#1 I am afraid that what Tony suggested may be needed regardless ....Let's imagine I have a site (home/office) with two xTRs xTR1 & xTR2 connected to the upstream via DSL.My DSL IP is auto-negotiated every 24h. That set by the provider. Both do it asynchronously.The fact that I loose xTR1-out interface will not trigger the xTR2 to set loc-reach-bit indicating that xTR1 is down as from the IGP of my home it xTR1 is looking just fine.
Yes, it will if you run an IGP. Other alternatives is to run HSRP or VRRP between them.
"An LISP+ALT router near the edge learns EID prefixes which are originated by authoritative ETRs, either by eBGP peering with them or by configuration."
And what does this reference have to do with your point?
#2If xTR talks BGP to first ALT how is this less BGP or multihoming without BGP ?
It is eBGP only over a GRE tunnel. The configuration is basically a neighbor and network command. And in the case of your DSL router, it wouldn't run BGP, it would be a low-opex router and the other side of the GRE tunnel would be the ALT router that originates your EID-prefix into the ALT.
#3If xTR talks BGP to first ALT how are you dealing with my side changing it's IP dynamically ? AFAIK BGP would require some IP address to come up.
The eBGP peering session is on the IP address configured on the tunnel. That could be a private address. You are just establishing a TCP connection from your box to the ALT box on a directly connected subnet. The subnet for the GRE tunnel.
The point of LISP is to decouple locators. Those locators which are changing will only cause a change to your mapping entry and no other configuration.
Or is the LISP/ALT solution only for those who would have static locator addresses ? I
Absolutely not.
know that mobility is last on your slide but this is not mobility. After all my house does not move that fast :)))
That is good. Your house, like others, will put less stress on the mapping system.
Dino
Cheers, R.There is no chance in hell this will ever get deployed. If it would, people would have not objected in using PIM Rendezvous Points in other SPs. It's the same issue.|If you put xTRs at the PE, they will never talk to each other when |deployed in different SPs. And you want them to know about each|other's reachability so they can set loc-reach-bits for themselves as|well as each other. That has to happen at the site. Plus it scales |better there. This doesn't follow. One could, for example, have a protocol orconfiguration on the xTRs that establishes reachability (from the site side) of each of the xTRs. If there was an agent on the site that coordinated this actively, it would only result in customer-provider interactions, whichcould reasonably scale.Dino _______________________________________________ GROW mailing list GROW at ietf.org https://www.ietf.org/mailman/listinfo/grow
_______________________________________________ GROW mailing list GROW at ietf.org https://www.ietf.org/mailman/listinfo/grow