[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: LC comments on considerations extranet section
Hi Eric, folks,
12/06/2009 21:02, Eric Rosen:
The section on extranet 3.3.6 is not accurate.
It suggests that it is not known how extranet could be achieved without
using BGP-based C-multicast routing. This is odd considering (a) a solution
providing this feature has been shipping for years, and (b) the solution is
written up in draft-rosen-l3vpn-mvpn-mspmsi-02.txt.
One could point out that :
- the section you quote does not say "not known" just "not specified"
- that a feature shipping is not a spec (!)
- that draft-rosen-l3vpn-mvpn-mspmsi-02 is not a WG doc, and that the
scope of the draft is only WG docs
- (also, that draft-rosen-l3vpn-mvpn-mspmsi was not even talking about
extranet when the paragraph in question was written one year ago)
So, nothing "odd" really.
It also suggests that all it takes to do extranet with BGP C-multicast
routing is to reuse the "already defined and well understood BGP route
target import/export semantics". In fact, this is not true. To support
extranet you also have to know which C-streams in which PMSIs are the ones
you need to receive. In the PIM control plane, you make this determination
by associating each PMSI with a particular PIM interface. I don't think the
corresponding functionality is specified for the BGP control plane.
With PIM, the fact that the dataplane and control plane are so tied, is
a double-edged sword, which happens to impact how extranet can be done.
I think it is premature to make a recommendation based upon which extranet
scheme works best until we have both extranet schemes written up in a
standards track document.
The (not) funny thing is that, if there had been an agreement to include
Extranet in the base specs, for both PIM and BGP, consistently with mVPN
requirements (RFC4834), we would have an interesting comparison to do.
But I have to agree with you that, since draft-ietf-l3vpn-2547bis-mcast
has unfortunately finally not included Extranet, this section 3.3.6 is
in a delicate position. I don't mind removing it.
And I hope further work will be done soon on the Extranet subject.
Cheers,
-Thomas