答复: RE: The status of the approaches to the E-Tree solution?

[Ran Chen<[e-mail]>]

Hi Sasha,
Plese see inline, thanks



Alexander Vainshtein <[e-mail]> 
发件人:  [e-mail]
2012/04/19 13:31

收件人
"Henderickx, Wim (Wim)" <[e-mail]>, "Rogers, 
Josh" <[e-mail]>, Lucy yong <[e-mail]>, Daniel 
Cohn <[e-mail]>, Sam Cao <[e-mail]>
抄送
"[e-mail]" <[e-mail]>
主题
RE: The status of the approaches to the E-Tree solution?






Wim,
Lots of thanks for a prompt response.

I think that operational aspects of the VLAN approach (e.g., how is the 
Global VID selected and distributed) should be examined in more detail.
[Ran] Agree, and there have already been such draft that describe "how is 
the Gloabal VID selected and distributed" in L2VPN WG, the link is:
http://tools.ietf.org/html/draft-chen-l2vpn-vpls-etree-vlan-01. 

At the same time it seems that the double PW approach carries with it too 
many operational problems.
E.g., no PWs are set up between Leaf-only PEs, but once one of these 
becomes a Mix PE, all the Leaf-only PEs must now recognize it as a valid 
peer. And if a Mix PE reverts to a Leaf only one, all its Leaf-only peers 
must drop it and shut down the relevant PWs...

My 2c,
     Sasha

________________________________________
From: Henderickx, Wim (Wim) [[e-mail]]
Sent: Thursday, April 19, 2012 7:21 AM
To: Alexander Vainshtein; Rogers, Josh; Lucy yong; Daniel Cohn; Sam Cao
Cc: [e-mail]
Subject: RE: The status of the approaches to the E-Tree solution?

Sasha, the VLAN approach allows for a similar operation as the CW does. It 
is orthogonal to the underlying PW deployment of VPLS.

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf Of 
Alexander Vainshtein
Sent: donderdag 19 april 2012 6:39
To: Rogers, Josh; Lucy yong; Daniel Cohn; Sam Cao
Cc: [e-mail]
Subject: RE: The status of the approaches to the E-Tree solution?

Hi all,
I fully understand that that what I am going to say is not very popular, 
but:

IMO one of the advantages of the CW-based solution is that it is 
orthogonal to usage (or non-usage) of P2MP PWs for effective delivery of 
BUN traffic.

Another advantage is preservation of full mesh of P2P PWs in a VPLS, and, 
in a more generic way, localization of effects of changes in the PE 
configuration.

In particular, adding a Leaf AC to a PE that previously has been only 
supporting Root ACs (or vice versa), removal of the last Leaf or last Root 
AC from a PE that previously has been supporting a mix etc. affect only 
the PE where this operation happens and not the rest of the PEs.

As for the need for HW changes that have been mentioned as a main 
disadvantage of the CW-based approach - I believe it strongly depends on 
specific implementations. And some changes in the forwarding process are 
required in any solution.

My 2c,
     Sasha



________________________________________
From: [e-mail] [[e-mail]] on behalf of Rogers, 
Josh [[e-mail]]
Sent: Wednesday, April 18, 2012 9:57 PM
To: Lucy yong; Daniel Cohn; Sam Cao
Cc: [e-mail]
Subject: Re: The status of the approaches to the E-Tree solution?

Again, the P2MP situation throws me.  Is this something that is used
commonly?

I'm under the impression that adding P2MP to any model results in a more
complex model.  Wether outside s-tag is used to differentiate, or
dedicated pw's are used for the same purpose, it seems both become more
complex.

Gile's comparison slide still concisely captures the differences between
these methods, in my opinion.  I haven't seen any new ideas or thoughts
brought to the group in the past week or two on the subject.  I would hate
for both proposed methods to die on the vine because we couldn't decide
between two methods that have nothing inherently wrong with either.

-Josh


On 4/18/12 1:53 PM, "Lucy yong" <[e-mail]> wrote:

>Send this again.
>
>Two PW approach can be complex too if the VPLS instance for E-Tree uses
>P2P PW for unicast traffic and P2MP PW for broadcast and unknown
>unicast traffic, and some P2MP PWs for multicast traffic. It may double
>all of them.
>
>Lucy
>
>-----Original Message-----
>From: Daniel Cohn [mailto:[e-mail]]
>Sent: Wednesday, April 18, 2012 1:42 PM
>To: Lucy yong; Rogers, Josh; Sam Cao
>Cc: [e-mail]
>Subject: RE: The status of the approaches to the E-Tree solution?
>
>I think the first option its the natural way to go. How is the processing
>in this case more complex?
>
>Thumb typed - please be tolerant
>
>Lucy yong <[e-mail]> wrote:
>
>
>
>Snipped..
>
>Multi-PW - On ingress PE, frame is placed onto either a Leaf-only P2MP PW
>(for traffic coming from a leaf AC), or onto a Root/Leaf P2MP PW (for
>traffic coming from a root AC)
>[[LY]] Not that simple. You construct either two P2MP PWs to all other
>PEs and let egress PE performing filtering, or construct one P2MP PW to
>leaf-only PEs and two P2MP PWs to root and leaf PEs and let ingress PE
>perform forwarding and filtering. Both make node process complex.
>
>[[LY]] VPLS is built with the mechanism utilizing P2P and P2MP PW for
>delivering the frames to remote PEs. We should utilize them with the
>minimized changes. Dual VLAN solution is simpler than Dual PW.
>
>Regards,
>Lucy
>
>
>I see how 2VLAN is simpler when P2MP PW's are involved, I think.  I
>haven't had any first hand experience with P2MP PW's, however, so don't
>feel terribly strong about this objection.  Is this a real problem for
>others (now or in the future), or is this objection in the weeds?
>
>I'm not sure the 'additional complexity' is notable, or even relevant.  I
>encourage others to speak up if they disagree, as P2MP PW is only
>conceptual to me, and I am unfamiliar with real-life use cases for it.
>
>Thanks,
>Josh
>
>
>
>On 4/18/12 10:30 AM, "Lucy yong" <[e-mail]> wrote:
>
>>Please see inline.
>>
>>-----Original Message-----
>>From: Sam Cao [mailto:[e-mail]]
>>Sent: Tuesday, April 17, 2012 7:14 AM
>>To: 'Daniel Cohn'; Lucy yong; 'Rogers, Josh'; 'Henderickx, Wim (Wim)';
>>[e-mail]; [e-mail];
>>[e-mail]
>>Cc: [e-mail]; [e-mail];
>>[e-mail]; [e-mail];
>>[e-mail]; [e-mail]
>>Subject: RE: The status of the approaches to the E-Tree solution?
>>
>>Yes, 2 pws are only needed between pes with both root and leaf acs after
>>improving Dual-PW approach. If consider P2MP, Dual-PW approach setup 2
>>P2MP
>>PWs if need. There is no difference between P2MP or normal PW setup. 
But,
>>can Leaf-ACs be bound to Root PE of P2MP PW?
>>
>>[[LY]] No, it makes complex in setting up P2MP PW. Should a PE with both
>>root and leaf ACs set up two or one P2MP PW to other PEs (some PE have
>>both root and leaf AC and some only have leaf ACs)?
>>Regards,
>>Lucy
>>
>>Regards,
>>
>>Yuqun (Sam) Cao
>>E-mail: [e-mail]
>>
>>
>>-----Original Message-----
>>From: Daniel Cohn [mailto:[e-mail]]
>>Sent: Tuesday, April 17, 2012 4:56 PM
>>To: Lucy yong; Rogers, Josh; Henderickx, Wim (Wim);
>>[e-mail];
>>[e-mail]; [e-mail]; Sam Cao
>>Cc: [e-mail]; [e-mail];
>>[e-mail]; [e-mail];
>>[e-mail]; [e-mail]
>>Subject: RE: The status of the approaches to the E-Tree solution?
>>
>>Adding Sam (as l2vpn@ is holding messages)
>>
>>DC
>>
>>-----Original Message-----
>>From: Lucy yong [mailto:[e-mail]]
>>Sent: Tuesday, April 17, 2012 12:39 AM
>>To: Daniel Cohn; Rogers, Josh; Henderickx, Wim (Wim);
>>[e-mail]; [e-mail];
>>[e-mail]
>>Cc: [e-mail]; [e-mail];
>>[e-mail]; [e-mail];
>>[e-mail]; [e-mail]
>>Subject: RE: The status of the approaches to the E-Tree solution?
>>
>>
>>Snipped,
>>
>>As we discussed extensively in the list, and as reflected in giles
>>slide, 2 pws are only needed between pes with both root and leaf acs,
>>which will typically be a small minority.
>>[[LY]] Don't know if the assumption is true. Even it is the case, both
>>approaches can benefit from it. I was off for a while and captures some
>>discussions now.
>>
>>Also as per giles slide, dual vlan can have scalability issues due to
>>additional lookup and double use of vlans in internal emulated lan
>>interface. Also there are potential backward compatibility issues with
>>silicon that doesn't support vlan mapping.
>>[[LY]] I was not in IETF83 meeting and wait on the meeting minutes. I am
>>not clear on all the issues. Could you be more specific? As I mentioned
>>in below, two PW approach makes VPLS transport construction and packet
>>forwarding more complex, I can see potential backward compatibility
>>issues with 2 PW solution.
>>
>>Regards,
>>Lucy
>>
>>Regards,
>>
>>Daniel
>>
>>Thumb typed - please be tolerant
>>
>>Lucy yong <[e-mail]> wrote:
>>
>>In my mind, the VLAN approach means dual vlan method.
>>
>>The main concern for CW approach is hardware support.
>>
>>Two PW approach can be complex too if the VPLS instance for E-Tree uses
>>P2P PW for unicast traffic and P2MP PW for broadcast and unknown unicast
>>traffic, and some P2MP PWs for multicast traffic. It may double all of
>>them.
>>
>>E-tree is an Ethernet service and there is already VLAN based solution
>>in IEEE, can we just utilize it without complicating VPLS transport
>>construction? This also makes interworking with Eth only network easier.
>>
>>Cheers,
>>Lucy
>>
>>-----Original Message-----
>>From: Rogers, Josh [mailto:[e-mail]]
>>Sent: Monday, April 16, 2012 8:35 AM
>>To: Lucy yong; Henderickx, Wim (Wim); '[e-mail]';
>>'[e-mail]'; '[e-mail]'
>>Cc: '[e-mail]'; '[e-mail]';
>>'[e-mail]'; '[e-mail]';
>>'[e-mail]'; '[e-mail]'
>>Subject: RE: The status of the approaches to the E-Tree solution?
>>
>>I believe the initial question was in regard to the CW method.  Are you
>>saying that you no longer are interested in that method in preference of
>>the dual vlan method?
>>
>>
>>
>>Lucy yong <[e-mail]> wrote:
>>
>>
>>Agree with Wim. VLAN approach is the best solution for E-Tree.
>>
>>Lucy
>>
>>-----Original Message-----
>>From: [e-mail] [mailto:[e-mail]] On Behalf
>>Of Henderickx, Wim (Wim)
>>Sent: Thursday, April 12, 2012 2:03 AM
>>To: '[e-mail]'; '[e-mail]';
>>'[e-mail]'
>>Cc: '[e-mail]'; '[e-mail]';
>>'[e-mail]'; '[e-mail]';
>>'[e-mail]'; '[e-mail]'
>>Subject: Re: The status of the approaches to the E-Tree solution?
>>
>>The vlan approach is superior as it also works for eth only networks,
>>etc. On top some vendors indicate hw issues with the cw approach. As
>>such we have dropped more or less the cw approach.
>>
>>Cheers,
>>Wim
>>_________________
>>sent from blackberry
>>
>>----- Original Message -----
>>From: Giles Heron [mailto:[e-mail]]
>>Sent: Thursday, April 12, 2012 08:22 AM
>>To: Simon Delord <[e-mail]>; Alexander Vainshtein
>><[e-mail]>
>>Cc: [e-mail] <[e-mail]>; Vladimir Kleiner
>><[e-mail]>; Andrew Sergeev
>><[e-mail]>; Idan Kaspit <[e-mail]>;
>>Mishael Wexler <[e-mail]>; Rotem Cohen
>><[e-mail]>
>>Subject: Re: The status of the approaches to the E-Tree solution?
>>
>>Sorry - the "anonymous presentation" was mine.  I should possibly have
>>put in a third column on the CW approach.  And hopefully the minutes
>>will be posted soon.
>>
>>We had various discussions, as Simon stated, and consensus seemed to be
>>forming around the two alternatives of two PWEs or two VLANs.  I believe
>>three of the authors of the CW approach are also authors of the two VLAN
>>approach and one is also an author of the two PWE approach. So perhaps
>>it's best to let those four individuals say which approach they prefer
>>and why?
>>
>>Giles
>>
>>On 10/04/2012 00:45, "Simon Delord" <[e-mail]> wrote:
>>
>>> Hi Alexander,
>>>
>>> You are right, no discussion on the WG mailing list recently, but
>>> there have been substantial discussions among the authors of various
>>> solution drafts off the mailing list. As far as I know, no consensus
>>> yet before ietf83, not sure the progress in the Paris WG meeting. I
>>> think the CW approach has not been rejected by the WG yet, or the WG
>>> has not yet decided on which one to adopt.
>>>
>>> Simon
>>>
>>> 2012/4/8 Alexander Vainshtein <[e-mail]>
>>>
>>>>  Hi all,
>>>>
>>>> Unfortunately I have not been able to attend the Paris IETF.
>>>>
>>>> However, looking up the L2VPN proceedings, I've found a short
>>>> anonymous presentation called "E-Tree Update"  (
>>>> http://www.ietf.org/proceedings/83/slides/slides-83-l2vpn-1.pptx).
>>>> This presentation discusses the differences of the E-Tree approaches
>>>> based on dedicated VLANs (as in
>>>> http://datatracker.ietf.org/doc/draft-cao-l2vpn-vpls-etree/?include_t
>>>> ext=1) and multiple PWs between the PEs (as in
>>>> http://datatracker.ietf.org/doc/draft-ram-l2vpn-etree-multiple-pw/?in
>>>> clude_te
>>>> xt=1)
>>>> and completely ignores the solution based on usage of the CW in the
>>>> PWs connecting the PEs (as in
>>>> http://datatracker.ietf.org/doc/draft-key-l2vpn-vpls-etree/?include_t
>>>> ext=1
>>>> ).
>>>>
>>>>
>>>>
>>>> The Minutes of the Paris L2VPN session are not yet available, but I
>>>> wonder whether the WG has taken a decision to reject the approach
>>>> based on the CW usage? I do not remember any recent discussion of
>>>> this topic on the WG mailing list.
>>>>
>>>>
>>>>
>>>> Regards, and lots of thanks in advance,
>>>>
>>>> Sasha
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> This e-mail message is intended for the recipient only and contains
>>>> information which is CONFIDENTIAL and which may be proprietary to ECI
>>
>>>> Telecom. If you have received this transmission in error, please
>>>> inform us by e-mail, phone or fax, and then delete the original and
>>>> all copies thereof.
>>>>
>>
>>
>>
>>
>>
>>This E-mail and any of its attachments may contain Time Warner Cable
>>proprietary information, which is privileged, confidential, or subject
>>to copyright belonging to Time Warner Cable. This E-mail is intended
>>solely for the use of the individual or entity to which it is addressed.
>>If you are not the intended recipient of this E-mail, you are hereby
>>notified that any dissemination, distribution, copying, or action taken
>>in relation to the contents of and attachments to this E-mail is
>>strictly prohibited and may be unlawful. If you have received this
>>E-mail in error, please notify the sender immediately and permanently
>>delete the original and any copy of this E-mail and any printout.
>>
>
>
>This E-mail and any of its attachments may contain Time Warner Cable
>proprietary information, which is privileged, confidential, or subject to
>copyright belonging to Time Warner Cable. This E-mail is intended solely
>for the use of the individual or entity to which it is addressed. If you
>are not the intended recipient of this E-mail, you are hereby notified
>that any dissemination, distribution, copying, or action taken in
>relation to the contents of and attachments to this E-mail is strictly
>prohibited and may be unlawful. If you have received this E-mail in
>error, please notify the sender immediately and permanently delete the
>original and any copy of this E-mail and any printout.


This E-mail and any of its attachments may contain Time Warner Cable 
proprietary information, which is privileged, confidential, or subject to 
copyright belonging to Time Warner Cable. This E-mail is intended solely 
for the use of the individual or entity to which it is addressed. If you 
are not the intended recipient of this E-mail, you are hereby notified 
that any dissemination, distribution, copying, or action taken in relation 
to the contents of and attachments to this E-mail is strictly prohibited 
and may be unlawful. If you have received this E-mail in error, please 
notify the sender immediately and permanently delete the original and any 
copy of this E-mail and any printout.
This e-mail message is intended for the recipient only and contains 
information which is CONFIDENTIAL and which may be proprietary to ECI 
Telecom. If you have received this transmission in error, please inform us 
by e-mail, phone or fax, and then delete the original and all copies 
thereof.
This e-mail message is intended for the recipient only and contains 
information which is CONFIDENTIAL and which may be proprietary to ECI 
Telecom. If you have received this transmission in error, please inform us 
by e-mail, phone or fax, and then delete the original and all copies 
thereof.