Skip to main content

Minutes IETF110: bess

Meeting Minutes BGP Enabled ServiceS (bess) WG
Title Minutes IETF110: bess
State Active
Other versions plain text
Last updated 2021-03-11

Recording :

Working Group Status                                      Chairs            15

Martin :
       - For draft-ietf-bess-evpn-inter-subnet-forwarding  the draft will loose
       ballots because of iesg change and it will have to go through another
Jorge :
       - draft-ietf-bess-evpn-proxy-arp-nd all DISCUSS cleared for proxy arp
       and other (ND) all DISCUSS being resolved. - need some more time to look
       at comments and reply to emails.

          -jorge - ready for WGLC

         -Sussan : first week of April expect reply .

         - it defines two different solution, one solution is complex, may be
         split it in two document.

         - draft expired after WGLC failure.
         - Author have not responded to comments.
         - If sami can not edit it, we may need new editor to complete the work

draft-mishra-bess-ipv4nlri-ipv6nh-use-cases-08            Gyan Mishra       05
 - feedback seeking from working-group.
 - WG adoption is expected.
 Stephane - is it best practice document
 Gyan  - changed it to best practice instead of standard.
 Eduard V : deliver next hop for IPv4 over IPv6 infrastructure is mandatory for
 the progress to IPv6-only

draft-ietf-bess-evpn-bfd                                  Donald Eastlake   05
min Jorge : last IETF there were some comment provided, i did not see its being
addressed email sent Donald : comment would be addressed.

draft-brissette-bess-evpn-l2gw-proto-06                   Luc Andre Burdet  05`

- draft-ietf-bess-evpn-mh-pa is mature draft and has multiple vendor support
- draft ready for WGLC
- draft-brissette-bess-evpn-l2gw-proto-06 , based on feedback there has been
incremental updates. - it is ready for adoption.

draft-krattiger-evpn-modes-interop                        Lukas Krattiger   10
 - document ready for adoption
Wen lin : syymetric and asymetric IRB support . what is forwarding behavior ?
whether it would be L2 or L3 ? Lukas : yes, it would be L2 forwarding Wen line
: it would be good to write in draft about forwarding model. Wen lin: Have
comment on Vlan aware and Vlan based service. Will have offline discussion
about some of the MUST statement in draft.

draft-dskc-bess-bgp-car-problem-statement-01              Dhananjaya Rao    10
min draft-dskc-bess-bgp-car-00.txt                            Dhananjaya Rao   
10 min

Jorge: NLRI including key len in route. RR behavior is new the way we do today.
are we changing behavior 7606 or is it adding new security risk. DJ: based on
experience we have come up with this solution. security implication would be
discussed. Jorge: encapsulation part of NLRI, no label DJ : we can discuss
offline Swadesh : function & locator can be carried as part of NLRI Jorge:
Swadesh : no label in NLRI . Srihari : NLRI whether is right direction. DJ :
non key TLV, we can discuss further Shraddha : MPLS you need prefix to label ,
why prefix to SID mapping is needed. it would be useful to propose next draft
Swadesh : composed of locator and function Haibo Wang : Jeffery Haas: The
packing conversations will become interesting with any of the PDU formats.
Having some per-NLRI+key state that is optional leads to an interesting
discussion about what bits get refactored from path attributes to
nlri-optional. AIGP for example would currently cause packing to break in some
situations as well.

Ketan : @Srihari about the NLRI keys - the keys are just E and C, the non-key
is the encapsulation data like label & SID function

Kaliraj : I just wanted to note that the end-to-end intent will not be
preserved, if the mechanisms of the CAR validation/resolution are followed as
described this presentation (in slide 21, perhaps).

Ketan : @Shraddha the same remote/egress IPv6 PE loopback may be reachable via
different intents and hence may use different End.* SIDs for the encapsulation.

Srihari : @Ketan - I understand, the non-key were attributes and now being
proposed to be part of NLRI. My point is, as wg (both IDR and bess) needs to
assess if this is the direction that we want to go. My opinion, this is
reinventing wheel.

Ketan: We can have discussion .

Kaliraj : "Transport-layer" CAR routes must not use "default/best-effort"
tunnels. In order to preserve end-to-end intent

Ketan : @Kaliraj, IIRC, the default/best-effort was used as a fallback and that
too based on the defined policy

Jeffery Haas : The idea of updated NLRI is interesting. Exploring the
consequences of it will be good engineering discussion.

Ketan : @ Jeff - I don't believe the proposal is for "unconstrained" stuff
getting into the NLRI. There needs to be a line drawn here.

Jeffery: Agreed, ketan. That's where I think the discussion needs to go

kaliraj : @ketan, when u allow transport-layer routes to fallback to default,
end-end intent cannot be preserved. you may figure it out when you implement,
test :) Ketan : @ Kaliraj, consider operator is most concerned with delay in
the core (i.e. long haul) and says deploys FlexAlgo with delay metric there.
There is no need to do that in the access layer within the metro. So use FA in
the core and best effort in the access. :-)

draft-hegde-spring-mpls-seamless-sr                       Shraddha Hegde    10
 - across vendor discussion in process if it needs to be merged.

draft-kaliraj-idr-bgp-classful-transport-planes-07        Kaliraj           10
min Swadesh : i see you carrying colored route target. do you supprt import
sementic. can route carry multiple RT ? Kaliraj: Route will map to one
transport class and it can carry multiple RT

draft-wang-bess-l3-accessible-evpn                        Wei Wang          10

draft-head-rift-auto-evpn-00                              Tony Przygienda   10
 Stephane : is there any relationship with BGP autfo conf in IDR.
 Tony : I am not aware of it.