Skip to main content

Minutes IETF105: bier
minutes-105-bier-01

Meeting Minutes Bit Indexed Explicit Replication (bier) WG
Date and time 2019-07-24 17:30
Title Minutes IETF105: bier
State Active
Other versions plain text
Last updated 2019-08-06

minutes-105-bier-01
Meeting notes:

BIER
Wed Jul 24
1:30 PM

WGLC

Drafts waiting for last call. Need more activity on the mailing list.

Hooman Bidgoli
Nokia
Last call on ietf-pim-signaling

Stig
Cisco
Should we make addition to mld-signaling
Will be presented today

1-draft-ietf-bier-pmmm-oam, Greg Mirsky

Tony
Have all markings been disclosed
Greg Mirsky
yes

Alvaro
This draft does not talk about requirement draft
How many requirements are met?
Looks like only 2 are currently met.
Need to explain why others are not met.
Will the requirement draft be published?
Or is it informational?

Greg Shepherd
Not one tool will address all requirements
More of a semantic discussion.
Process was in order; publication may not have been
Draft can be informative
Tools can be published
Room agrees

Toerless
Don’t agree.

Akbar
Cannot make requirements normative

Alvaro
Tools cannot be published without the requirements
Alternative can be publish them together

BIER header has 2 bits that can be used for OAM; does not explain how.
This draft is taking ownership of the bits and not leaving room for others.
Can the bits be used by any mechanism or are we going to assign ownership to a
specific mechanism and is this the mechanism Take it to the list

WGLC
Take it to list with issues and do LC after.

2-draft-zzhang-bier-tether-02, Jeffrey Zhang

Greg Shepherd
Cannot follow the example for illustrating SPF rooted at BFRx
Why would there be a loop if there is no bit set for BFR1?

Jeffrey
Can happen if BFRx decides path to go to BFR3 is through BFR1

Greg Shepherd
Who thinks it is ready for adoption?
People who read it agree

Sam
Non BFR capable router X; will it hold state for BFERs
Can we call BFRx a proxy?

Jeffrey
Proxy is used in another draft to to mean something else
Helper does not keep any additional information other than saying it is the
helper for X Other BFRs only need to remember the helper for a non bier-capable
router.

Hooman
Most of the examples show only one router that is not BIER capable.
How realistic is this in a provider network?
If the one router becomes a cluster like a ring.
How will we bypass it?
Where will this be handy?

Jeffrey
Usually there are many BIER incapable routers.
In theory, we could pick one for each or a few to satisfy all of them
This will be handy in any network where we want to do BIER

Daniel Awduche
Tethering can be used anywhere to enable BIER capable network

Hooman
Prefer approach that avoids non-bier capable router

Jeffrey
If you want to get around, you still need to ensure connected graph to reach
all BFERs Sometimes you may not be able to do that. Tethering helps here.

Sandy Zhang
Two types of nodes are defined in the draft. Reminds of graceful restart.
Should we just define “helper” and not “helped”

Jeffrey
Two way to advertise helping information. With the second approach “I am the
helper of X” the helped node need not do anything.

Greg Shepherd
Example to show incremental deployment as opposed to tail-end would be useful

3-draft-hb-bier-mldp-signaling-over-bier, 10 mins Hooman Bidgoli

WG-adoption
4 ppl have read it and support
Take it to list

4-draft-ietf-bier-multicast-http-response-01, 10 mins, Dirk Trossen
WGLC
2 ppl have read it and support
Take it to list

5-draft-ietf-bier-te-arch 10min, Toerless Eckert
WGLC
Please read the draft and respond on the list

6-draft-ietf-bier-mld-02, 10min, Stig Venaas
Stig
Should we extend IGMP/MLD messages to include sender info

Hooman
Isn’t this the same comments made for PIM.
It is fine the way it is

Jeffrey
We should do this for IGMP since we did it for PIM/MVPN

Stig
Discuss on mailing list

7-draft-zhang-bier-bierin6-03, Sandy Zhang
Toerless
Partial deployment capabilities?
Cost of copying entropy to flow-id.

Sandy/Tony
Primary intended use is HOMENET
Includes auto-discovery

Tony
Intended purely to be hop-by-hop but not directly-connected routers are an
unintended consequence

Greg Shepherd
Entropy is just a tag.

WG adoption
People who read agree (around 5 ppl in the room)
Take it to list

8-draft-zhang-bier-source-protection-00, 15min, Sandy Zhang
Jabber
BIER ping or traditional ping?

Sandy
Either can be used

Greg Shepherd
Is there a reason not to run BFD between both BFIRs and BFERs
It is quicker to know all my neighbors ahead of time
Is there an example where this is better

Jeffrey
MVPN fast failover already handles this for BGP-MVPN
You can use unicast FRR to switch faster.
This method can take longer timer?

Stig
Agree with Jeffrey. Maybe do forwarder election between BFIRs to decide when to
forward on a failure.

Greg Shepherd
Need to discuss all failure scenarios

Need to adopt?
Few hands up

Alia
MLDP FRR RFC talks about detecting link vs node failures. Might be interesting.

9-draft-hu-bier-bfd-04, 5min, Sandy Zhang
Greg
Take it to list

10-draft-ietf-bier-ipv6-requirements, 15min, Mike McBride
Mike
Please provide comments/pros/cons if solutions is in the document.