Minutes IETF117: bier
minutes-117-bier-00
Meeting Minutes | Bit Indexed Explicit Replication (bier) WG | |
---|---|---|
Date and time | 2023-07-28 19:00 | |
Title | Minutes IETF117: bier | |
State | Active | |
Other versions | plain text | |
Last updated | 2023-09-12 |
minutes-117-bier-00
Meeting: IETF117, Friday 28 July 2023 Location: Hilton Union Sq, Plaza B, 12:00-13:30 Chairs: Shep <gjshep@gmail.com>, Tony <tonysietf@gmail.com> Secretary: Sandy <zhang.zheng@zte.com.cn> Minutes: Sandy <zhang.zheng@zte.com.cn>, Hooman <hooman.bidgoli@nokia.com> 0 WG status no comments. 1 BIER MLD Tony Przygienda: there are six authors in the draft. Stig Venaas: we may explain or move some authors to the contributors. Tony Przygienda: not for holding the process of this draft. but in case there are multiple DF candidates, DF may be selected for this. Stig Venaas: it may be solved by configuration to make sure you only enable one DF, or run PIM to use PIM DF election. Ice has a draft to discuss the election may years ago. We think to move this draft to publish with restriction and then work on the election of non-PIM DF election. Tony Przygienda: we can publish this draft, push the problem and let people know that there is any other solution except PIM DR election. Chairs: start the polling for “should we wait for the general election draft?” The poll result of raise_hand: do_not_raise_hand is 8:1. Greg Shepherd: since we have a draft talking about the general election function years ago, when you’re going to work with Ice or take it over. Stig Venaas: will talk with Ice. 2 BIER OAM • BIER Ping Tony Przygienda: please check the number of author list according to the author number limitation. Greg Mirsky: already done. Tony Przygienda: BIER ping is the very important OAM draft. more people review is better. Greg Mirsky: implementations sharing welcomed. Hooman Bidgoli: Nokia has implementation. but needs to check the TLV changes. Greg Mirsky: core message is still the same. may send you questionnaires for the implementation information collection. • BIER BFD no comments. • BIER performance measurement Tony Przygienda: about the individual draft you mentioned in IPPM WG, did you call for WG adoption for it? Greg Mirsky: during our presentation in IPPM WG, IPPM WG chairs raised some technical issues but they are not object to the adoption call if we want before IETF118 meeting. Tony Przygienda: what’s the relationship of the two drafts? if BIER BFD applies automatically? Greg Mirsky: the individual draft in IPPM WG defines the function and is underlay agnostic. it can work with IP, MPLS or BIER. Tony Przygienda: applicability deployment consideration or something else may be considered to see if there is any special to BIER. Greg Mirsky: Yes. so we solicit BIER WG to provoke the thinking. Tony Przygienda: so the bitmap mapping in BIER has been defined, right? Greg Mirsky: Yes. definitely • BIER OAM requirements Greg Mirsky: would like to know how the draft going. Tony Przygienda: preserve it. Greg Mirsky: wants WGLC for it. Alvaro raised the question of consistency between performance measurement and OAM requirements. it may apply to all the BIER OAM documents. Tony Przygienda: at least BIER ping should be considered. 3 BIERin6 The co-author request WGLC. Chairs start the polling for it, 10 positive and 0 negative. 4 BIER-TE • IS-IS/ OSPF/ OSPFv3 extensions for BIER-TE Jeffrey Zhang: suggest combine all the BP and BIFT-ID signaling of the three IGP extensions (IS-IS, OSPF and OSPFv3) into one draft. Huaimo Chen: we follow the older style of BIER IGP extensions. Jeffrey Zhang: I think it’s in early BIER developing process. Now BIER is already in premature and the BP and BIFT-ID advertisements are just incremental additions to the existing protocols. And IS-IS, OSPF and OSPFv3 has been covered by the same LSR working group. So I think it makes sense to combine all the BIER-TE IGP underlay signaling such as BPs and BIFT-ID into one draft. Huaimo Chen: Maybe we can merge them into two drafts, one for IS-IS, one for OSPF including OSPFv2 and OSPFv3. Greg Shepherd: I see there is few information in each of the BIER-TE extensions drafts. It may not always make sense if we’re defining the same extension for different protocols in different drafts. A clear document is better. • BIER-TE LAN Sandy Zhang: IMO this situation may not existed because this can be avoided by configuration or controller. The controller knows all the network details and can find the way to avoid wrong encoding. Huaimo Chen: for the example I listed there is duplicated packet existed even if we use the controller. the existed BIER-TE RFC (9262) indicates that all the nodes only know the local and adjacent BP in the network and don’t know the remote BP. Sandy Zhang: if you read RFC9262 carefully you will know if you want to build a multicast tree in this scenario you must make some necessary pieces. The controller is intelligent and will know the duplication and try to avoid it. Andrew Alston (AD): good discussion and please continue in the mailing list. 5 BIER extension header Greg Mirsky: we may consider the discussion of MNA. Andrew Alston (AD): if you want to make alignment with MNA you should be careful. Tony Przygienda: it’s interesting topic and we should pay attention to MNA because BIER is 2.5 layer. Greg Mirsky: as Andrew said, MNA has adopted framework, requirements, use cases and ISD encapsulation documents. But we are still looking for if there is any use case that cannot be solved by using ISD. Now all the use cases listed in use cases document can be solved by using ISD solution. So for BIER use case if it’s real that it cannot be solved by ISD. it’s good to join the MPLS discussion and give the feedback from BIER perspective. Jeffrey Zhang: we will consider if it’s urgent if PSD will not make progress in MPLS. The reason we want to go with the PSD solution is that the ISD bit set is not aligning with IPv6 encoding. The extension is not only for MPLS, but also for IPv6 and Ethernet. So mostly we are considering the applicability of IPv6. Greg Mirsky: it’s valid concerns. but for BIER use case we need to find that if there is anything cannot be solved by ISD. Sandy Zhang: Whatever PSD will be adopted by MPLS or not, we should consider the extension header from BIER perspective. Jeffrey Zhang: originally we try to do it for next header concept, we may no longer say that we are trying to align with MPLS and take it more for IPv6 next header. Tony Przygienda: we should look at the requirements document to see if it applies to BIER. Tony Przygienda: for the details of extension header, it’s confused that there may have similar encoding and different semantics. you may have a common block for silicon stuff but the op code at the end is different semantics. Jeffrey Zhang: yes, it may be more complicated for the assignments. we may talk more about it. Tony Przygienda: encourage somebody to take the requirements document and see whether we have more requirements and which of the requirements apply to BIER. Jeffery Zhang: now a concrete use case for BIER is IOAM. Fragmentation and security are things we could do but may not be mandatory. Tony Przygienda: for freedom in design so it doesn’t have to be done now but if it’s architecturally important, that’s already a requirement as far as I see. Greg Mirsky: Since MPLS is still discussing the IOAM topic, it’s good for BIER to provide the points from BIER perspective to see the difference between ISD-based solution and PSD-based solution. 6 Discussion of BIER interop events Jeffrey Zhang: there are some off-line discussion about BIER interop events as part of hackathon in Prague. I will send the email to public BIER mailing list. Greg Shepherd: who do we have confirmed so far? Jeffrey Zhang: Nokia, Juniper, Huawei and ZTE have the indication but that’s just unofficial. I forgot to send email to Cisco. Greg Shepherd: don’t forget B4 implementation. Hooman Bidgoli: we may need to make sure how do we make the devices connection., because some implementations using simulator and some implementations like Nokia using the actual hardware. Jeffrey Zhang: even in simulator there may need hardware though the connection by simulator may be easier. They all are the details we need to figure out. we should have the discussion ahead. Greg Shepherd (without chair’s hat): let me know if you need an independent third party to sit in the middle. ----end of the session----