On Fri, Mar 04, 2022 at 02:39:17PM -0500, Joel M. Halpern wrote:
> I do not believe the community has an agreed definition of identifier or
> locator. We do have some relatively common usage for locator. As far as I
> know there is no fully acurate and written down definition even for that.
Thanks for reconfirming. I just became a fan of the white knight approach,
because it removes the need for any absolute truth.
> Note also that while Dino likes LISP for lots of things (for good reason),
> the LISP working group has been working to move the LISP RFCs to Proposed
> Standard with descriptions of usage that are for particular use case (not
> limited domains, but specific uses). While originally developed to address
> Internet-Wide scaling problems, that is NOT what the PS documents cover.
Yes. I think we will have to wait for even Tony's proposed prefix improvement
work
to not suffice, but instead for the Internet Core/ABSR routers to finally
not scale anymore before enough operators would come back asking for
architectural improvements. But i think even before that happens, one could
offer
maybe some partial deployments of LISP as a service to tier-2++ smaller
ISPs who want to multihome to more than one Internet uplink provider but
still not carry the whole BGP routing table. But i have no good idea of
where the operational pressure points are these days that could lead to such
incremental evolution from the edge.
Cheers
Toerless
>
> Yours,
> Joel
>
> On 3/4/2022 6:39 AM, Toerless Eckert wrote:
> > On Thu, Mar 03, 2022 at 09:28:23AM -0800, Dino Farinacci wrote:
> > > > of its address structure helps the underlay to locate the entity (xTR)
> > > > that the
> > > > address is assigned to (xTR). So the name 'locator' is 'just' a good
> > > > name for what LISP calls/uses the address for, not for how the under
> > > > itself would maybe call the address or use the address for.
> > >
> > > Well the locator you put in an outer header destination address is
> > > called/used/assign to whatever the rules of the underlay are. If the
> > > underlay is ethernet, then its a 6-byte address where the high-order 3
> > > bytes is an organizational ID, just to cite an example.
> >
> > Indeed.
> >
> > I have not seen an answer to the question i posed earlier in the thread:
> > whether and if so what general (not technology specific) definition of
> > locator
> > and identifier the IETF may have. But i have seen a lot of confusion about
> > it and people shying away from using these terms.
> >
> > If (as i think) we do not have a commonly applicable definition of
> > locator/identifier
> > (beyond its use in indivdual technologies like LISP), then i think this is
> > because
> > folks who tried to apply these terms (incorrectly) may have failed to
> > see the difference between what an address is and what someone (like an
> > application) calls it (/uses it for). In that respect the reference to
> > the White Knight in IEN19 is very helpful to remember.
> >
> > Cheers
> > Toerless
> >
> > > Dino
> >
--
---
[email protected]
_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area
