Re: [mpls] New Version Notification for draft-mirsky-spring-bfd-00.txt

Greg,

> On May 10, 2017, at 5:23 PM, Greg Mirsky <gregimirsky@gmail.com> wrote:
> 
> Hi Carlos,
> RFC 7110 defined sub-TLVs by extensively re-using TFS sub-TLVs. Even more, they've referenced explanations of fields to the TFS-defining RFCs. I guess only Flags field was introduced in RFC 7110 with Primary and Secondary bit flag fields being defined.
> 
> As, I've said in the discussion on BFD directed, this is proposal, it make sense to me as the head-end has all the information already. I always welcome technical comments and appreciate well-argumented discussion.

From my perspective, the well argument discussion already happened on this exact text and this proposed approach. They happened already twice on multiple WGLCs in BFD, and that text was removed. Asking again under a different draft filename does not change the arguments already presented.

The proposal is broken.

> What would be the reason not to use the proposed approach but do it TFS-like style?
> 

You can read these reasons on the list archives on the previous discussion. I answered this question already. But, for completeness:

Label values can change. With labels there is no validation possible that what distributed by a given label distribution protocol is what is meant in the data plane. 

More importantly, again, a technical example of why this is broken and backwards:

https://tools.ietf.org/html/draft-mirsky-spring-bfd-00#section-4 says:

   The IANA is requested to assign new sub-TLV type from "Multiprotocol
   Label Switching Architecture (MPLS) Label Switched Paths (LSPs) Ping
   Parameters - TLVs" registry, "Sub-TLVs for TLV Types 1, 16, and 21"
   sub-registry.

     +---------+-------------------------------------+---------------+
     | Value   | Description                         | Reference     |
     +---------+-------------------------------------+---------------+
     | X (TBD) | Segment Routing MPLS Tunnel sub-TLV | This document |
     +---------+-------------------------------------+---------------+

Now, TLV Types 1, 16, and 21 for MPLS LSP Ping are 

         1   Target FEC Stack
      16    Reverse-path Target FEC Stack
      21    Reply Path 

MPLS LSP Ping TLVs 1, 16, and 21 need a sub-TLV with a FEC. Not with a Label value. That is why, TLV 1 is called “Target FEC Stack” (sometimes referred to informally as TFS)

However, you are defining a protocol structure that holds numeric Label values, and you somehow want that to be used in TLV 1 for MPLS LSP Ping…

How do you envision this to work with https://tools.ietf.org/html/rfc8029#section-3.2? 

That is yet another reason why FECs are being defined at:
https://tools.ietf.org/html/draft-ietf-mpls-spring-lsp-ping-02

Hope that helps,

— Carlos.
PS: As I find this repetitive, this is my last email on the subject.

> Regards,
> Greg
> 
> On Wed, May 10, 2017 at 1:01 PM, Carlos Pignataro (cpignata) <cpignata@cisco.com> wrote:
> You are right — sorry about that! “TFS” is not used in any of those RFCs or drafts, although it is used on email discussions about LSP Ping.
> 
> Indeed, TFS for “Target FEC Stack” from Section 3.2 of RFC 8029.
> 
> Thanks,
> 
> — Carlos.
> 
>> On May 10, 2017, at 3:41 PM, Robert Raszuk <robert@raszuk.net> wrote:
>> 
>> 
>> Never mind .. I guess you made it up from "Target FEC Stack" :)
>> 
>> 
>> 
>> On Wed, May 10, 2017 at 9:40 PM, Robert Raszuk <robert@raszuk.net> wrote:
>> Hi Carlos,
>> 
>> Sorry what is "TFS" ? 
>> 
>> RFC 7110 does not even use such abbreviation neither do draft-ietf-mpls-bfd-directed :) Google also seems to be pretty clueless about it. 
>> 
>> Just curious as you keep using this term in each email :) 
>> 
>> Thx,
>> R.
>> 
>> On Wed, May 10, 2017 at 9:24 PM, Carlos Pignataro (cpignata) <cpignata@cisco.com> wrote:
>> Greg,
>> 
>> In the MPLS data plane, FECs are also instantiated through a label stack. But RFC 7110 does not use numeric label values, it uses TFSs. That does not create any additional state. E.g.,: https://www.ietf.org/mail-archive/web/mpls/current/msg16091.html
>> 
>> Thanks,
>> 
>> — Carlos.
>> 
>> 
>> 
>>> On May 9, 2017, at 3:43 PM, Greg Mirsky <gregimirsky@gmail.com> wrote:
>>> 
>>> Hi Carlos,
>>> I probably would characterize anything that starts with Why not as a technical comment but rather as a question.
>>> According to draft-ietf-spring-segment-routing-mpls, "In the MPLS dataplane,the SR header is instantiated through a label stack".
>>> At the same time, one of advantages of SR is that "per-flow state only [maintained] at the ingress node to the SR domain".
>>> Thus, for the case of monitoring unidirectional SR tunnels, I consider that there's no need to create any additional state on the egress node.
>>> Of course, if there were bidirectional SR tunnels, then control of the reverse direction of the BFD session would not require use of the Return Path sub-TLV.
>>> As for LSP-Ping, I just propose that the Segment Routing MPLS Tunnel sub-TLV MAY be used Reply Path TLV defined in RFC 7110. I viewed the proposal as invitation to technical discussion.
>>> 
>>> Regards,
>>> Greg
>>> 
>>> On Tue, May 9, 2017 at 9:07 AM, Carlos Pignataro (cpignata) <cpignata@cisco.com> wrote:
>>> Thank you Greg!
>>> 
>>> Since https://tools.ietf.org/html/draft-mirsky-spring-bfd-00 seems quite similar to the text removed at https://tools.ietf.org/rfcdiff?url2=draft-ietf-mpls-bfd-directed-05.txt, then the complete set of outstanding technical comments that triggered the removal of that text from draft-ietf-mpls-bfd-directed-05.txt might peek your interest :-)
>>> 
>>> One that I recall is: why use label values when every other return-path sub-TLV for BFD and for LSP-Ping, including draft-ietf-mpls-bfd-directed, uses TFSs? 
>>> 
>>> Best,
>>> 
>>> — Carlos.
>>> 
>>>> On May 9, 2017, at 12:00 PM, Greg Mirsky <gregimirsky@gmail.com> wrote:
>>>> 
>>>> Dear Carlos,
>>>> I've decided to re-start the discussion and am interested to hear technical comments to the proposed solution. 
>>>> 
>>>> Regards,
>>>> Greg
>>>> 
>>>> On Tue, May 9, 2017 at 8:51 AM, Carlos Pignataro (cpignata) <cpignata@cisco.com> wrote:
>>>> Dear Greg,
>>>> 
>>>> Cursorily scanning through this, it seems that most concerns raised and comments made about the SR sections of draft-ietf-mpls-bfd-directed-0N (with N < 5) apply to your new draft.
>>>> 
>>>> This is one of those: https://www.ietf.org/mail-archive/web/mpls/current/msg15860.html — the list archive shows a few more. The copy/paste did not address the comments.
>>>> 
>>>> Best,
>>>> 
>>>> — Carlos.
>>>> 
>>>>> On May 8, 2017, at 11:33 PM, Greg Mirsky <gregimirsky@gmail.com> wrote:
>>>>> 
>>>>> Dear All,
>>>>> perhaps this new draft may is of interest to you.
>>>>> Your comments, suggestions are most welcome and greatly appreciated.
>>>>> 
>>>>> Regards,
>>>>> Greg
>>>>> 
>>>>> ---------- Forwarded message ----------
>>>>> From: <internet-drafts@ietf.org>
>>>>> Date: Mon, May 8, 2017 at 8:29 PM
>>>>> Subject: New Version Notification for draft-mirsky-spring-bfd-00.txt
>>>>> To: Gregory Mirsky <gregimirsky@gmail.com>
>>>>> 
>>>>> 
>>>>> 
>>>>> A new version of I-D, draft-mirsky-spring-bfd-00.txt
>>>>> has been successfully submitted by Greg Mirsky and posted to the
>>>>> IETF repository.
>>>>> 
>>>>> Name:           draft-mirsky-spring-bfd
>>>>> Revision:       00
>>>>> Title:          Bidirectional Forwarding Detection (BFD) in Segment Routing Networks Using MPLS Dataplane
>>>>> Document date:  2017-05-08
>>>>> Group:          Individual Submission
>>>>> Pages:          7
>>>>> URL:            https://www.ietf.org/internet-drafts/draft-mirsky-spring-bfd-00.txt
>>>>> Status:         https://datatracker.ietf.org/doc/draft-mirsky-spring-bfd/
>>>>> Htmlized:       https://tools.ietf.org/html/draft-mirsky-spring-bfd-00
>>>>> Htmlized:       https://datatracker.ietf.org/doc/html/draft-mirsky-spring-bfd-00
>>>>> 
>>>>> 
>>>>> Abstract:
>>>>>    Segment Routing architecture leverages the paradigm of source
>>>>>    routing.  It can be realized in the Multiprotocol Label Switching
>>>>>    (MPLS) network without any change to the data plane.  A segment is
>>>>>    encoded as an MPLS label and an ordered list of segments is encoded
>>>>>    as a stack of labels.  Bidirectional Forwarding Detection (BFD) is
>>>>>    expected to monitor any kind of paths between systems.  This document
>>>>>    defines how to use Label Switched Path Ping to bootstrap and control
>>>>>    path in reverse direction of a BFD session on the Segment Routing
>>>>>    network over MPLS dataplane.
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 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.
>>>>> 
>>>>> The IETF Secretariat
>>>>> 
>>>>> 
>>>>> _______________________________________________
>>>>> mpls mailing list
>>>>> mpls@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/mpls
>>>> 
>>>> 
>>> 
>>> 
>> 
>> 
>> 
> 
> 