[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE : [mpls] draft-vasseur-ccamp-te-router-info-00.txt clarificationneeded.
Hi Dillikar,
Sorry for this delayed answer.
Thanks for these useful comments, that will help clarifying this spec.
Please see inline.
Regards,
JL
PS: I'm copying ccamp
>-----Message d'origine-----
>De : mpls-bounces@lists.ietf.org
>[mailto:mpls-bounces@lists.ietf.org] De la part de
>Satyanarayana Dillikar
>Envoyé : mercredi 6 octobre 2004 10:38
>À : mpls@ietf.org; mpls-ops@mplsrc.com
>Objet : [mpls] draft-vasseur-ccamp-te-router-info-00.txt
>clarificationneeded.
>
>
>Hi,
> We have some confusion in understanding the Data
>Plane Capability Flags (B-bit & E-bit) from
>draft-vasseur-ccamp-te-router-info-00.txt
>
>(a) Does E-bit ON implies B-bit ON always ? (assuming
>ON = set and OFF = unset).
Basically bud (transit + Egress) capability requires some branching in the data plane so in general E ON will imply B ON,
but note that these capabilities does not necessarily reflect real hardware capabilities as they
may be activated/deactivated by the operator for various reasons.
We will clarify this point in next revision.
>(b) E-bit = ON & B-bit = OFF, is it a valid
>combination.
Yes see above, there may be cases where the operator want to deactivate branch capability on a node (He does't want that the node act as a branch LSR), even if its data plane is physically branch capable, but he allows the node to act as a bud-LSR (transit + egress).
This gives more operational flexibility.
>(c) Please tell us the E-bit and B-bit status for a
>node which is a destination node but does not have
>branch capability.
If its data plane is not branch capable then it will also probably not be bud capable so
E = 0 and B = 0
In return, if its data plane is branch capable but branch LSR capability has been deactivated by configuration and bud-LSR capability is activated, then
E = 1 and B = 0
>
>
>We are also curious to know
>(1)The idea behind combing two things (egress status &
>transit status) in a single E-bit. rather than making
>use of B-bit(branch) and having E-bit just for egress
>status.
Remind that these capabilities are used for tree computation purpose, and the egress is an entry
of the computation. So, IMHO it does't really make any sense to advertise egress capability only.
>(2) Why the TE Node Capability Descriptor TLV should
>have E-bit & how it should be used in CSPF path
>computation.
This allows advertising if an LSR can be transit and egress.
This is particulary useful for steiner tree topologies.
See the following example:
Tree T: Ingress = R1 Egresses = R2, R3, R4
R1
|
R2--R3---R4
Such tree can be setup only if R3 has Egress + Transit capability.
Regards,
JL
>
>Thanks
>Satya
>
>
>
>_______________________________
>Do you Yahoo!?
>Declare Yourself - Register online to vote today! http://vote.yahoo.com
>
>_______________________________________________
>mpls mailing list
>mpls@lists.ietf.org
>https://www1.ietf.org/mailman/listinfo/mpls
>