Hi Dhruv,
In "draft-king-pce-hierarchy-fwk-02", section 5.6.2 [Hierarchical PCE
End-to-End Path Computation Procedure]:
Domain paths are denoted as (S-BN11-BN21-D2-BN23-BN31-D,
S-BN11-BN21-D2-BN24-BN32-D, and S-BN13-BN41-D4-BN42-BN33-D).
i.e. Parent PCE must know about the BNs and there interconnections.
What is the need for this?
I think you are refering to a specific example of the potential paths
determined in a parent PCE that will give rise to computation requests sent
to child PCEs to compute specific segments of the potential paths.
Nothing here says that there is a specific representation of domain paths.
But note that there is no need to represent the domain path to anyone by
anyone in this model unless one of
- the ingress PCC has some restriction it wants to apply
- the path computation will be hybrid using some elements of BRPC
- the resulting path is a hybrid that will require some elements of PDPC
Thus, the question simply does not arrise.
Why can't the domain sequence be represented just by a sequence of domains
i.e. [S-D2-D and S-D4-D]?
BRPC procedure can take domain sequence as input and figure out the
appropriate BNs and there interconnection while applying the VSPT
algorithm.
Thanks, yes, that is how BRPC works.
Also note that, Child PCE can use the PCED-discovery [NEIG-PCE-DOMAIN
Sub-TLV] to specify the neighbor PCE-Domain(s) to Parent PCE, without
mentioning any BNs link-interconnection information.
That is true.
However, in a complex mesh of domains, knowledge of the inter-domain links
can help the parent PCE to reduce the potential domain paths and so reduce
the processing that the child PCEs have to do.
Cheers,
Adrian
_______________________________________________
Pce mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/pce
