Re: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

"Les Ginsberg (ginsberg)" <ginsberg@cisco.com> Fri, 10 August 2018 03:52 UTC

Return-Path: <ginsberg@cisco.com>
X-Original-To: lsr@ietfa.amsl.com
Delivered-To: lsr@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A8CC6130EF2; Thu, 9 Aug 2018 20:52:10 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -14.51
X-Spam-Level:
X-Spam-Status: No, score=-14.51 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_HI=-5, SPF_PASS=-0.001, T_DKIMWL_WL_HIGH=-0.01, USER_IN_DEF_DKIM_WL=-7.5] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (1024-bit key) header.d=cisco.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id fGtPRC_Zs_y8; Thu, 9 Aug 2018 20:52:03 -0700 (PDT)
Received: from rcdn-iport-4.cisco.com (rcdn-iport-4.cisco.com [173.37.86.75]) (using TLSv1.2 with cipher DHE-RSA-SEED-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id CB474127148; Thu, 9 Aug 2018 20:52:02 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=cisco.com; i=@cisco.com; l=194770; q=dns/txt; s=iport; t=1533873122; x=1535082722; h=from:to:cc:subject:date:message-id:references: in-reply-to:mime-version; bh=ulYvUs4QQnvyT/RFCTvG/EqRLAX7S+HTw2JRihahfgU=; b=AmX0OdXBoFaso0CR2hEOPdNOH9CnDhvR4gL4x3R3hUdIXxjitquWvjyu 8iUJaR5Ig3+PtBhD2CecM8Nn5+1cB7PIMpC1XWPtTRhhIWNQByK4Y9LWF L9KEV4jNfSI7Ql55UaYnOdxaLuWUJh0RQprrLCPX62hxXSxGLcL6VmiNs M=;
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: A0CiAQB2C21b/40NJK1cGQEBAQEBAQEBAQEBAQcBAQEBAYJXSS5jfygKg1aUPoINlX4UgWMDCxgBDIRHAheDCiE2FgECAQECAQECbRwMhTcBAQEEAQEMDAECBgQGOAIEAwsMBAIBBgIRAQMBASEBBgMCAgIlCxQDBggCBAENBQgTgwaBHGQPjzybS3szimOJFBeBQT+BEQGCFH6DGwEBAgEBFoEUARIBBy8PBgoIgkOCVQKHZhmKHwqIJQkChh2Cb4Y6gVZIg2CCdX+ESIp9h1sCERSBJCQELSY7XA0IcBUaIYJpCYIcF4hZhT5vAYxAgR+BGwEB
X-IronPort-AV: E=Sophos;i="5.53,218,1531785600"; d="scan'208,217";a="435990549"
Received: from alln-core-8.cisco.com ([173.36.13.141]) by rcdn-iport-4.cisco.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 10 Aug 2018 03:51:59 +0000
Received: from XCH-RCD-003.cisco.com (xch-rcd-003.cisco.com [173.37.102.13]) by alln-core-8.cisco.com (8.15.2/8.15.2) with ESMTPS id w7A3px8G001232 (version=TLSv1.2 cipher=AES256-SHA bits=256 verify=FAIL); Fri, 10 Aug 2018 03:51:59 GMT
Received: from xch-aln-001.cisco.com (173.36.7.11) by XCH-RCD-003.cisco.com (173.37.102.13) with Microsoft SMTP Server (TLS) id 15.0.1320.4; Thu, 9 Aug 2018 22:51:58 -0500
Received: from xch-aln-001.cisco.com ([173.36.7.11]) by XCH-ALN-001.cisco.com ([173.36.7.11]) with mapi id 15.00.1320.000; Thu, 9 Aug 2018 22:51:58 -0500
From: "Les Ginsberg (ginsberg)" <ginsberg@cisco.com>
To: "stephane.litkowski@orange.com" <stephane.litkowski@orange.com>, "Van De Velde, Gunter (Nokia - BE/Antwerp)" <gunter.van_de_velde@nokia.com>, DECRAENE Bruno IMT/OLN <bruno.decraene@orange.com>
CC: "lsr@ietf.org" <lsr@ietf.org>, "draft-ietf-isis-mpls-elc@ietf.org" <draft-ietf-isis-mpls-elc@ietf.org>, "徐小虎(义先)" <xiaohu.xxh@alibaba-inc.com>
Thread-Topic: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt
Thread-Index: AQHUKv4QtGYMEiS0k0yl0LxTrFwJ/qSuMyWQgAAIHNCAAEvDcIAENQUAgAAF04CAABPFAIADsieggADy/QCAAOG6EA==
Date: Fri, 10 Aug 2018 03:51:58 +0000
Message-ID: <1ce096814d6f450a816c8def38054d20@XCH-ALN-001.cisco.com>
References: <153304602040.5962.1405809920091386791@ietfa.amsl.com> <24381_1533241460_5B636874_24381_173_1_53C29892C857584299CBF5D05346208A47B01EEB@OPEXCLILM21.corporate.adroot.infra.ftgroup>, <27791_1533243037_5B636E9D_27791_243_1_53C29892C857584299CBF5D05346208A47B01F41@OPEXCLILM21.corporate.adroot.infra.ftgroup> <4065204c-4b69-472f-94d0-e47b62d603e1.xiaohu.xxh@alibaba-inc.com> <27380_1533282342_5B640826_27380_325_1_53C29892C857584299CBF5D05346208A47B02616@OPEXCLILM21.corporate.adroot.infra.ftgroup> <f2b05c3ef578459d852965d812e4e0a1@XCH-ALN-001.cisco.com> <11215_1533315641_5B648A39_11215_243_1_53C29892C857584299CBF5D05346208A47B03107@OPEXCLILM21.corporate.adroot.infra.ftgroup> <585acbfaf4a44cc8aeaf0c9cf7dc8246@XCH-ALN-001.cisco.com> <12855_1533543961_5B680619_12855_159_1_f7bd4b69-908c-46a9-bcac-a8540d719f8c@OPEXCLILM6C.corporate.adroot.infra.ftgroup> <AM5PR0701MB17293B463037BA08310D062BE0200@AM5PR0701MB1729.eurprd07.prod.outlook.com> <13288_1533549460_5B681B94_13288_147_1_4121a642-0212-4203-8b9a-a69877e23c49@OPEXCLILM5E.corporate.adroot.infra.ftgroup> <b443d09eb5c64e88b11f9d0afaab7cc4@XCH-ALN-001.cisco.com> <30692_1533804824_5B6C0118_30692_478_1_ed40eb26-835a-46ca-b479-11e706789c19@OPEXCLILMA1.corporate.adroot.infra.ftgroup>
In-Reply-To: <30692_1533804824_5B6C0118_30692_478_1_ed40eb26-835a-46ca-b479-11e706789c19@OPEXCLILMA1.corporate.adroot.infra.ftgroup>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach:
X-MS-TNEF-Correlator:
x-ms-exchange-transport-fromentityheader: Hosted
x-originating-ip: [10.24.44.250]
Content-Type: multipart/alternative; boundary="_000_1ce096814d6f450a816c8def38054d20XCHALN001ciscocom_"
MIME-Version: 1.0
X-Outbound-SMTP-Client: 173.37.102.13, xch-rcd-003.cisco.com
X-Outbound-Node: alln-core-8.cisco.com
Archived-At: <https://mailarchive.ietf.org/arch/msg/lsr/w-YV7nCI5xUXVeCiV595qkwxSqo>
Subject: Re: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt
X-BeenThere: lsr@ietf.org
X-Mailman-Version: 2.1.27
Precedence: list
List-Id: Link State Routing Working Group <lsr.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/lsr>, <mailto:lsr-request@ietf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/lsr/>
List-Post: <mailto:lsr@ietf.org>
List-Help: <mailto:lsr-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/lsr>, <mailto:lsr-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 10 Aug 2018 03:52:11 -0000

Stephane –

Responses inline – look for LES2

From: stephane.litkowski@orange.com <stephane.litkowski@orange.com>
Sent: Thursday, August 09, 2018 1:54 AM
To: Les Ginsberg (ginsberg) <ginsberg@cisco.com>; Van De Velde, Gunter (Nokia - BE/Antwerp) <gunter.van_de_velde@nokia.com>; DECRAENE Bruno IMT/OLN <bruno.decraene@orange.com>
Cc: lsr@ietf.org; draft-ietf-isis-mpls-elc@ietf.org; 徐小虎(义先) <xiaohu.xxh@alibaba-inc.com>
Subject: RE: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Hi Les,

Please see inline comments.

Brgds,

Stephane


From: Les Ginsberg (ginsberg) [mailto:ginsberg@cisco.com]
Sent: Thursday, August 09, 2018 01:53
To: LITKOWSKI Stephane OBS/OINIS; Van De Velde, Gunter (Nokia - BE/Antwerp); DECRAENE Bruno IMT/OLN
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>; 徐小虎(义先)
Subject: RE: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Stephane –

There are some issues you raise that I think are easily resolved/agreed upon – some others that may require more discussion.
Responses inline.

From: stephane.litkowski@orange.com<mailto:stephane.litkowski@orange.com> <stephane.litkowski@orange.com<mailto:stephane.litkowski@orange.com>>
Sent: Monday, August 06, 2018 2:58 AM
To: Van De Velde, Gunter (Nokia - BE/Antwerp) <gunter.van_de_velde@nokia.com<mailto:gunter.van_de_velde@nokia.com>>; Les Ginsberg (ginsberg) <ginsberg@cisco.com<mailto:ginsberg@cisco.com>>; DECRAENE Bruno IMT/OLN <bruno.decraene@orange.com<mailto:bruno.decraene@orange.com>>
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>; 徐小虎(义先) <xiaohu.xxh@alibaba-inc.com<mailto:xiaohu.xxh@alibaba-inc.com>>
Subject: RE: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Hi Gunter,

IMO, there are multiple cases to distinguish.

The basic case (Adj-SIDs & Prefix-SIDs, single area, single domain, single protocol):

-          ELC as a node property is a good thing. I personally do not see a use case to get some segments to be ELC and some others not to be ELCs.

[Les:] Good – I think there is general agreement on this point.

Multiarea case:

-          When prefixes are leaked from one area to the other, we are losing the per node properties (like the ERLD). If we want to get benefit of EL, the ingress node in one area should get the ELC status of nodes in the other areas.

-          One solution would be to propagate the ELC as part of the prefix advertisement (when the prefix is leaked). We are still losing the ERLD which may impact the loadbalancing (EL/ELI placement not efficient).

-          Another solution could be for the ingress node to retrieve the ERLD from another source of information like BGP-LS but that’s not trivial as the ingress must know the best route to the tail from the ABR point of view.

[Les:] There is no issue with multiarea.  Repeating some points made earlier in the thread:

Both IGPs have the ability to leak router capabilities/info between areas. Which means the node based attributes ELC and ERLD can be available domain-wide.
We also have the ability to leak source router-id with the leaked prefixes (OSPF has a gap here but we have a commitment to address that with a modest extension to the protocol – which needs to be done for multiple reasons.)
So all the information necessary is available in all areas.

[SLI2] Ok cool. If we go this way, don’t you think it would be useful to add a statement in the draft telling that such information (router cap+source-id) needs to be leaked ?

[Les2:] Agreed. I think I already made that statement earlier in the thread.

Multiprotocol case:

-          It’s not really different than the multiarea case.

[Les:] Multiprotocol is quite different than multiarea.
In the case of multiprotocol we do not have the ability to know/advertise the original source router-id nor the node capabilities in the destination protocol. It is not even guaranteed that reachability to the router-id of the source protocol is available in the area of the destination protocol.
This is where a controller based solution is most advantageous as it has the ability to learn all of that information from all of the protocols enabled in the domain.

[SLI2] Based on what you previously said, I agree that it is different.
We have such deployment in our network, so we need a solution for the multiprotocol stuff.
For this use  case, I agree with Bruno that setting the ELC on a per prefix basis may be the easiest solution.
As a consequence, if per prefix ELC is required for the multiprotocol case, there is no more reason to maintain the per node ELC.

[Les2:]  It is not that simple. It is still necessary to know ERLD – and there is no mechanism to leak that between protocols. There are also the issues with router-id I mentioned above.
I appreciate that multiprotocol may be a real world deployment case – but I think the best suggestion here is to use a controller. If that does not suffice then there is much more required than advertising ELC along with prefixes.

SRMS and SR-LDP stitching:

•         SR to LDP: the LDP tail-end may be ELC, if an SR ingress node wants to use an ELI/EL on top of the prefix SID associated to the LDP FEC, we need to propagate the ELC known from LDP (attached to the FEC) to the SR advertisement. The SR-LDP stitching node is not a tail-end node, it will do a label swap, not a pop then push. So an entropy label pushed by the SR ingress node will be carried to the tail-end LDP node. Am I wrong ?

•         LDP to SR: the stitching node knows if the tail-end SR node is ELC (through the IGP advertisement). It may propagate the ELC into LDP. All the prefixes of the tail-end SR node which are LDP FECs can get the ELC set if the SR tail-end node is ELC.
I do not see any strict prescription requirement for the LSP here.

[Les:] Here there is more room for differences of opinion.
One viewpoint is that SR-LDP interworking is a transition strategy. Any solution which attempts to exchange ELC/ELRD information between the SR<->LDP portions of the network will be “messy” and perhaps not worth the trouble as we hope that the transition will be “short-lived”.

Another viewpoint could be that if the transition period is long enough it may be useful to have such support.

But as any solution (even if we were to advertise ELC per prefix) would be non-trivial, I think there needs to be a strong case for taking this on.
Do you believe we have a strong case here?

[SLI2] No, I’m not telling that there is a strong case. If we explicitly do not want to support it, it would be worth mentioning it in the draft.
I see a real complexity in the advertisement if we want to make it working. Disaggregation of the SRMS range may be necessary. In the worst case, a per /32 range will be required with properties for each range. I personally do not want to go in this but again the draft should mention that we are aware of this.


[Les2:] OK – I take this to mean we agree we do not have to solve this issue. As for covering this in a draft, I agree it would be good to discuss such issues – but I don’t think the ELC drafts are the best place. It would be better to put this into a generic policy framework draft as the issues are not confined to SR-LDP. I see some similarity to the multiprotocol case. Sharing attributes between different protocols – be they two IGPs or two label distribution mechanisms – is problematic.

Binding SID:

-          Binding SID associated to an SRTE LSP: if an ingress node wants to put an ELI/EL next to a binding SID, it needs to ensure that the tail-end of the binding SID is ELC. In addition, the expansion of the binding SID may introduce new ELI/ELs  as part of the new added label stack.

-          Binding SID associated to an RSVP-TE LSP: this case was removed from the SR extensions, but if it is revived in future, the ELC status of the RSVP-TE LSP may need to be propagated to the SR domain as part of the binding SID advertisement if we want the ingress to be able to push an ELI/ELI next to the BSID.

[Les:] I don’t appreciate the problem here.
I think you are talking about two possible EL use cases.

1)An ingress node which imposes the binding-SID may wish to utilize entropy on the path to the owner of the Binding-SID. As the node sid (or possibly an adj-sid advertised by the owner of the binding sid) will be needed in the label stack, the ingress node still knows – independent of the binding-sid, whether the owner is ELC.

2)When the packet arrives at the node which owns the binding SID, that node may wish to include EL in the label stack which it imposes to implement the policy defined by the binding SID. But as the ELC of the LSP tail end is known this should not be a problem. And I do not see why the ingress node that imposed the binding-sid needs to know whether the owner of the binding-sid will/will not impose EL.

[SLI2] I think there is a slight misunderstanding here.
Let me try to do a simple figure:

A --- B ---- C ---- D ---- E ---- F --- G

A is the ingress node.
C is owner of a BSID. The BSID corresponds to the path CDEF.

A pushes a label stack {Adj_BC, BSID_C, Node_G}
A may insert an ELI/EL next to the BSID_C like : {Adj_BC, BSID_C, ELI, EL, Node_G}.
In order to do so, A must know if the tail-end of the LSP associated with BSID_C is ELC.
When the packet enters C with { BSID_C, ELI, EL, Node_G}, C expands the BSID with the following stack { Adj_DE, Adj_EF }. It swaps BSID_C to Adj_EF and pushes Adj_DE. The packet exists C with the following label stack {Adj_DE, Adj_EF, ELI, EL, Node_G}. Optionally, C may be able to add an additional ELI/EL when pushing adj_DE. But let’s forget this for this example.
When the packet enters F, F needs to be able to process the ELI/ELI. That’s the reason why A needs to be aware of node F ELC through the binding SID advertisement.

The problem comes because of the label swap on C. The other approach could be to tell that the LSP associated with the BSID label stops on the owner of the BSID and another one starts, each one being independent in term of ELC.
When the packet enters C with { BSID_C, ELI, EL, Node_G}, C expands the BSID with the following stack { Adj_DE, Adj_EF }. It pops BSID_C and the ELI/EL and pushes {Adj_DE, Adj_EF}.
In this solution, the ELC property of the BSID is the ELC property of C.

From an MPLS architecture point of view, doing a swap+push or pop+double push does not mean the same thing in term of LSP continuity and there is some implication for the ELC behind.

I do not see any document telling about the forwarding behavior expected for the binding SID.

[Les2:] A is trying to setup an LSP from A->G. As such, A should only be interested in ELC of G.
The fact that C may insert an “inner LSP” to F should be irrelevant to A. But I take your point that there are two possible behaviors on C and at the moment which behavior is expected is not specified.

1)(ELI, EL) following a BSID is associated with the BSID owner and should be popped by C when it imposes the LSP associated with the BSID

Or

2)(ELI,EL) following a BSID is associated with the endpoint of the LSP defined by the ingress node (A in this case) and should NOT be popped when the inner LSP associated with the BSID is pushed.

I prefer #2 – it means the introduction of BSID does not change existing EL behavior. But I am interested in what other folks think.

  Les




From: Van De Velde, Gunter (Nokia - BE/Antwerp) [mailto:gunter.van_de_velde@nokia.com]
Sent: Monday, August 06, 2018 10:47
To: LITKOWSKI Stephane OBS/OINIS; Les Ginsberg (ginsberg); DECRAENE Bruno IMT/OLN
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>; 徐小虎(义先)
Subject: RE: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

“
The other case to handle is the binding SID. The binding SID needs to reflect the ELC of the associated LSP.
The SRMS case is also interesting if we want to enable entropy when doing SR-LDP interworking.
“

Does this not assume that a LSP is strictly defined hop-by-hop? If LSP is a loose LSP, then it is very well possible that
depending on the state of network, sometimes the LSP is ELC capable and another time it is not capable or the ERLD changes.

I’m not convinced or have clear understanding on the value of ELC/ERLD for a particular LSP without making some strict prescriptions on LSP.

To me, having ERLD (and maybe of less importance) ELC has value only for a controller to make more educated decisions upon packet flows.
Having such information per LSP seems complex and of lesser value as per node in the network.

G/

From: Lsr <lsr-bounces@ietf.org<mailto:lsr-bounces@ietf.org>> On Behalf Of stephane.litkowski@orange.com<mailto:stephane.litkowski@orange.com>
Sent: Monday, August 6, 2018 10:26
To: Les Ginsberg (ginsberg) <ginsberg@cisco.com<mailto:ginsberg@cisco.com>>; DECRAENE Bruno IMT/OLN <bruno.decraene@orange.com<mailto:bruno.decraene@orange.com>>
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>; 徐小虎(义先) <xiaohu.xxh@alibaba-inc.com<mailto:xiaohu.xxh@alibaba-inc.com>>
Subject: Re: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Hi,

Hi Les, Bruno,


[Bruno] I also raised the case of redistribution of IP prefix/SID between IGP domains. Possibly one using OSPF and one using IS-IS. This case needs to be also covered.

[Les:] If a prefix is leaked between protocols then you lose the identification of the source. Which means you have other considerations which are not met e.g., what is the value for ERLD (which also is clearly not a per prefix value). To know this you need to know the source and have the Node Capabilities.

[SLI] I agree with Les that we are losing the ERLD information. But I think it is worth allowing the propagation of the ELC even if we are losing the ERLD info. We make this propagation optional. Moreover it will be still optional for the ingress node to use it.
I agree that we must not leak the node infos.

The other case to handle is the binding SID. The binding SID needs to reflect the ELC of the associated LSP.
The SRMS case is also interesting if we want to enable entropy when doing SR-LDP interworking.


From: Les Ginsberg (ginsberg) [mailto:ginsberg@cisco.com]
Sent: Friday, August 03, 2018 23:27
To: DECRAENE Bruno IMT/OLN
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>; 徐小虎(义先)
Subject: RE: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Bruno –

Inline.

From: bruno.decraene@orange.com<mailto:bruno.decraene@orange.com> <bruno.decraene@orange.com<mailto:bruno.decraene@orange.com>>
Sent: Friday, August 03, 2018 10:01 AM
To: Les Ginsberg (ginsberg) <ginsberg@cisco.com<mailto:ginsberg@cisco.com>>
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>; 徐小虎(义先) <xiaohu.xxh@alibaba-inc.com<mailto:xiaohu.xxh@alibaba-inc.com>>
Subject: RE: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Les,

Please see inline [Bruno]

From: Les Ginsberg (ginsberg) [mailto:ginsberg@cisco.com]
Sent: Friday, August 03, 2018 6:32 PM
To: DECRAENE Bruno IMT/OLN; 徐小虎(义先)
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>
Subject: RE: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Bruno –

I appreciate why you suggest per-prefix signaling for ELC, but I would prefer that we not employ that model.
[Bruno] Thanks for the feedback. That’s part of the discussion that I was calling for.

ELC is clearly a node capability

[Bruno] It’s a node capability of the egress of the LSP. Hence it could also be seen as a property of the LSP ;-) RFC 6790 says  “one may choose to associate ELs with MPLS tunnels (LSPs) or
   with MPLS applications […] We take the former approach”


– signaling it in per node scope is therefore most appropriate. And it aligns with the SR model where we do not need to depend on hop-by-hop signaling as in the LDP case.
[Bruno] I would argue that advertising ELC per prefix is not a hop-by-hop signaling as in the LDP case. That’s one single advertisement per area/level.

[Les:] I probably expressed this poorly.
As LDP does hop-by-hop signaling the “natural” way to signal this in LDP is in the neighbor label exchanges.
However, SR signaling is scoped by the area/domain and therefore we have the ability to advertise capability in a way that matches how the support is actually enabled. There is no basis for a node to advertise ELC support for local Prefix A but not for Local Prefix B. So advertising this per prefix is redundant (though I agree is still possible).

As regards the interarea issues you raise:

Both Router Capability TLV (IS-IS) and Router Information LSA (OSPF) support domain-wide flooding scope. This is not a new capability
[Bruno] Agreed.

– though I do agree with you that the ELC drafts should explicitly mention the flooding scope requirement.
[Bruno] May be the scalability impact may need to be discussed. I would assume an impact on the size of the LSDB (but a priori no impact on the churn). This is also a new behavior which may surprise operational teams.
[Les:] The significant issue as regards scale is the leaking of the prefixes – not the leaking of per-Node information. So I am not greatly concerned about leaking Node Capability info.

[Bruno] I also raised the case of redistribution of IP prefix/SID between IGP domains. Possibly one using OSPF and one using IS-IS. This case needs to be also covered.

[Les:] If a prefix is leaked between protocols then you lose the identification of the source. Which means you have other considerations which are not met e.g., what is the value for ERLD (which also is clearly not a per prefix value). To know this you need to know the source and have the Node Capabilities.

    Les

Thanks,
Regards,
--Bruno

As regards identifying the source of a prefix advertisement domain-wide, IS-IS has a complete solution for this as defined in RFC 7794.
OSPF is lacking support for advertising the source Router-ID, but this can be easily remedied by defining an extension using Extended Prefix LSA (as has been mentioned by Peter in another thread). And this functionality is needed for other reasons e.g., to know when PHP should/should not be done. So it is probably past time when this should be defined.

So I think it is better if we use the per-node ELC scope proposed in the ELC drafts.

As an aside, I would prefer that we utilize the existing TE Node Capability Descriptor registry defined in RFC 5073 rather than invent a new codepoint/registry (the proposed Non-IGP Functional  Capabilities registry) – but that is a minor point.

   Les


From: Lsr <lsr-bounces@ietf.org<mailto:lsr-bounces@ietf.org>> On Behalf Of bruno.decraene@orange.com<mailto:bruno.decraene@orange.com>
Sent: Friday, August 03, 2018 12:46 AM
To: 徐小虎(义先) <xiaohu.xxh@alibaba-inc.com<mailto:xiaohu.xxh@alibaba-inc.com>>
Cc: lsr@ietf.org<mailto:lsr@ietf.org>; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>
Subject: Re: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Hi  Xiaohu,

Thanks for the reply.
You seem to assume/require that (router) capability advertisement be propagated across IGP areas/domains. If so,
- this is a new requirement for existing multi-area networks that need to be indicated in the draft
- I find this debatable. This point should be explicitly discussed. I’d rather advertise the ELC capability on a per Segment basis. This would also be better aligned with RFC 6790 hence safer if EL extensions are defined. The (Segment Routing) Prefix-SID sub-TLV has 2 flags remaining. This may be a good candidate as this information seems SR specific. Alternatively, RFC7794 may be used.

Thanks
Best regards,
--Bruno

From: Lsr [mailto:lsr-bounces@ietf.org] On Behalf Of ???(??)
Sent: Friday, August 03, 2018 4:13 AM
To: Lsr; draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>
Cc: lsr@ietf.org<mailto:lsr@ietf.org>
Subject: Re: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Hi Bruno,

Thanks for raising this important issue.

In fact, the Routable IP Address TLVs/sub-TLVs as described in (https://tools.ietf.org/html/draft-ietf-ospf-routable-ip-address-02) and (https://tools.ietf.org/html/draft-xu-isis-routable-ip-address-01) respectively were intended to address the problem that you had mentioned (i.e., it is required for OSPF routers in one area to find correlations between routable IP addresses and capabilities of OSPF routers in another area).

The following text is quoted from

"    There are several situations where it is required for OSPF routers in

   one area to find correlations between routable IP addresses and

   capabilities of OSPF routers in another area.  One example is the

   Entropy Label Capability (ELC) advertisement [I-D.xu-ospf-mpls-elc<https://tools.ietf.org/html/draft-ietf-ospf-routable-ip-address-02#ref-I-D.xu-ospf-mpls-elc>]

   across the OSPF domain.  In this example, assume the ELC TLV

   originated by a router in one area is propagated to another area.

   Those routers in the latter area need to find routable IP addresses

   of the router originating that ELC TLV before inserting the Entropy

   Label (EL) for packets going to the Label Switch Path (LSP) tunnel

   towards one of the above routable IP addresses..."

Later, such correlation requirement in the ISIS domain was addressed by introducing the source IPv4/IPv6 router ID sub-TLVs into the Extended Reachability TLVs (see https://tools.ietf.org/html/rfc7794). I forget whether the same extension to OSPF as RFC7794 has been done.

Best regards,
Xiaohu
------------------------------------------------------------------
From:bruno.decraene <bruno.decraene@orange.com<mailto:bruno.decraene@orange.com>>
Send Time:2018年8月3日(星期五) 04:50
To:draft-ietf-isis-mpls-elc@ietf.org <draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>>
Cc:lsr@ietf.org <lsr@ietf.org<mailto:lsr@ietf.org>>
Subject:Re: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt

Hi authors,

"4.  Advertising ELC Using IS-IS

   One bit of the Non-IGP Functional Capability Bits (Bit 0 is desired)
   is to be assigned by the IANA for the ELC [RFC6790]."

RFC6790 defines ELC capability on a per FEC/LSP egress basis.
Please defines what you mean exactly with this per node capability. If this is expected to advertise ELC capability in spring networks, it's not crystal clear to me how it works in multi-area/domain network with IP prefix/SID redistribution.
Possibly the ELC flag would need to be advertised on a per prefix basis.

Thanks,
Regards,
--Bruno


 > -----Original Message-----
 > From: Lsr [mailto:lsr-bounces@ietf.org] On Behalf Of bruno.decraene@orange.com<mailto:bruno.decraene@orange.com>
 > Sent: Thursday, August 02, 2018 10:24 PM
 > To: draft-ietf-isis-mpls-elc@ietf.org<mailto:draft-ietf-isis-mpls-elc@ietf.org>
 > Cc: lsr@ietf.org<mailto:lsr@ietf.org>
 > Subject: Re: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt
 >
 > Hi authors,
 >
 > Please find below some minor comments:
 >
 > 1) Abstract:
 > " In addition, this document introduces the Non-IGP Functional
 >    Capabilities Sub-TLV for advertising IS-IS router's actual non-IGP
 >    functional capabilities.  ELC is one of such non-IGP functional
 >    capabilities."
 >
 > It's a matter of opinion but reducing the number of occurrences of " non-IGP functional
 > capabilities" may improve the S/N ration.
 >
 > 2)
 >    The format of the Router Non-IGP Functional Capabilities Sub-TLV is  as follows:
 >
 >         0                   1                   2                   3
 >         0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
 >        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 >        |    Type=TBD1  |    Length=4   |
 >        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 >
 >
 > The sub-TLV is not hard coded/defined with a length of 4, hence this value should not be part of
 > the definition.
 >
 > 3)
 > "Length: Indicates the length of the value portion in octets and  will be a multiple of 4 octets"
 >
 > Possibly :s/will/MUST
 > Please specify the error handling. (e.g. disregards the whole sub-TLV, disregards the last 1 to 3
 > octets, accept the whole sub-TLV...)
 >
 >
 > 4)
 > "One bit of the Non-IGP Functional Capability Bits (Bit 0 is desired)  is to be assigned by the
 > IANA for the ELC [RFC6790]."
 >
 > Since this document defines the new sub-TLV, it can freely do any allocation itself.
 >
 > 5)
 > "The registration procedure is "Expert Review" as defined in   [RFC8126]."
 >
 > You may want to read RFC 8126 https://tools.ietf.org/html/rfc8126#section-4.5
 > Which, In particular, states:
 > " The registry's
 >    definition needs to make clear to registrants what information is
 >    necessary.
 >
 >   [...]
 >
 >    The required documentation and review criteria, giving clear guidance
 >    to the designated expert, should be provided when defining the
 >    registry.  It is particularly important to lay out what should be
 >    considered when performing an evaluation and reasons for rejecting a
 >    request.  It is also a good idea to include, when possible, a sense
 >    of whether many registrations are expected over time, or if the
 >    registry is expected to be updated infrequently or in exceptional
 >    circumstances only. "
 >
 > 6)
 > "This capability, referred to as Entropy  Readable Label Depth (ERLD) as defined in  [I-D.ietf-
 > mpls-spring-entropy-label] "
 >
 > This probably calls for this document to be a normative reference.
 >
 >
 > "   A new MSD-type of the Node MSD b-TLV
 >    [I-D.ietf-isis-segment-routing-msd], called ERLD is defined to
 >    advertise the ERLD of a given router."
 >
 > May be adding the reference to the document defining the ERLD:
 > OLD: advertise the ERLD
 > NEW: advertise the ERLD [I-D.ietf-mpls-spring-entropy-label]
 >
 > 7)
 > "If a router has
 >    multiple line cards, the router MUST NOT announce the ELC [RFC6790]
 >    unless all of its linecards are capable of processing ELs."
 >
 > May be you mean
 > OLD: all of its linecards
 > OLD: all of the linecards of the links advertised as IS-IS adjacencies.
 >
 > Regards,
 > --Bruno
 >
 >  > -----Original Message-----
 >  > From: Lsr [mailto:lsr-bounces@ietf.org] On Behalf Of internet-drafts@ietf.org<mailto:internet-drafts@ietf.org>
 >  > Sent: Tuesday, July 31, 2018 4:07 PM
 >  > To: i-d-announce@ietf.org<mailto:i-d-announce@ietf.org>
 >  > Cc: lsr@ietf.org<mailto:lsr@ietf.org>
 >  > Subject: [Lsr] I-D Action: draft-ietf-isis-mpls-elc-05.txt
 >  >
 >  >
 >  > A New Internet-Draft is available from the on-line Internet-Drafts directories.
 >  > This draft is a work item of the Link State Routing WG of the IETF.
 >  >
 >  >         Title           : Signaling Entropy Label Capability and Entropy Readable Label Depth Using
 >  > IS-IS
 >  >         Authors         : Xiaohu Xu
 >  >                           Sriganesh Kini
 >  >                           Siva Sivabalan
 >  >                           Clarence Filsfils
 >  >                           Stephane Litkowski
 >  >  Filename        : draft-ietf-isis-mpls-elc-05.txt
 >  >  Pages           : 7
 >  >  Date            : 2018-07-29
 >  >
 >  > Abstract:
 >  >    Multiprotocol Label Switching (MPLS) has defined a mechanism to load
 >  >    balance traffic flows using Entropy Labels (EL).  An ingress Label
 >  >    Switching Router (LSR) cannot insert ELs for packets going into a
 >  >    given tunnel unless an egress LSR has indicated via signaling that it
 >  >    has the capability of processing ELs, referred to as Entropy Label
 >  >    Capability (ELC), on that tunnel.  In addition, it would be useful
 >  >    for ingress LSRs to know each LSR's capability of reading the maximum
 >  >    label stack depth and performing EL-based load-balancing, referred to
 >  >    as Entropy Readable Label Depth (ERLD), in the cases where stacked
 >  >    LSPs are used for whatever reasons.  This document defines mechanisms
 >  >    to signal these two capabilities using IS-IS.  These mechanisms are
 >  >    useful when the label advertisement is also done via IS-IS.  In
 >  >    addition, this document introduces the Non-IGP Functional
 >  >    Capabilities Sub-TLV for advertising IS-IS router's actual non-IGP
 >  >    functional capabilities.  ELC is one of such non-IGP functional
 >  >    capabilities.
 >  >
 >  >
 >  > The IETF datatracker status page for this draft is:
 >  > https://datatracker.ietf.org/doc/draft-ietf-isis-mpls-elc/
 >  >
 >  > There are also htmlized versions available at:
 >  > https://tools.ietf.org/html/draft-ietf-isis-mpls-elc-05
 >  > https://datatracker.ietf.org/doc/html/draft-ietf-isis-mpls-elc-05
 >  >
 >  > A diff from the previous version is available at:
 >  > https://www.ietf.org/rfcdiff?url2=draft-ietf-isis-mpls-elc-05
 >  >
 >  >
 >  > Please note that it may take a couple of minutes from the time of submission
 >  > until the htmlized version and diff are available at tools.ietf.org.
 >  >
 >  > Internet-Drafts are also available by anonymous FTP at:
 >  > ftp://ftp.ietf.org/internet-drafts/
 >  >
 >  > _______________________________________________
 >  > Lsr mailing list
 >  > Lsr@ietf.org<mailto:Lsr@ietf.org>
 >  > https://www.ietf.org/mailman/listinfo/lsr
 >
 > __________________________________________________________________________
 > _______________________________________________
 >
 > Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou
 > privilegiees et ne doivent donc
 > pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par
 > erreur, veuillez le signaler
 > a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant
 > susceptibles d'alteration,
 > Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.
 >
 > This message and its attachments may contain confidential or privileged information that may
 > be protected by law;
 > they should not be distributed, used or copied without authorisation.
 > If you have received this email in error, please notify the sender and delete this message and
 > its attachments.
 > As emails may be altered, Orange is not liable for messages that have been modified, changed
 > or falsified.
 > Thank you.
 >
 > _______________________________________________
 > Lsr mailing list
 > Lsr@ietf.org<mailto:Lsr@ietf.org>
 > https://www.ietf.org/mailman/listinfo/lsr

_________________________________________________________________________________________________________________________

Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.

This message and its attachments may contain confidential or privileged information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
Thank you.

_______________________________________________
Lsr mailing list
Lsr@ietf.org<mailto:Lsr@ietf.org>
https://www.ietf.org/mailman/listinfo/lsr


_________________________________________________________________________________________________________________________



Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc

pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler

a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,

Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.



This message and its attachments may contain confidential or privileged information that may be protected by law;

they should not be distributed, used or copied without authorisation.

If you have received this email in error, please notify the sender and delete this message and its attachments.

As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.

Thank you.

_________________________________________________________________________________________________________________________



Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc

pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler

a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,

Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.



This message and its attachments may contain confidential or privileged information that may be protected by law;

they should not be distributed, used or copied without authorisation.

If you have received this email in error, please notify the sender and delete this message and its attachments.

As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.

Thank you.

_________________________________________________________________________________________________________________________



Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc

pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler

a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,

Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.



This message and its attachments may contain confidential or privileged information that may be protected by law;

they should not be distributed, used or copied without authorisation.

If you have received this email in error, please notify the sender and delete this message and its attachments.

As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.

Thank you.

_________________________________________________________________________________________________________________________



Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc

pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler

a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,

Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.



This message and its attachments may contain confidential or privileged information that may be protected by law;

they should not be distributed, used or copied without authorisation.

If you have received this email in error, please notify the sender and delete this message and its attachments.

As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.

Thank you.

_________________________________________________________________________________________________________________________



Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc

pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler

a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,

Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.



This message and its attachments may contain confidential or privileged information that may be protected by law;

they should not be distributed, used or copied without authorisation.

If you have received this email in error, please notify the sender and delete this message and its attachments.

As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.

Thank you.