Re: [mpls] I-D Action: draft-zheng-mpls-lsp-ping-yang-cfg-06.txt

[Greg Mirsky <[e-mail]>]

Hi Faisal,
apologies for the delay to respond to your thorough review of the MPLS LSP
ping data model and great comments. Please find my answers and notes
in-line and tagged GIM>>.
I'm looking forward to be working with you as co-author on this draft. Hope
you'll agree.

Regards,
Greg

On Tue, Nov 14, 2017 at 2:27 PM, Faisal Iqbal (faiqbal) <[e-mail]>
wrote:

> Dear Authors,
>
> Thank you for initiating the work on Yang Model for LSP Ping and sharing
> your draft. I have gone through the document and find it very useful and
> important work item. Please see below for some comments on the document.
> I’m still a beginner in Yang modeling so I apologize if I’m missing
> something obvious or well-understood.
>
>
>
> Sec 3.1 – I think the document aims to specify various existing MPLS OAM
> targets (including P2MP) through a combination of target-fec-type and
> target-fec structure. Is my understanding correct?
>
GIM>> Yes, you're absolutely correct.

>
>
> Sec 3.1 – I did not find a way to specify some important parameters in the
> container control-info that operators can use to control echo request path
> or other parameters. Specifically:
>
> i.                 Operator should be allowed to specify RSVP interface
> by tunnel name as well. For some operators, name may be more meaningful
> than tunnel number.
>
GIM>> I think we may use tunnel-ref and tunnel-p2mp-ref defined
in draft-ietf-teas-yang-te.
GIM>> And extending your suggestion, should it be for other FECs, for
example, static?

> ii.                A field to provide a destination address in 127/8
> range.
>
> iii.               An option to specify a range of destination addresses
> to test e.g. give destination start address 127.0.0.1 and end address
> 127.0.0.8, system will initiate 8 echo requests correspondingly.
>
GIM>> I like the idea to provide optional controls over entropy of LSP
Ping. These may be useful not only for LSP Ping but BFD over MPLS LSP.

> iv.               Given that TTL is a field in the container
> control-info, I feel that operator should be given an option to specify
> whether or not to use Downstream Detailed Mapping (DDMAP) TLV, or even the
> deprecated DSMAP TLV.
>
> v.                Operator should also be allowed to specify DDMAP TLV
> sub-TLV fields such as the presence, type, and size of multipath TLV.
>
GIM>> Agree. And what the defaults could be?

>
>
> Sec 3.3 – I had following questions about the result-info structure.
>
> i.                 If the echo reply contains a TLV (e.g. DDMAP TLV for
> request to a transit node), how will we communicate that information to the
> user? Operator may use LSP Ping to identify downstream path(s) from a
> particular downstream node.
>
> ii.                What does probe-index refer to?
>
GIM>> Index of an LSP ping within a test session.

>
>
> Sec 6 – For target-fec-type, is there a way for operator to specify a
> particular control plane or FEC (as defined in RFC-8029 e.g. LDP, Generic
> FEC etc.) to be used for echo request?
>
GIM>> It must be extended, agreed. I think it is r/w. Not?


> Does this document only tackle LSP Ping operation? Would other echo
> request operations such as LSP Traceroute for LSP path discovery be
> documented separately?
>
GIM>> Great observation! Traceroute should be supported. Path discovery?

>
>
Some minor editorial comments below.
>
> -“Model presented in [RFC4560];” instead of “Model presented in[RFC4560]
> ;” in Section 1.2
>
> -Text in Section 3 should read “…and result information for multiple
> instances of LSP-Ping test.” Instead of “…and result information for multi
> instances of LSP-Ping test.”
>
> - "IETF Multiprotocol” instead of “"IETF Multiprotocl” in Section 6.
>
> -Description for enum success should read “The test probe is successful”
> instead “The test probe is successed”
>
> -Description for leaf sum-of-squares in Section 6 should read “replies”
> instead of “replys”
>
GIM>> Thank you, Faisal. Will include in the next update.


>
>
> Regards,
>
> Faisal Iqbal
>