Re: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"

["Jeffrey (Zhaohui) Zhang" <[e-mail]>]

Hi Greg, all,

Please see zzh> below.

From: Greg Mirsky <[e-mail]>
Sent: Sunday, November 17, 2019 7:07 PM
To: Ketan Talaulikar (ketant) <[e-mail]>
Cc: John E Drake <jdrake=[e-mail]>; [e-mail]; Alexander Vainshtein <[e-mail]>; [e-mail]; Robert Raszuk <[e-mail]>; <[e-mail]> ([e-mail]) <[e-mail]>
Subject: Re: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"

Hi Ketan,
thank you for your thoughtful consideration of my comment. In the draft-voyer-spring-sr-replication-segment I find the following statement:
   A Replication segment at ingress node of Multi-point
   service replicates packets directly to each egress node of the
   service, without need for any state in the core of SR domain.
   Multiple Replication segments can be stitched together to build a
   tree in SR domain for Multi-point service; this is outside the scope
   of this document.
confusing. Firstly, what is the definition of  "core of SR domain"?

zzh> We can change the wording, but the intention is “internal routers”.

Secondly, isn't the suggestion of the potential stitching of Replication segments, an acknowledgment that the Replication Segment introduces the state to internal nodes of a single SR domain in order to realize a scalable Multi-point service within an SR domain?

Zzh> If one does not care about efficient replication, then one can use Ingress Replication (which has been used before SR and continue to apply in SR networks) and that does not have any state inside the network (you still have per-flow/tree state on the edge). Let me call that option 1.
Zzh> If you do care about efficient replication, then you either use BIER (which does not have per-flow state inside the network, though it does require new or programmable ASIC so that the new encapsulation and forwarding algorithm required by BIER can be handled) or introduce/keep per-flow/tree state inside the network. Let me call the two as option 2 and 3 respectively.
Zzh> The replication segment is intended for option #1 and #3. As I explained in another thread (of the same subject line), the replication segment can be used for both #1 and #3, and #1 is really just a special form of #3.
Zzh> BTW – traditional mLDP/RSVP-TE P2MP tunnels and PIM trees can all be considered as #3. In the SR era, (some) people don’t want to use PIM/mLDP/RSVP-TE P2MP anymore but as long as you want efficient replication w/o using BIER (option #2), then you still have to keep state on internal routers.

Zzh> So why we’re doing this replication segment?
Zzh> I see two important aspects of SR – a) no per-flow state on internal routers b) use of controllers to steer traffic.
Zzh> When it comes to efficient replication/multicast we can’t achieve a) w/o using BIER, but we can still achieve b), and the replication segment is the building block to achieve b). The separate “draft-voyer-pim-sr-p2mp-policy” describes how the building block can be used to achieve b).
Zzh> Thanks.
Zzh> Jeffrey

Regards,
Greg

On Sun, Nov 17, 2019 at 5:21 PM Ketan Talaulikar (ketant) <[e-mail]<mailto:[e-mail]>> wrote:
Hi Greg,

Please check inline.

From: Greg Mirsky <[e-mail]<mailto:[e-mail]>>
Sent: 17 November 2019 13:14
To: Ketan Talaulikar (ketant) <[e-mail]<mailto:[e-mail]>>
Cc: John E Drake <jdrake=[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>; Alexander Vainshtein <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>; Robert Raszuk <[e-mail]<mailto:[e-mail]>>; <[e-mail]<mailto:[e-mail]>> ([e-mail]<mailto:[e-mail]>) <[e-mail]<mailto:[e-mail]>>
Subject: Re: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"

Hi Ketan,
thank you for your suggestion. As you've pointed out, the draft in discussion introduces a new segment type, Replication Segment, to realize p2mp behavior in an SR domain. Looking into RFC 8402, I find the following statement regarding multicast:
6.  Multicast

   Segment Routing is defined for unicast.  The application of the
   source-route concept to Multicast is not in the scope of this
   document.

Hence, I believe, is the valid question to where the possible impact of multicast on the architecture of segment routing should be discussed, described.
[KT] Sure and I think I understand your point. Multicast in SR networks is for the WG to discuss and the chairs/AD to guide. However, I believe this is not related or blocking for the document in the subject line (as also echoed by Sasha). Just wanted to clarify this part.

Thanks,
Ketan

I hope that clarifies what has been the topic of discussion on this thread.

Regards,
Greg

On Sun, Nov 17, 2019 at 12:09 PM Ketan Talaulikar (ketant) <[e-mail]<mailto:[e-mail]>> wrote:
Hi Greg/Sasha/All,

I really wonder whether we are talking about the same document anymore. The subject of this thread is https://tools.ietf.org/html/draft-voyer-spring-sr-replication-segment-00<https://urldefense.com/v3/__https:/tools.ietf.org/html/draft-voyer-spring-sr-replication-segment-00__;!8WoA6RjC81c!XeZJtRzefFgiJLe2wLYn90cJp25eXTwbA0DX8pxmJctNWhRTkFfbtETOZZB2nDCK$>

It is indeed possible that you and others are referring to some other document(s)?

From reading of the draft, one can see that :

  *   It does not deal with multicast group joins/receivers or senders
  *   It does not build multicast trees
  *   It does not talk about multicast flows
  *   It simply introduces a new type of segment called Replication Segment (p2mp) for a specific local node forwarding behaviour that is in line with the Spring Charter (see below)

o New types of segments mapping to forwarding behaviour (e.g., local
ingress replication, local forwarding resources, a pre-existing
replication structure) if needed for new usages.

Can you please take another quick read over the draft with the above context in mind? I am positive that you will see that this is not getting multicast work in Spring – that is being worked on in other WGs.

Thanks,
Ketan

From: spring <[e-mail]<mailto:[e-mail]>> On Behalf Of Greg Mirsky
Sent: 17 November 2019 11:39
To: John E Drake <jdrake=[e-mail]<mailto:[e-mail]>>
Cc: [e-mail]<mailto:[e-mail]>; Alexander Vainshtein <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>; Robert Raszuk <[e-mail]<mailto:[e-mail]>>; <[e-mail]<mailto:[e-mail]>> ([e-mail]<mailto:[e-mail]>) <[e-mail]<mailto:[e-mail]>>
Subject: Re: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"

Dear All,
I concur with Sasha and John. Intended ingress replication of a particular flow, though using a unicast destination address, is still a multicast.

Regards,
Greg

On Thu, Nov 14, 2019 at 5:36 AM John E Drake <jdrake=[e-mail]<mailto:[e-mail]>> wrote:
Robert,

As Sasha and I have indicated, your position is your own and is not consistent with the majority of work on this topic.  I’m fine w/ agreeing to disagree.

John
Sent from my iPhone

On Nov 14, 2019, at 2:57 AM, Robert Raszuk <[e-mail]<mailto:[e-mail]>> wrote:

John,

> Your claim that ingress replication is not multicast is, at best, a stretch.

I use a very basic and simple rule of thumb ... if address of my packet is a multicast address then it is multicast if not it is unicast.

Ref: https://www.iana.org/assignments/multicast-addresses/multicast-addresses.xhtml<https://urldefense.com/v3/__https:/www.iana.org/assignments/multicast-addresses/multicast-addresses.xhtml__;!8WoA6RjC81c!QFbPjRVo7hB9622FCxHnivS8PVicSm5TCW9kaF-KRqhC3G7uLL0tCrGUUxL2sAQ$>

Solution as described in draft-voyer-spring-sr-replication-segment does not seems to be requiring multicast addresses hence it is applicable to pure unicast networks.

Thx,
Robert.


On Wed, Nov 13, 2019 at 10:20 PM John E Drake <[e-mail]<mailto:[e-mail]>> wrote:
Robert,

I’m sorry for the confusion.  My only point was that MVPN provides the reference architecture for dealing w/ multicast using a multiplicity of tunnel types in a consistent manner, as Sasha alluded to in his mention of PMSI.  Your claim that ingress replication is not multicast is, at best, a stretch.

Yours Irrespectively,

John



Juniper Business Use Only
From: Robert Raszuk <[e-mail]<mailto:[e-mail]>>
Sent: Wednesday, November 13, 2019 3:55 PM
To: John E Drake <[e-mail]<mailto:[e-mail]>>
Cc: Alexander Vainshtein <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>; [e-mail]<mailto:[e-mail]>; <[e-mail]<mailto:[e-mail]>> ([e-mail]<mailto:[e-mail]>) <[e-mail]<mailto:[e-mail]>>
Subject: Re: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"

Hi John,

> Further, ingress replication has been part of MVPN since forever.

Just curious how is this at all relevant for this discussion ?

Do I have to roll out MVPN monster to split my unicast UDP stream to few receivers at selected network point ?

And last but not least who said this is at all related to "ingress replication" ??? Ingress to p2mp segment can be at any SR midpoint in the network. Are you suggesting to run MVPN apparatus with manual tree building ? Whow :)

Thx,
R.






On Wed, Nov 13, 2019 at 9:40 PM John E Drake <[e-mail]<mailto:[e-mail]>> wrote:
Hi,

I think Sasha has a valid point.  Further, ingress replication has been part of MVPN since forever.

Yours Irrespectively,

John



Juniper Business Use Only
From: spring <[e-mail]<mailto:[e-mail]>> On Behalf Of Alexander Vainshtein
Sent: Wednesday, November 13, 2019 9:26 AM
To: Robert Raszuk <[e-mail]<mailto:[e-mail]>>
Cc: spring@ietf..org<mailto:[e-mail]>; [e-mail]<mailto:[e-mail]>; <[e-mail]<mailto:[e-mail]>> ([e-mail]<mailto:[e-mail]>) <[e-mail]<mailto:[e-mail]>>
Subject: Re: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"

Robert,
Lots of thanks for a prompt response.

You seem to imply that a multicast distribution tree that is built, say, by an SDN controller and used, say, to act as a PMSI in the mVPN application, is not really a multicast.  Personally I disagree, but this is a matter of taste and terminology.

What looks unambiguous to me is that:

  *   The WG charter explicitly mentions ingress replication as one of “new types of segments mapping to forwarding behavior” that “may require architectural extensions”
  *   The current architecture document does not cover any such segment type (whether because such segments have been considered as related to multicast by the authors, or for some other reason is not all that important. )
Therefore my concern remains unresolved regardless of whether ingress replication is or is not formally considered as multicast.

Regards,
Sasha

Office: [phone]
Cell:      [phone]
Email:   [e-mail]<mailto:[e-mail]>

From: Robert Raszuk <[e-mail]<mailto:[e-mail]>>
Sent: Wednesday, November 13, 2019 4:15 PM
To: Alexander Vainshtein <[e-mail]<mailto:[e-mail]>>
Cc: <[e-mail]<mailto:[e-mail]>> ([e-mail]<mailto:[e-mail]>) <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>; [e-mail]<mailto:[e-mail]>
Subject: Re: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"

Sasha,

If I have some content and I send it to you and your neighbour as two unicast streams am I suddenly doing multicast ?

IMHO N number of replicated unicasts is still not a multicast.

Multicast in my definition requires  multicast groups, receiver joins, tree building protocols etc ... and this draft does not suggest any of this. IN contrast it just describes how can we have p2mp unicast distribution ... call it fan out node.

Thx,
R.





On Wed, Nov 13, 2019 at 12:42 PM Alexander Vainshtein <[e-mail]<mailto:[e-mail]>> wrote:

Hi all,

I have a question regarding adoption of draft-voyer-sr-spring-replication-segment as a SPRING WG document.



These concerns are based on the following:

1.       This draft (both based on its title and on its content) deals with local (in the Root node) ingress replication which, in its turn, is one of the issues that could be used for delivery of multicast.

2.       Local ingress replication is mentioned in the SPRING WG Charter as one of the “New types of segments mapping to forwarding behavior”. The charter further says that “Any of the above <Sasha: New types of segments> may require architectural extensions”

3.       The current (and, AFAIK, the only existing) Segment Routing Architecture document (RFC 8402<https://urldefense.com/v3/__https:/clicktime.symantec.com/34qM9QogJnh1eY5nZPXYAkA6H2?u=https*3A*2F*2Ftools.ietf.org*2Fhtml*2Frfc8402__;JSUlJSU!8WoA6RjC81c!TeDGZsCZsxVU3U1A-_hQaYhZsmLZFF4oF-lGSpNnOmTa-zUl6jfGkGEUOvwkLSU$>) explicitly states in Section 6 that “Segment Routing is defined for unicast. The application of the source-route concept to Multicast is not in the scope of this document”.

The combinations of observations above strongly suggests to me that a document defining multicast-related extensions of segment routing architecture should be very useful (if not mandatory) for progressing the Replication Segment draft. From my POV the Replication Segment draft is not (and is not intended to be) such a document.



I wonder if there is an intention to produce such a document in the timeframe that could be relevant for discussion of the Replication Segment draft.



Nothing in this message should be interpreted as my objection to (or support of) adoption of the Replication Segment draft as a WG document per se.

Bit I find it difficult to take a position any which way without a clear and commonly agreed upon framework for multicast in segment routing.



Regards,

Sasha



Office: [phone]

Cell:      [phone]

Email:   [e-mail]<mailto:[e-mail]>



-----Original Message-----
From: spring <[e-mail]<mailto:[e-mail]>> On Behalf Of IETF Secretariat
Sent: Tuesday, November 12, 2019 7:06 PM
To: [e-mail]<mailto:[e-mail]>; spring-chairs@ietf..org<mailto:spring-chairs@ietf..org>; [e-mail]<mailto:[e-mail]>
Subject: [spring] The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state "Candidate for WG Adoption"





The SPRING WG has placed draft-voyer-spring-sr-replication-segment in state Candidate for WG Adoption (entered by Bruno Decraene)



The document is available at

https://clicktime.symantec.com/3EMJRgfTdX6UyWKGnMPiVwZ6H2?u=https%3A%2F%2Fdatatracker.ietf.org%2Fdoc%2Fdraft-voyer-spring-sr-replication-segment%2F<https://urldefense.com/v3/__https:/clicktime.symantec.com/3EMJRgfTdX6UyWKGnMPiVwZ6H2?u=https*3A*2F*2Fdatatracker.ietf.org*2Fdoc*2Fdraft-voyer-spring-sr-replication-segment*2F__;JSUlJSUl!8WoA6RjC81c!TeDGZsCZsxVU3U1A-_hQaYhZsmLZFF4oF-lGSpNnOmTa-zUl6jfGkGEUHVCWfyU$>



Comment:

IPR call:

https://clicktime.symantec.com/3KG7A2qM3Xf2eqDctGju1e66H2?u=https%3A%2F%2Fmailarchive.ietf.org%2Farch%2Fmsg%2Fspring%2F_stJjBM5K6vr7QYw0HRKf-z0_us<https://urldefense.com/v3/__https:/clicktime.symantec.com/3KG7A2qM3Xf2eqDctGju1e66H2?u=https*3A*2F*2Fmailarchive.ietf.org*2Farch*2Fmsg*2Fspring*2F_stJjBM5K6vr7QYw0HRKf-z0_us__;JSUlJSUlJQ!8WoA6RjC81c!TeDGZsCZsxVU3U1A-_hQaYhZsmLZFF4oF-lGSpNnOmTa-zUl6jfGkGEUfVccUWU$>



_______________________________________________

spring mailing list

[e-mail]<mailto:[e-mail]>

https://clicktime.symantec.com/3AtNGCKcyM5uigFH55oARZ86H2?u=https%3A%2F%2Fwww.ietf.org%2Fmailman%2Flistinfo%2Fspring<https://urldefense.com/v3/__https:/clicktime.symantec.com/3AtNGCKcyM5uigFH55oARZ86H2?u=https*3A*2F*2Fwww.ietf.org*2Fmailman*2Flistinfo*2Fspring__;JSUlJSUl!8WoA6RjC81c!TeDGZsCZsxVU3U1A-_hQaYhZsmLZFF4oF-lGSpNnOmTa-zUl6jfGkGEUhKjFqCs$>

___________________________________________________________________________

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.
___________________________________________________________________________
_______________________________________________
spring mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/spring<https://urldefense.com/v3/__https:/clicktime.symantec.com/3KSi9HHVnunMDQNLd2U3Sij6H2?u=https*3A*2F*2Fwww.ietf.org*2Fmailman*2Flistinfo*2Fspring__;JSUlJSUl!8WoA6RjC81c!TeDGZsCZsxVU3U1A-_hQaYhZsmLZFF4oF-lGSpNnOmTa-zUl6jfGkGEUZIWr6Wk$>

___________________________________________________________________________

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.
___________________________________________________________________________
_______________________________________________
spring mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/spring<https://urldefense.com/v3/__https:/www.ietf.org/mailman/listinfo/spring__;!8WoA6RjC81c!XeZJtRzefFgiJLe2wLYn90cJp25eXTwbA0DX8pxmJctNWhRTkFfbtETOZWB2K7_O$>