Excerpts from Brian E Carpenter on Fri, Apr 03, 2009 02:57:24PM +1300:
> On 2009-03-31 05:59, Scott Brim wrote:
> 
> > Excerpts from Dino Farinacci on Mon, Mar 30, 2009 09:31:25AM -0700:
> >> I think we need to keep all this terminology stuff simple. And not
> >> have too many terms. What comes out of this group will pro-market or
> >> negative-market locator/ID separation. And you don't want people to
> >> react "this locator/ID separation stuff is more trouble then it is
> >> worth".
> > 
> > For me it's about the meaning of 'locator/identifier separation', and
> > mainly in endpoints.  The essential point is to get identification
> > functions to stop using data that might have to change when topology
> > changes.  
> 
> Thankyou! This is a great insight. I think we're seeing that all
> these addressy-locatory-identifiery labels are a bit slippery and
> can change their nature according to context and usage. But the function
> that's using them needs to use the most stable label that has the correct
> semantics for the specific use. And if it fails, it's always necessary to
> go up a level and try again.
> 
> In primitive terms, in today's world:
> 
> if a MAC address fails, ARP again
> if an IP address fails, DNS again
> if an FQDN fails, Google again ;-)
> 
> In an ID/LOC split world:
> 
> if a MAC address fails, ARP again
> if an IP-locator fails, map again
> if an IP-identifier fails, DNS again
> if an FQDN fails, Google again ;-)
> 
>     Brian

All true.  And if I need to give you a referral, I might include the
chain of fallbacks I would take if the "primary" name I give you
doesn't work.
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to