Re: IP Layer 2 Transport over L2TPv3

Hi Carlos,

Carlos Pignataro wrote:
> Hi Prasad,
> 
> Circa 9/23/2005 8:00 AM, Prasad Yadati said the following:
> 
>>Hi Carlos,
>>
>>Please see in-line.
>>
>>Carlos Pignataro wrote:
>>
>>
>>>Hi Prasad,
>>>
>>>Please see inline.
>>>
>>>Circa 9/20/2005 7:01 AM, Prasad Yadati said the following:
>>>
>>>
>>>>Carlos,
>>>>
>>>>Please see in-line.
>>>>
>>>>Carlos Pignataro wrote:
>>>>
>>>>
>>>>
>>>>>Prasad,
>>>>>
>>>>>Thanks. Please see a couple of comments inline.
>>>>>
>>>>>List,
>>>>>
>>>>>Any further comments on the request to accept this draft as a working
>>>>>group document?
>>>>>
>>>>>
>>>>>Circa 9/17/2005 5:50 AM, Prasad Yadati said the following:
>>>>>
>>>>>
>>>>>
>>>>>>Good one Carlos.
>>>>>>How about just specifying it as "IP transport over L2TPv3".
>>>>>>
>>>>>
>>>>>
>>>>>Thanks.
>>>>>
>>>>>The "IP Layer 2" was added there for 2 reasons:
>>>>>1. To somehow differentiate raw-IP from framed-IP being transported
>>>>>over
>>>>>  L2TPv3 while trying to reference the NSP function that interfaces the
>>>>>  L2 AC with the IP PW Termination.
>>>>>2. Symmetry/parity name with the already defined PW Type for MPLS PWE3.
>>>>>
>>>>
>>>>But as per this draft, no "IP layer 2" information is transported.
>>>
>>>
>>>
>>>Yes, only IP datagrams are transported, although there is a Layer 2
>>>attachment circuit, and hence "IP Layer 2". The NSP function "extracts
>>
>>
>>Attachment circuit layer 2 is already terminated. Isn't it ? L2TPv3 in
>>this case does not aware of attachment circuit's layer 2. In fact it
>>does not matter what kind of layer 2 IP was carried at the attachment
>>circuit.
>>I am sorry Corlos, I am still not clear.
> 
> 
> Although the "IP Layer 2" was intended to make the NSP function
> explicit, I understand (and agree on second thoughts) the argument that
> can be confusing with what's actually being transported.
> 
> How about this to describe the main aspects of the draft:
> "Signaling and Encapsulation for the Transport of IP over L2TPv3"

Looks good to me, Carlos.

> 
> There are also two further items in the draft to possibly update:
> 1. "IP Layer 2 interface": I'd leave this one as is.
Carlos, why not we just use AC (attachment circuit) instead, so that it will be 
consistent with rest of the L2TPv3 RFCs.

> 2. PW Type/PW name: If changing the title, I'd also update:
> 	IPL2PW IP Layer 2 Transport Pseudowires
> 	IP Layer 2 Transport Pseudowire Type.
>    to:
> 	IPPW   IP Transport Pseudowires
> 	IP Transport Pseudowire Type.

Looks good.

Thanks a lot for taking care of the comments Carlos.

Regards
Prasad

> 
> Would this aid in clarity?
> 
> Thanks for your comments !
> 
> --Carlos.
> 
> 
>>>IP datagrams from the Layer 2 frames" and that's the motivation for the
>>>name, make a distinction from framed-IP. See below as well.
>>>
>>>
>>>
>>>>Rather we are trying to prevent the same. For me, "IP layer 2" seems to
>>>>be mis-leading.
>>>>
>>>>
>>>>
>>>>
>>>>>"IP transport" may possibly be too generic for the procedures defined.
>>>>
>>>>
>>>>
>>>>Corlos, for me, "IP transport" seems more appropriate than current. In
>>>>this case as far as L2TPv3 is concerned, it is transporting IP rather
>>>>than layer 2 frame.
>>>
>>>
>>>
>>>IMHO, "IP transport" is too generic, and does not fully convey what's
>>>happening in the interface between the physical layer 2 interface and
>>>the pseudowire termination. Other PW Types transport IP (framed-IP) as
>>>well. I thought about "raw ip" and "native ip" but opted against them
>>>for a similar reason. Any other suggestions?
>>
>>
>>How about "Abstracted/Extracted IP over L2TPv3" ?
>>If this draft can be, really generic, I mean if it  can be used even for
>>non-pw cases where L2TPv3 carries generic IP datagrams, we may have "IP
>>transport over L2TPv3" ?
>>
>>Kind Regards
>>Prasad
>>
>>
>>
>>
>>>
>>>>>>Wondering whether we can have a generic "Layer 3 transport over
>>>>>>L2TPv3"
>>>>>>and specify transport of all layer 3 protocols (IP, IPX,....) in it
>>>>>>including IP.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>Do you mean multiple L3 protocols at the same time over the session?
>>>>>I'd
>>>>>think we can use existing L2 transports to accomplish that. OTOH, IP as
>>>>>L3 stands out in the applications of an "IP L2" session.
>>>>
>>>>
>>>>Sorry I was not clear. I meant to combine specification of other layer 3
>>>>data packets over L2TPv3 in one draft. May be that is not needed. We may
>>>>have separate similar such of these drafts for other layer 3 protocols.
>>>>Please ignore this comment.
>>>
>>>
>>>
>>>OK. Thanks,
>>>
>>>--Carlos.
>>>
>>>
>>>
>>>>Thanks
>>>>Prasad
>>>>
>>>>
>>>>
>>>>>Thanks !
>>>>>
>>>>>--Carlos.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>>Regards
>>>>>>Prasad
>>>>>>
>>>>>>Carlos Pignataro wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>>Hi All,
>>>>>>>
>>>>>>>We submitted the following `IP Layer 2 Transport over L2TPv3' draft:
>>>>>>>http://www1.ietf.org/mail-archive/web/i-d-announce/current/msg06623.html
>>>>>>>
>>>>>>>
>>>>>>>that details the tunneling of IP datagrams directly over an L2TPv3
>>>>>>>session, and defines the `CE IP Address AVP' as well.
>>>>>>>
>>>>>>>We would like it to become an l2tpext working group document, and
>>>>>>>seek
>>>>>>>review, feedback and comments from the WG.
>>>>>>>So we ask here for the draft to be accepted as work item for the
>>>>>>>group.
>>>>>>>Your feedback will be most appreciated.
>>>>>>>
>>>>>>>Thanks,
>>>>>>>
>>>>>>
>