IETF-65 Autoconf WG Thursday 23rd March 1300-1500 Chantilly W ===================================================================== Minutes: Christopher Dearlove Emmanuel Baccelli, w. help from Philippe Jacquet Jabber: Justin Dean Chair: Thomas Heide Clausen (Apologies from Shrubhranshu Singh, who could not be present) Attendees: 78 (on blue sheets) Preliminaries (Thomas Heide Clausen) ------------------------------------- Two WG documents: architecture document (relevance wider than this WG) problem statement document Both overdue, need candidate WG documents as soon as possible. Item 1: MANET Architecture Document ------------------------------------ Dave Thaler "Multi-Subnet MANETs". Document: draft-thaler-autoconf-multisubnet-manets-00.txt Slides: In Proceedings Related: also see Dave Thaler's INT area presentation Discussion: Fred Templin: - are you planning /64 prefixes for MANET? Dave: - Yes Fred Templin: - have you considered if clients may be routers? Charles Perkins: - what are multicast/broadcast expectations, just to node itself? Ian Chakeres: - suggest using "link" rather than "subnet" as latter is overloaded term. Fred Templin: - what sort of addresses should routing protocols pass around? (Prefixes.) This is a change. Joe Macker: - this is what is being done in MANET-OSPF. Treat nodes as routers, supporting prefixes. Charles Perkins: - current MANET protocols can handle prefixes (...general agreement that this is the case...) Pedro Ruiz: - Concerned by compatibility with traditional IP. If two routers cannot see each other, can they be in the same subnet? Dave Thaler and Thomas Clausen: - out of the scope of the MANET field. - essentially the same situation as ANY network, MANET or otherwise Charles Perkins: - Is there something really wrong with host routes? Dave Thaler: - yes, see INT area presentation. Fred Templin: - what is a MANET-cast? What is the definition of the link on the wireless interface? Ian Chakeres: - MANETcast is holding term for link local multicast address. Dave Thaler: - loopback link and wireless link, latter is you and everyone who can hear your transmission over one hop. Charles Perkins: - concern: If you require non local signalling at layer 2 it will not scale. If you want to call a link local address as such. It must just be link-local unique, not site-unique. We should not try to detect layer-2 duplicate addresses. Dave Thaler: - must be careful when talking about "uniqueness" over a region, which may be different from the scope of the address Charles Perkins: - concern about detecting duplicate layer 2 addresses, not scalable. Joe Macker: - duplicate detection should be "should", not "must". Bob Hinden: - like this architecture, unnumbered interfaces are useful. Joe Macker: - agree with Bob Hinden, but can have globally routable addresses. Don't like this use of term MANETcast, not anything new. Craig: - I like things that do not change neighbor discovery Thomas Clausen: -should take follow-up discussion to mailing list. Item 2: Problem Statement Document ------------------------------------ Charles Perkins "Address Autoconfiguration for MANETs: Definition and Problem Statement" (partly on behalf of Shrubhranshu Singh) Document: draft-singh-autoconf-adp-03 Slides: (currently missing?) Discussion: Simone Ruffino "Internet gateway & MANETs Scenario presentation" Document: draft-ruffino-autoconf-conn-scenarios-00 Slides: In Proceedings Discussion: Fred Templin: - multiple gateways, one prefix considered by netlmm WG. Need interface between two. Dave Thaler: - multiple egress from a network is not a MANET specific problem. Joe Macker: - need to be careful of use of "optimal", shortest path is not necessarily optimal. Veering towards transit networks, most MANET work is stub network. (Not actually transit networks, but next step.)a Joe Macker: - About autoconfiguring routers, which is new, more than just addresses? Thomas Clausen: - results presented, are they specific to particular solution? Simone Ruffino: - represent problem any solution may incur. Dave Thaler: - most issues are independent of solution. A. Mustafa: - Are the gateways mobile too? What is the expected of mobility? S. Ruffino: - this is a MANET wg issue. Not an autoconf issue. Emmanuel Baccelli "A Common Framework for Autoconfiguration of Stand-Alone Ad Hoc Networks" (on behalf of Kenichi Mase) Document: draft-mase-autoconf-framework-01 Slides: In Proceedings Discussion: Thomas Clausen: - part of solution space, but as framework has relevance to problem statement. Joe Macker: - are you using routing plane for messages? Thomas Clausen: - there are different ways to do this, this is where draft moves into solution pace specific. For the moment, we should not look at what this drafts proposes in terms of actual solutions, rather what it does to scope the problem domain Joe Macker: - need to stick to general part, for example SMF can do this in data plane. Joe Macker: - Is the recommendation about generalized routing packet a requirement? Thomas Clausen: -the essential part of this draft is the Finite State machine proposal. Thomas Clausen (summary of last three presentations): - the working group needs to decide which parts of this work to include in Problem Statement. Item 3: Solution Space Announcements ------------------------------------ Pedro Ruiz "Extensible Manet Auto-configuration Protocol (EMAP)" Document: draft-ros-autoconf-emap-02.txt Slides: In Proceedings Discussion: Dave Thaler: - important to discover DNS, should be added to Problem Statement. - could carry DHCP options in TLVs. Philipp Hofmann "Multihop Radio Access Network (MRAN) Protocol" Document: draft-hofmann-autoconf-mran-00 Slides: In Proceedings Discussion: (No issues were brought up) Simone Ruffino "Automatic Configuration of IPv6 Addresses for MANETs with Multiple Gateways" Document: draft-ruffino-manet-autoconf-multigw-02 Slides: In Proceedings Discussion: (No issues were brought up) Final Announcement ------------------ Thomas Clausen: - maintaining http://manetautoconf.online.fr/Blog - with individual I-D announcements and such Bob Hinden: - make sure that discussion happens on the mailing list Thomas Clausen: - comments are (or should be) disabled on the blog Tom Henderson: - Does "Note Well" apply to the content of that page? Thomas Clausen: - Yes - needs to figure out formalities, if the text should be made explicit Dave Thaler: - consider adding it as "Alternative AUTOCONF www page" on page on ietf.org Thomas Clausen: - Thanks to our scribes! - Thanks for your attention.