Re: [mpls-tp] MPLS-TP identifiers clarification request

[Huub van Helvoort <[e-mail]>]

Hello Vencat,

You wrote:

>   Hi MPLS-TP group,
>   Please find below my comments in-line tagged VM>> on MPLS-TP identifiers.

My response [hvh]

> As per George's response:
>  >> The draft defines the Global_Tunnel_ID as:
>  >> Src-Global_ID::Src-Node_ID::Src-Tunnel_Num::
> Dst-Global_ID::Dst-Node_ID::Dst-Tunnel_Num
>  >>This seems to define a bidirectional tunnel. Is that the intent? If
> so please clarify. If not then why
>  >>Dst-Global_ID::Dst-Node_ID::Dst-Tunnel_Num is needed?
>
> *>GS: The scope is bidirectional.  But I suppose it can be applied to
> unidirectional as well.  I still don’t understand the requirement for
> unidirectional – seems to be that bidirectional with 0 data-bandwidth in
> the reverse direction would do better as you would have a return path
> for OAM and DCC etc.
> *
> VM>> Don't we need uni-directional tunnel for MPLS-TP if the scope is
> bi-directional?

[hvh] consider uni-directional as a special, or initial case of
unidirectional p-2-mp. So IMHO the answer is: yes, we need uni-dir.

> If the answer yes, then do we need to remove the below requirements of
> an MPLS-TP from the RFC-5654.
> Section 2.1.  General Requirements
>      7.   MPLS-TP MUST support unidirectional, co-routed bidirectional, and
>          associated bidirectional point-to-point transport paths.
>      8.   MPLS-TP MUST support unidirectional point-to-multipoint transport
>          paths.

[hvh] no need to remove.

> As per Mach Chen's mail dated Tue, 22 Jun 2010 11:47:17 +0800
>  >>2. In your draft, an LSP is identified by the combination of
> Src-Global_ID, Src-Node_ID, Src-Tunnel_Num, Dst->>Global_ID,
> Dst-Node_ID, Dst-Tunnel_Num, LSP_Num, this is fine for unidirectional
> and co-routed bidirectional LSP, but it is not enough for associated
> bidirectional LSP that is combined with two reverse unidirectional LSPs
> and IMHO two LSP_Nums are required.
>
> VM>> IMHO, we need two LSP_Nums for co-routed bidirectional LSP for
> unidirectional protection switching i.e only forward or reverse
> direction LSP protection switching instead of switching both forward and
> reverse direction working LSP into protection LSP.

[hvh] in transport forward and reverse co-routed LSPs are switched
simultaneous.

> So, I think we need a single LSP_Num for unidirectional LSP, two
> LSP_Nums for forward and reverse direction co-routed/associated MPLS-TP
> tunnel.

[hvh] we need a single LSP_Num for unidirectional LSP, one LSP_Nums
for forward and reverse direction co-routed, and two LSP nums for
associated MPLS-TP tunnel

> Can we have two tunnels created independenly and have an association
> with them for both co-routed and associated bi-directional MPLS-TP tunnel?

[hvh] no, it is important for co-routed that exactly the same
path is setup, it will be difficult to achieve when first the
forward path is setup and then the reverse path.

> As per Mukund's mail dated Fri, 28 May 2010 12:06:37 +0530
>  >The "draft-ietf-mpls-tp-identifiers" mention two forms of operator
> identification - Global_ID (for IP operational practice) and
> ICC (compatible with ITU).
>  >All the sections on identifiers further in the draft, mention the
> globally unique format of the identifiers by pre-pending the Global_ID.
> i.e they mention only the IP based identifiers.
>  > Eg; For LSP Identiffication:
>  > "Src-Global_ID::Src-Node_ID::Src-Tunnel_Num::
> Dst-Global_ID::Dst-Node_ID::Dst-Tunnel_Num::LSP_Num"
>  > Shouldnt the draft be mentioning the ICC based identification also
>   (substituting Global_ID by ICC for ICC based LSP Identifiers) or
> mentioning explicitly in the draft as follows:
>  >
> "Src-Operator_ID::Src-Node_ID::Src-Tunnel_Num::Dst-Operator_ID::Dst-Node_ID::Dst-Tunnel_Num::LSP_Num"
>  > Where Operator_ID can be Global_ID or ICC_ID
>  > Or is it expected that for ICC based identification, different set of
> identifiers will be defined similar to the MEG & MEP Identifiers??
>
> VM>> IMO, If we have a transport network where LSP can be operated over
> IP and Non-IP environments, then Global_ID and ICC should be combined
> and called as Operator ID. Can authors of MPLS-TP identifiers draft
> clarify this and make necessary changes in the draft?

[hvh] ICC is Global ID.

Best regards, Huub.


-- 
================================================================
Always remember that you are unique...just like everyone else...