TICTOC/NTP meeting at IETF-85 Thursday, 8 November 2012, 1300-1500 EDT (1800 - 2000 UTC) (Note: This is a joint meeting of the TICTOC and NTP WGs. All the presentations are filed under TICTOC in the proceedings.) Chaired by Karen O'Donoghue and Yaakov Stein Notes by Kevin Gross Note Well presented - Karen Agenda Presented - Yaakov --- WG status - Yaakov 3 mature workgroup drafts 6 active individual drafts 4 expired drafts Yaakov - the drafts draft-mjsraman-l2vpn-vpls-tictoc-label-hop and draft-mjsraman-l3vpn-tictoc-label-hop on securing Model-C interprovider L2 and L3 VPNs with label hopping will be presented next time, but please take a look at them. Karen: Security related drafts suspended and allowed to expire while security requirements were being defined Karen: Draft draft-marlow-tictoc-computer-clock-accuracy may be taken up by BMWG or IPPM. Dave Marlow has retired. Please contact chairs if you're interested in contributing. Chairs will then contact chairs of the appropriate WG. Doug Arnold: There's a lot of benchmarking work going on in ITU so coordination is needed. ---- ITU Q13/15 update - slides from Stefano, presented by Yaakov New work item to use transparent clock in telecom profile. Investigating how to handle layer violation caused by this. Involving IEEE 802.1. Doug Arnold: 1-step transparent clocks changing source address causes devices not to know who to respond to. PAR being initiated in IEEE to study this. Pat Thaler: IEEE 802 meets next week may respond then or in January. Changing source address does not address layer violation. Karen: This is not an issue for TICTOC to address. Just informative. ---- IEEE 1588 update - Karen Work revisiting 1588 PAR. Expectation that work on 1588v3 will start early 2013. ISPCS had several sessions on this. Will MIB work move to 1588? Doug Arnold: I will be co-chair of 1588 with John Eidson. Work is expected to be done in a manner compatible with 1588v2. Issues to be addressed: Security. Layer violation. CERN White Rabbit extensions. Need security experts from TICTOC and IETF to help. Brain Haberman: Security in what context? Authentication. Authorization. Information protection. We need a framework to be able to help. Karen: What are we talking about when we talk about security and 1588? Define stress scenarios? Current security draft has been distributed to ISPCS conference attendees. Doug: 1588v2 thinking was authentication was important. Encryption may now be important. 1588 may need a mapping to IETF security ideas. Doug: 40/100 Gb Ethernet synchronization issues. Where do you timestamp? Kevin Grioss: 802.3bf-2011 defines a time reference point for Ethernet packet transmission and reception. Pat: Skew is on the order of a byte or two. Yaakov: Current 1588 timestamp defined at leading edge of first bit. Pat: With lanes - where is the first bit? Doug: That points to some murkiness of new transports. Some applications care about picoseconds. Karen: There has not been a 1588 working group for several years. That is expected to change early next year. ---- 1588 over MPLS - Presented by Yaakov Draft proposed to be split into 4 to be sent to respective working groups. Forwarding plane draft submitted in time for IETF85. Other 3 will follow later. Greg Mirsky: Control word requirement is informational. Yaakov: Do we need to wait wait until IETF86 for last call? Karen: Other drafts are dependent on this. Eric Gray: Sharam was saying IETF86. Are we volunteering Sharam to finish early? Yaakov: He volunteered to get the next rev by January. We will check with him before changing this. --- MIB draft - Vinay Shankarkumar Brian: There isn't a MIB doctor review until prior to last call. If you need it, find someone informally to help. Karen: Andy Bierman has reviewed it. Official MIB doctor will happen once we declare last call. Vinay: I need help with some errors reported by some of the MIB tools. Karen: How long would it take to add write operations? Vinay: There are groups who can't use this MIB unless this is included. Co-authors and I want to do it. Inclusion encouraged by ISPCS community. Yaakov: Specifically what are we adding? Is this significantly expanding the scope ? Vinay: IP addresses and other configuration options. Not performance tuning. Doug: My recommendation is to keep MIB in TICTOC. There should be one 1588 MIB. Karen: This will delay completion of the MIB. Doug: two reasons to keep MIB in TICTOC as opposed to IEEE 1588: 1/ Need to get it out and 1588 group doesn't exist 2/ No MIB expertise in 1588 Kevin: We want to have one authoritative 1588 MIB Karen: Is there opposition to adding write items to MIB? No opposition in room. Karen: We also need to take a look at 802.1AS MIB if we want one 1588 MIB. Vinay: Authors will look at 802.1AS MIB ---- Security Requirements - Tal Mizrahi Only discussing changes in most recent draft Draft sent to 1588 community via the ISPCS attendees list. Doug: 1588 will be started in a few months and an email list will be formed. Doug: No on-path support in telecom profile makes for easier security - don't have to share secrets with routers. Karen: Security definitions are good. Need more feedback from community as to which are requirements. Please forward to friends and family for review. Yaakov: There are apparently trade-offs between accuracy and security. This deserves mention in the draft. Karen: We need to wait a little bit before WGLC. Will discuss timing on mailing list. Volunteers to read: Doug Arnold, Kevin Gross, Danny Meyer ---- Enterprise profile - Doug Arnold Name has changed to "Enterprise profile" since "hybrid profile" was already taken by an ITU project. Kevin: Are IPv4 and IPv6 separate clock domains? Doug: Yes. Brian: We're going to have IPv4 and IPv6 on the same network. Kevin: This is good reason to do it in IETF - IPv6 expertise Doug: IPv6 has a different multicast model Yaakov: Is hit-less alternate master recovery required? Doug: This is not a requirement in 1588 so not proposed to be a requirement here. Kevin: Should multicast management messages be allowed at all? Doug: Enterprise applications may prefer to use SNMP. Tal: Are we talking about IP source addresses? NATs may also translate addresses and might cause a problem for address in TLV for 1-step transparent boundary clocks. Brian: Do you actually want to carry this in PTP data payload or should it be in a IP payload? Doug: There are already problems in 1588 use of IPv6. This could potentially patch 1588v2 until 1588v3 arrives. Karen: These are engineering decisions we can discuss once we have a draft. Doug: Is this the right place to do this work? Karen: Do we have volunteers? Doug: Yes and we have 2 implementations. Kevin: Is it a concern that there are already too many profiles and this would be one more? Doug: There are many profiles but only two are widely used: Telecom and power. Kevin: We should establish strong liaisons with other organizations developing profiles. ---- draft-shpiner-milti-path-synchronization - Tal Mizrahi Danny Meyer: Can't paths constantly vary? Tal: Yes. There will be transitions when paths changes. Doug: Will this work when you can control the path? Tal: Draft discusses IP only where this is not possible. We can define over other transport types. Karen: Discuss working group adoption on mailing list. ---- Autokey - Dieter Siebold Karen: We need more discussion on this. We had put together a small design team and that needs to be reassembled. Anyone interested in joining contact me. This is a big change so we need to think about how to proceed. ---- NTP extensions - Tal Mizrahi Yaakov - we are almost out of time. Tal can you present in 1 minute ? Tal - yes ! Clarify whether a MAC is required when you use and extension field. Believes an erratum is needed. Yaakov: You can submit an erratum on you own. There are some people who submit them all the time ... Tal: I'd like the workgroup to review the draft first Deiter: I support submitting an errata. Yaakov: +1 Meeting ended 1 minute after official time.