technologies for wireless networking. Al Javed offered a comprehensive survey of these technologies (PDF). There is an embarrassment of riches here; Javed noted that ensuring interoperability among an assortment of deployed wireless technologies requires better "etiquette" built into the technologies for better sharing of shared bands. On the other hand, more stringent etiquette means less user-friendliness; an ideal solution would involve machine learning built into the radios themselves, so they know when not to transmit, but that's likely to be a few years away. In a Vendor Soup session, Madan Jagernauth discussed how Nortel is deploying "wireless meshes" across large areas, using access points in weatherproof cans that can be placed anywhere there's power, e.g. on utility poles or the sides of buildings. There were also presentations on various aspects of wireless from Philippe Hanset (PDF), Matthew Gast, and Terry Simons, and JT15 was immediately preceded by a successful 802.1X and Wireless Network Security hands-on workshop.
   
• New security threats. In addition to the major security challenges of wireless, JT15 attendees were brought up to date on the growing problem of "botnets" (PDF), networks of remotely commandeered computers. These networks can be very large and very resilient, with any compromised computer able to take over the role of network controller in the event the exising one is discovered and shut down. Botnets have become big business; 70% of spam now originates from them. One key capability needed to address emerging challenges to network security is line-speed packet processing using languages such as SNORT and BRO. There was a well-attended Vendor Soup Metanetworks demo of hardware that will do this at 10 Gbps. It seems clear that security will remain as much art as science for the foreseeable future; in Tim Toole's recap of security at SC2004, the first item listed under lessons learned was "Intrusions were caught by good judgment."

Presentations from JT15 are available at http://events.internet2.edu/2005/JointTechs/SaltLake/agenda.cfm?event=228. Writeups from the sessions sponsored by the Internet2 End-to-End Performance Initiative are at http://e2epi.internet2.edu/JT05/JT-winter05-summaries.html.