James, ........................................Determination of how this router is designated is outside the scope of this document. An RA that is immediately unicast to the sender rather than delayed is known as a "fast RA".
I do not believe it should be outside the scope of this draft. I think it imperative that this be defined clearly and a method proposed before FAST RA can be bought into by IPv6 implementors. I also believe at this time the IESG should require a fairly good set of reports from multiple implementations of this and other proposals for wireless efforts that want to change DAD parameters and extend attributes to ND and Stateless Addrconf. Right now the only work within this space that has received the test of fire at UNH, ETSI, or TAHI as examples is MIPv6. These tests beat hard on IPv6 implementation enhancements as was just done for MIPv6 D-18 in Sophia Antipolis at ETSI. I would suggest to the IESG doing any changes to ND or Addrconf without serious hard core testing from this work is dangerous. If you, I, and others believe this is important then the code will get done for this and Optimistic DAD and show up at the tests. Otherwise this is purely a research and at best AD exercise. IPv6 is now being deployed we cannot mess with DAD or Addr Conf in casual ways any more than we do with IP (for v4), TCP, or RTT algrorithm for TCP as examples. I also believe all these efforts should begin to think like MIPv6 has and that is they are separate documents and I strongly am against any enhancments to ND or Addrconf from Fast RA, Optimistic DAD etc. They should be new RFCs. All this also presents new security problems too that does not exist in current ND and Addrconf on wired links. The base reason is that these mess with the ND architecture and wired links that are firewalled or PKI protected are more insulated than say 802.11b. That being said I would like to see all security sections for this work define what new security problems exist simply because of this behavior not just say exists now in DAD or whatever. None of these options should be permitted for wire links (status quo) at all. That gives me an idea at least how the router to do Fast RA should be defined. I would also like to see hard core router or switch "PRODUCT" developers comment on this work and other work. I know routing code but not like those that build ASICs and the like. I would like to hear them say the pain and gain they feel or see from this strategy. /jim [Have you ever seen the rain coming down on a sunny day] -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
