Re: [mpls-tp] [mpls] R: Use of term "interface" in MPLS-TP Identifiers and OAM Framework drafts

["Adrian Farrel" <[e-mail]>]

Hi Sasha,

I don't think the fact that there was a label above in the stack is important.
What is important is whether the packet in hand has been received over an
interface (including a virtual interface) or not. An implementation and a
deployment should be free to choose whether an LSP tunnel is implemented as a
virtual interface or not. If it is not, the label in hand will be associated to
the interface over which the packet (with the previous label) was received. So,
in fact, no change in processing or semantics.

I think that some implementations do not support LSPs being presented as virtual
interfaces. This is fine and acceptable and within the scope of the
specifications. IIRC the MIB module allows an LSP tunnel to be configured either
way.

Adrian

> -----Original Message-----
> From: Alexander Vainshtein [mailto:[e-mail]]
> Sent: 10 December 2010 17:35
> To: [e-mail]; [e-mail]
> Cc: [e-mail]; [e-mail]
> Subject: RE: [mpls-tp] [mpls] R: Use of term "interface" in MPLS-TP
Identifiers
> and OAM Framework drafts
> 
> Adrian,
> I agree with you that "interface" has a long history in the IETF and is not
always
> used consistently.
> 
> My concern is about the term "per-interface label space" as it appears in 3031
and
> the consequent IETF documents.
> 
> Does your email imply that, from your point of view,  per-interface label
space
> includes label space per terminated tunnel?
> And that, as a consequence, processing of an incoming downstream-allocated
> label may depend on the label (or labels) that have been present in the label
> stack above it and have been popped?
> 
> I'd like to quote from Section 3.9 of RFC 3031 just to make sure we are all on
the
> same page:
> 
>    Although, as we shall see, MPLS supports a hierarchy, the processing
>    of a labeled packet is completely independent of the level of
>    hierarchy.  The processing is always based on the top label, without
>    regard for the possibility that some number of other labels may have
>    been "above it" in the past, or that some number of other labels may
>    be below it at present.
> 
> Regards,
>      Sasha
> 
> ________________________________________
> From: [e-mail] [[e-mail]] On Behalf Of
> Adrian Farrel [[e-mail]]
> Sent: Friday, December 10, 2010 4:20 PM
> To: [e-mail]; [e-mail]
> Subject: Re: [mpls-tp] [mpls] R: Use of term "interface" in MPLS-TP
Identifiers
> and OAM Framework drafts
> 
> I agree with Italo (try not to gasp so loudly :-)
> 
> The I-Ds do appear to use the term consistently, and bringing this important
> term to the front so it is really noticed seems important. Also adding it to
the
> Rosetta Stone I-D in a consistent way.
> 
> A *separate* issue is determining whether we are happy with the definition
> being
> used. "interface" is a wriggly term with a long history in the IETF. Those
> wanting to pick up some background could do worse than read IF-MIB (RFC2863)
> and
> then look at the information about the use of "interfaces" in Section 8 of RFC
> 3813.
> 
> In RFC 4397 we worked with experts in ITU-T SG15 and got a bit anal about the
> definitions. Section 3.6 discusses "link interfaces", and it would be nice to
> stay consistent.
> 
> The question of "layers" is delicate, and should probably include
"sub-layers".
> But note that the interface is not the layer (which seems to be what Sasha is
> objecting to), it is the "gateway" to the layer. One might think of it in
terms
> of adaptation functions, and indeed, I think this *is* consistent with the
> current MPLS data plane architecture. Consider, for example, that an LSP end
> may
> be represented as a virtual interface (i.e. the LSP is a virtual link).
> 
> Cheers,
> Adrian
> 
> > -----Original Message-----
> > From: [e-mail] [mailto:[e-mail]] On Behalf Of
> > BUSI, ITALO (ITALO)
> > Sent: 10 December 2010 12:07
> > To: Alexander Vainshtein; [e-mail]; [e-mail]
> > Subject: [mpls] R: [mpls-tp] Use of term "interface" in MPLS-TP Identifiers
> and
> > OAM Framework drafts
> >
> > Sasha,
> >
> > The OAM Framework and Identifier drafts are already using the term
> "interface"
> > in a consistent way among each other.
> >
> > According to the Identifiers draft the attachment point to a server MPLS-TP
> > Tunnel is an interface.
> >
> > The change #2 I am proposing is just aimed at clarifying that the server
> MPLS-TP
> > tunnel is a server sub-layer and not a server layer as per past discussion.
> >
> > Italo
> >
> > > -----Messaggio originale-----
> > > Da: Alexander Vainshtein [mailto:[e-mail]]
> > > Inviato: venerdì 10 dicembre 2010 10.48
> > > A: BUSI, ITALO (ITALO); [e-mail]; [e-mail]
> > > Oggetto: RE: [mpls-tp] Use of term "interface" in MPLS-TP Identifiers and
> > > OAM Framework drafts
> > >
> > > Italo,
> > > Lots of thanks for a prompt response.
> > > I agree that different MPLS-TP documents treat the term "interface"
> > > differently, and this adds to the overall havoc.
> > >
> > > I strongly object to treating sub-layers as interfaces in MPLS-TP because
> > > IMO  this would break the MPLS data plane in an irreparable manner.
> > >
> > > As a consequence I suggest we keep the definition used in the MPLS-TP
> > > Identifiers draft and rework the OAM framework document accordingly.
> > >
> > > Regards,
> > >      Sasha
> > >
> > >
> > >
> > > From: [e-mail] [[e-mail]] On Behalf Of
> > > BUSI, ITALO (ITALO) [[e-mail]]
> > > Sent: Friday, December 10, 2010 11:00 AM
> > > To: [e-mail]; [e-mail]
> > > Subject: [mpls-tp] Use of term "interface" in MPLS-TP Identifiers and OAM
> > > Framework drafts
> > >
> > >
> > > I see a lot of comments related to the term "interface" used within the
> > > OAM Framework draft in the context of the per-interface MIP definition and
> > > NO comments on the same term used within the Identifiers draft.
> > >
> > > I would like to clarify that the same term "interface" actually represents
> > > the same concept in both the MPLS-TP Identifiers and MPLS-TP OAM
> > Framework
> > > drafts.
> > >
> > > I would therefore propose that the following definition is added to both
> > > drafts:
> > >
> > >  "
> > > Interface: An interface is the attachment point to a server (sub-)layer
> > > e.g., MPLS-TP section or MPLS-TP tunnel.
> > > "
> > >
> > > This definition is taken from the text in section 4 of the Identifiers
> > > draft with the following proposed changes:
> > >
> > > 1) c/Access Point (AP)/attachment point/ as proposed by ITU-T
> > >
> > > 2) c/layer/(sub-)layer/ to align the definition with the outcome of past
> > > discussion: label stacking is a form of sub-layering with the MPLS and
> > > MPLS-TP layer network.
> > >
> > > I have a strong opinion that both the MPLS-TP Identifiers and the MPLS-TP
> > > OAM Framework must use the same term (and definition) to identify the
> same
> > > entity.
> > >
> > > I have a preference to keep the term "interface" on both drafts to speed
> > > up the work.
> > >
> > > Italo
> > _______________________________________________
> > mpls mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/mpls
> 
> _______________________________________________
> mpls-tp mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/mpls-tp=