IETF 82 6RENUM Working Group 18th November 2011 11:20 - 11:25 Note Well, Jabber relay - none/Wes relaying comments if necessary note taker(s) - Gunter Van de Velde 11:25 - 11:30 Agenda bashing, WG status update 11:30 - 11:35 Homenet update (Tim Chown) Tim: Homenet has not taken considerations on renumbering. Wes: Update on DHC - spoke at DHC about the fact that some of the things that they are working on (multihoming cases) may be applicable as renumbering cases - since we may be sending gaps their direction, encouraged them to think about renumbering in that context. also recommend that renum members participate in DHC 11:35 - 11:40 SDN BoF update (Wes George) Wes: initial discussion on automation of provisioning to identify various API?s. Interesting for Renum is to understand where new tools are needed. SDN is not moving fwd to become a WG just yet. 11:40 - 12:10 IPv6 Enterprise Network Renumbering Scenarios and Guidelines draft-jiang-6renum-enterprise-01 Presenter: Brian Carpenter Presentation of slides Changes from -00 Static address is defined in separate document ULA is out of scope Discusses ND and DHCPv6 co-existing Secure dynDNS is recommended Slide-3 Structure draft Slide-4 Considerations and BCP Prefix delegation Usage of FQDN ? Service location protocol is not widely used and multicast DNS with SRV records. Question is how to convince people to Tim: why would people use FQDN Speaker (???): controlled access lists for services/server and to restrict access to certain domains. If one uses addresses, make sure that one can find them due to maybe wrong caching. Speaker2 (???): if QDN option is existing, then can it be used for renumbering? Should a host spec be defined. Answer from Brian: it seems an implementation option and not a protocol option Slide-5 Avoid manual configured addresses ULA?s should be used for only local communications Speaker (???): why is this a restriction for address configuration? Wes: important to clarify the situation for the enterprise domain and force to choose one or the other. Tim: differentiation needs to be made to understand if it's on same subnet or not Brian: Updating DNS should be procedural, and DNS on demand is probably not that such good idea. Speaker (???): what if I have 1000?s of hosts with strong crypto, that could be a DoS attack Wes: SDNS may in some case not be able to deal with it. Tim: is SeND used? Question is raised if people use SeND (answer: none) or authenticated DHCP (answer none). RA-Guard should be mentioned in the draft. Jim Harper: not having this is not a big hole. Brian: RFC4092 (??) like procedures for renumbering. Not all people have long lived sessions. Just a few and they need to be informed on any change. That will be management work and not technical work. Speaker (???): apps that don?t do re-issue when address changes need to be taken into consideration During renumbering: with or without a flag day... If DHCP lease is longer than when the address Eric Vyncke: some ISP?s change IP?s quite frequently. Q from Brian: Is that a homenet issue or Enterprise issue: Answer from Eric: I don?t know. Question from Brian to Eric to send a note to email list for checking if this is an issue in Enterprise or homeNET Brian: If DNS servers themselves get renumbered then that needs to be propagated to hosts as well. Tim: would that be a usage case for ULA Fred Baker: Would it be useful to have another prefix as usage case, and does that ?have? to be a ULA? Is session survivability important Wes: there is major difference between when session starts and ends, and when session starts and gets expired. Tim: Did the draft look into the older renumbering RFC Fred: The RFC needs to be updated before it can be used Brian: should this go to BCP or an Informational document John Harper: depends on how well we know the answers to the questions. Jarri Arko: demonstrated concerns in this section Benedict Stockebrand and Fred proposed to review and provide feedback on the list. 12:10 - 12:30 Problem Statement for Renumbering IPv6 Hosts with Static Addresses (Brian Carpenter) draft-carpenter-6renum-static-problem-00 New draft describing an observation. If there are static addresses, then the subnet needs to be static also. Eric Vyncke: Printers use often multicast DNS instead of static addresses. Tim: stateful DHCPv6 is a solution as well Brian: Virtual machines tend to have static addresses, and there seems need for a managed stateful procedure. Wes: It would be interesting to get some information on other applications Fred: It needs to be handled in a single place, but is this the correct place? Tim: what about the HOMENET case where the there is delegation and Nat? Brian: What about asset user management and user tracking. For this many organizations use stateful DHCPv6 Brian: there are still folks that have licenses attached to IP addresses. Wes: recommendation is to not do it Brian: this is reality, we must describe it Bob Hinden: In many cases a loopback would do the trick Brian: or placing a NAT ? Brian: Network Elements is often done through config and script files. Fred Baker: Can silliness be automated? Benedict: there are reasons for static addresses. Often a shortcut, silliness or both. Brian: if addresses are static, then automation will be rather hard. And there is also still issue for licensing on IP addresses. Tim: at HomeNET the goal is that in the home the addresses remain rather static. Q: anything missed, any useful recommendations, does 6Renum want to work on these aspects. Wes: we need to see and make distinction between a random device and a server Benedict: Good monitoring is very important to detect problems Brian: is there a need for special monitoring outside normal operation Benedict: Wes: asking authors to create a new revision and publish it to the list 12:30 - 13:00 IPv6 Site Renumbering Gap Analysis draft-liu-6renum-gap-analysis-02 Speaker: Bing Liu Aim: Promote Automate renumbering Promote ?make before break? ? what about session survivable Promote unplanned immediate Tim: Flash renumbering will most likely be required to be taken into account Some sections are moved to annex. Wes/Tim: add two sections for protocol gaps - unsolvable and out-of-scope Wes: We are not creating solutions and describe protocol changes Speaker (???) ? we need some help with Multicast Wes: challenges what should be in or removed, and that should be the group defining this. Bing: exclude SLAAC/DHCP gap and standard gaps Router restart issue ? unclear what the impact is Parameterized router configuration is a GAP Lee Howard ? should we speak about an issue that may or not may not exist. It is an implementation issue Lee Howard ? in HomeNET there are some potential issues identified, so we should take that into account also Fred: use DNS and make the lifetime shorter than the restart time. Wes: Netconf is mentioned by Bing as an incomplete solution in some cases. Wes: Who will review the document? Volunteers: Gunter Van de Velde, Chris (and I could not understand) Brian: There are just few editors, maybe there is need for more editors Wes: concerns that there is not enough participation Liu: mentioned that this session is maybe too late in the week Lee Howard: Is there a document missing according to the charter? Tim: a new document on static addresses showed up as that seemed important to focus upon 13:00 - 13:30 Open Discussion Suggestion (???): What is the impact on the routing and should a draft be written on that aspect? Brian Carpenter: This seems as the wrong WG for this type of work, however its important Wes: if this would be done, then requirements will be important to describe 13:30 Close