L2VPN minutes...

[Giles Heron <[e-mail]>]

Hi everyone,

First off an apology that I didn't manage to get the minutes out from the
Prague IETF out before the deadline.  I've included them below in this
message...

Secondly a promise from Nabil and I that we'll get the minutes out on time
after Quebec City. This time I won't rush off on holiday immediately after
IETF and forget all about the minutes ;-)

And thirdly a call for volunteers to take minutes or jabber scribe in Quebec
City.  There are a few gaps from our minutes in Prague, as we only had one
minute taker.  At a minimum I'd like to see one minute taker and one jabber
scribe so we can use the jabber transcript to fill in any gaps in the
minutes.

Thanks!

Giles


------

L2VPN Minutes - IETF 80 (Prague)


1)    Working Group Status Update - Giles

a)    2 RFCs published nice the last meeting, RFC 6074 and 6136

There is another in the queue, draft-ietf-l2vpn-vpls-bridge-interop-06 on
which there are a few comments from the mailing list, to be addressed and
will be fixed in the next week.

b)    ARP mediation draft - Giles asks Himanshu tasked to finish this draft;
agreed

c)    Nabil: OAM VPWS. ATM interworking is not seem as significant, only
vendor response was for FRF5. Next Steps - plan to wrap up the document as
soon as the vendor survey is completed.

d)    Giles - remaining draft updates.
            Multicast - VPLS, looking for more debate around this draft
            VPMS, looking for a republish from Yuji.
            VMPS, will cover multi-source but not multi-coverage

e)    multihoming02 - Wim at the mic saying it is ready for last call.

Giles - Please can we have more comments on existing drafts.


2)    Re-Charter of L2VPN - Giles

a)    Who has read it? A show of 6 or so hands of people who have read it.
Giles encouraging people to read the re-charter. 

b)    puts charter up on the screen and has fun with text sizing, then does
a quick overview as to the nature of L2VPN charter. 

c)    E-Tree competing approaches and needs to be resolved. E-VPN, same, a
few drafts so a need to resolve this.

d)    Are the milestones realistic? Need for everyone to think on this and
comment back to the group. Milestones for new work, and also for
old/existing work.

Stewart Bryant arrives.

Giles - Again comments requested, deletions comments are as welcome as
additions.

Giles - Anyone interested in MIB work is welcome also.

Giles - Work being done in the data-center area by L2VPN

Mic comments:

Luca - are we looking at this in both L2VPN and TRILL? Exactly what area is
L2 looking at?

Nabil - we need to ensure that work that gets picked up is real L2VPN work,
this can include data-center to data-center, but not a flat L2 network, VLAN
etc.

Rahul, couple of things to add. He likes the charter, the milestones are
missing VPMS auto discovery there is a draft for this he will resurrect.

Stewart, anyone doing MIBs? Call from Giles again for MIB people. Stewart as
a concern as to are there enough people prepared to do the work? Do we take
things off the charter if people are not interested in the work.

Sharam from Broadcom - Is VLAN in the charter or not? Stewart reply, if it
is a L2VPN then it is in.

Giles - can we also discuss E-Tree?

Luca - Are we going back to look at all the ways and details of doing
Ethernet frames (IEEE) over L2? 

Ali - If IEEE has done the work we should not duplicate, we should
interoperate. 

Unknown speaker (Broadcom) - QCN question, will L2VPN do interworking with
this? 

Giles - PWE drafts have this in the control word so do we need this again as
it is already done? Is GRE based VPN in the scope, is there any interest?

Giles (continuing with the charter)

New E-VPN and E-Tree in as items 7 and 10. Not seeing this group doing work
on TP or static VPLS

Milestones, some are done and the bad news is many have not.

Mic comments:

Stewart - do we have doc authors for July docs?


3)    VPLS analysis - Jie Dong

Overview of the ways to deploy VPLS and select the correct mechanism.

Mic Comments:

Ali - what is the point of this draft? Concern over the lack of detail of
doing a draft with less details, if you need the details then you still need
to read the draftsŠ

Unknown speaker - Similar comment

Luca - no draft needed on having a comparison between methods. How about a
draft on what operators thing is missing and or need to be improved.

Sasha - agrees with Luca, and perhaps an operators survey would be helpful
instead.

Unknown speaker (China Telecom) - does not see that there is enough detail
and gave some examples.

Nabil - anyone read the draft? Looks like it is going to the mailing list.
Market has asked for 2 solutions, they have been delivered, we don't track
the solutions in this group.


4)    VPLS IS-IS - Xiaohu Xu

Overview of the notion of adding IS-IS extension to support lightweight VPLS
without other protocols. (BGP/LDP etc)

Mic comments:

Ali - is this learning Mac addresses against IP addresses? If this is pure
data centre then should this be in TRILL.

Giles - is this a VPN?

Ali - existing solution can be used.

Ali - how is multicast done?

Xia - use ingress multicast

Ali - not efficient

Sasha - is this MPLS over IP or GRE

Unknown speaker - saying this is already done with TRILL and L2, this draft
needs to define exactly what space this is for.

Giles - in summary can we define the problem space and perhaps take this to
the mailing list?


5)    LDP Extensions for Optimized MAC Address Withdrawal in VPLS - Ran Chen

Mic comments:

Ali - not sure what the problem actually is,

Giles - can we see what in 4762 is broken via the mailing list.


6)    BGP MPLS Based Ethernet VPN - Rahul Aggarwal

Mic comments:

Unknown speaker - Interesting work

Giles - sounds like Rahul is ready for final comments on version 2.


7)    PBB E-VPN Ali Sajassi

Mic comments:

Luca - just to clarify the purpose.

Florin - comment on the TRILL part, TRILL interworking will be challenging.

Ali - TRILL interop is TBD and we need to work out how.


8)    VPLS Active/Active Multi-homing - Clarence Filsfils

Mic comments:

Unknown speaker - Very useful draft, but can we break the text in the draft
in 2, can we look at E-VPN and then the changes to VPLS signalling.

Clarence - only draft 1 so we can make these types of changes

Wim - there are 2 potential solutions, can we combine this one with PBB vpn
one, but perhaps analysis needed.

Rahul - thinks we need to combine what we mean by VPLS in this draft, and is
concerned that this draft might change the data plane

Clarence - disagrees

Lucy - wonders how we will do the signaling in the control plane

Clarence - early draft

Robert Raszuk - likes the draft

Unknown Speaker - question on how would we signal the MHID label under the
PW label.

Rahul - still thinks this changes the data plane in VPLS, and thinks that we
can use existing VPLS technology rather than changing the data plane.

Nabil - goes to mailing list to continue discussion. VPLS bridge model seems
to be changed here so this needs to be discussed.

Unknown speaker (Ericsson) - wanted to clarify the use of the labels 30 and
31 in the slides

Clarence - they are just examples, not proposed reserved labels.


9)    Extension to LDP-VPLS for E-Tree using Two PW - Daniel Cohn

Mic comments:

Florin - root and leaf PEs need to co-ordinate

Unknown speaker - Comment this breaks the bridge model

Sharam - backwards compatibility, legacy limitations if root and leaf not
support on legacy.

Unknown speaker - this could be done with a existing PW label instead

Luca - thinks the idea is possible, but doing this with interface parameters
would make this possible.


10)    VPLS PE model for E-Tree Support - Yuanlong Jiang

Mic comments:

Ali - are you suggesting mapping all C-VLAN in to a S-VLAN? If so this will
complicate this on bridging side of things. Once it is received you need to
double tag and or the VPLS instance?  Currently we map PW into the Bridge
ID, not the tag as well.

Unknown speaker - If the CE is tagged twice or QinQ, are we adding a tag of
modifying a tag?

Yuanlong - we could do both.


11)    Requirement and Framework of VPN-Oriented Cloud Services - Ning So

Mic comments:

Ali - are we saying that not all the requirements in this draft are for
L2VPN? Things like disk space are not part of this working group.

Lucy - On slide 3 are these functions done at operator layer of customer
level

Ning - can be done on both, depends on the tools made available by the
operator.

Himanshu - the dynamic creation of VPN via an operator is very interesting
and should be considered.

Giles - this if is an ops work then the requirements might want to be feed
back to the L2 group via the Ops WG.


12)    Wrap up from Giles and Nabil. 

Please can we see active mailing lists and the completion of some of the
drafts.