RE: [Fwd: [PWE3] Requesting last call on Wildcard PW type]

["Hamid Ould-Brahim" <hbrahim at nortel.com>]

As a general comment, the draft seems to imply that
the PW wildcard type is only applicable in the case
of single-sided signaling model (I think maybe this
is due to the fact that the draft was written with a
specific application in mind). In my view PW wildcard type
can be used in any provisioning and signaling model and
in any PW deployment (single, multi) scenario it doesn't
have to be restricted to only single-sided signaling 
models.

Since the draft is short I included my other comments
within the text (below).

        "Target Choice of Pseudowire Type"

Hamid> Why not just call the draft "PW Wildcard Type". That's
       what the draft is all about.

"Abstract

   The Generalized PWid FEC permits a procedure know as single-sided
   signaling."

Hamid> Single-sided signaling term is no longer used in l2vpn.
       I think you maybe referring here to 'single-sided *provisioning*
       with discovery' model. This is the only model left in
       draft-ietf-l2vpn-signaling-07 (besides the double-sided
       provisioning model).

   "In this procedure, one end of the pseudowire always
   initiates the pseudowire setup and the target of that label mapping
   message only signals in response. For certain applications of
   pseudowires it is advantages to configure the pseudowire type (PW
   type) at the target of the initial label mapping message."

Hamid> The above text seems indicating that the current model implies
       that for most cases the configuration of pw type is done at
       initiating PE and most applications are using that.
       I suggest either add "only" before "at the target of...",
       and replace "advantages" with "necessary" or just replace
       the above paragraph completely with:

      "For certain applications of pseudowires, the pseudowire type
     (PW type) needs to be determined by the target of the
     initial label mapping message."

...

1. Introduction

   Pseudowire signaling is defined in "Pseudowire Setup and Maintenance
   using the Label Distribution Protocol" [CONTROL].  The Generalized
   PWid FEC permits a procedure know as single-sided signaling as docu-
   mented in "Provisioning, Autodiscovery, and Signaling in L2VPNs"
   [L2VPNSIG]. 

Hamid> See comment above. draft-ietf-l2vpn-signaling-07 defines only
       the single-sided provisioning with discovery model. I believe
       that the PW widlcard type can be used in any provisioning model,
       including the double-sided provisioning model if one wants to use
       it this way. The only difference it just for double-sided
       provisioning, both PEs know sufficient information to start the
       signaling but one of the PE does not know the PW type or prefer
       to let the target PE to decide on PW type to use.

....


" 3. Procedures

   The initiating PE indicates a PW type of "wildcard" to indicated to
   the target PE that it should determine the type for this pseudowire."

Hamid> Replace "indicated" with "to indicate".


   "When the targeted PE receives the wildcard PW type, it follows the
   normal procedures in checking the AGI and TAII values.  If a label
   mapping message has already been issued, it MUST respond to this mes-
   sage with a Label_Release message."

Hamid> It looks this is not necessary. If a PE has already initiated
       a label mapping and is okay with responding to the PW wildcard 
       type, it may choose not to release the label it already sent.

       On the other hand, if a target PE has already sent the
       label mapping message it means that it knows sufficient 
       information about the initiating PE (so this cannot be a 
       single-sided provisioning model but another model, this is 
       another reason why this draft applies to any provisioning model 
       not just single-sided signaling...).

 "If PE2 cannot map the TAI to one of its Forwarders, then PE2 sends a
  Label Release message to PE1, with a Status Code of "Unas-
  signed/Unrecognized TAI", and the processing of the Label Mapping
  message is complete."

Hamid> The above text (besides that it introduces a new term "PE2" for
       the first time) is not in fact needed since it is already covered
       in the base control protocol spec and this is part of *normal*
       procedure as part of handling the TAI (unrelated to PW wildcard
       procedure). So I suggest removing it.


"4. Security Considerations

   This draft has little impact the security aspects of [CONTROL]."

Hamid> add "on" before 'the'.


Hamid.

> -----Original Message-----
> From: Stewart Bryant [mailto:stbryant at cisco.com]
> Sent: Thursday, March 23, 2006 6:16 PM
> To: pwe3
> Subject: [Fwd: [PWE3] Requesting last call on Wildcard PW type]
>
>
> This is the start of last call for
>
> draft-ietf-pwe3-wildcard-pw-type-01.txt
>
> Last call will end 10th April 2006
>
> - Stewart
>
>
>
>
> _______________________________________________
> pwe3 mailing list
> pwe3 at ietf.org
> https://www1.ietf.org/mailman/listinfo/pwe3
>
> 


_______________________________________________
pwe3 mailing list
pwe3 at ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3