The MPLS WG Archive

["Gibson, Mark" <mgibson@orchestream.com>]

> I believe that making a decision on whether RSVP-TE or CR-LDP should be on
> standards  track based  upon marketshare  should  NOT be  how a  standards
> organization works 

This is really  a misrepresentation.  They are both  on the standards track.
The question  is whether  a marketshare of  approximately zero is  reason to
remove a document from the standards  track.  In fact, it's fairly common to
remove documents from the standards track for just this reason. 

The only real  issue is whether there is any  realistic prospect (as opposed
to,  say,   wishful  thinking)  that  CR-LDP  will   ever  gain  significant
deployment.  A few of the folks favoring CR-LDP have actually addressed this
issue, but not, it seems to me, with any great degree of confidence. 

[MG] The fact that CR-LDP has little market share should be no surprise at
all.  Given that the two dominant core router manufacturers offer only the
RSVP-TE solution will tend to ensure this becomes the de facto standard
whatever the technical rights and wrongs of the implementation.  To stop
this process would require the IETF to enforce a fair fight-off by requiring
all MPLS-TE LSR manufacturers to supply both protocols and have the market
decide.  This is clearly an unsustainable situation.  Also, given that, from
a service point of view, both offer broadly the same functions, there is no
drive from service providers to require one over the other to drive the need
for CR-LDP support.  (Neither have vrf-to-tunnel mapping support but that's
another issue...)

[MG] So, using the absence of a clear service requirement *and* support from
vendors for the implementation of both protocols, there seems little point
in progressing two standards.  That's what the market share would dictate.

[MG] But surely the IETF is not in the business of deciding which protocols
should be used, rather ensuring that those developed are done so to the most
exacting of standards.  Therefore there should be no decision to move CR-LDP
to informational but move it to historical once it becomes so. 

[MG] If not, can we only have one PWE3 protocol please, one IGP please, one
L2VPN discovery protocol please....

Mark