Currently, both AS30036 and AS13335 are BGP peers with the IXP acting as a transparent route server (so even though that's the gateway of last resort for both ASBRs, nobody cares about it for routing). AS30036 redistributes OSPF external type-2 routes to BGP (aka, consumer routes), and redistributes BGP routes to OSPF as type-1. Only has OSPF area 0 for now, thigh as I expand them they might go to IS-IS.
The PE router there is connected to "me" over _serial_ with PPP (for the heck of it) with unidirectional (CE to PE) CHAP auth.
PE router redistributes its connected routes into OSPF, meaning my assigned IP block is sent to the ASBR, who redistributes that into BGP.
Fun fact: the AS13335 router is the origination point of it's announced prefixes (for 126.96.36.199 and 188.8.131.52) with loopback interfaces. AS30036 doesn't announce anything "internal" (yet)
Excuse me while I brainstorm out loud.
The nice thing about this is that, in theory, not every PE router needs a copy of the global routing table, only the backbone routers and ASBRs do. Of course, that means having multiple areas.
In the real world this is a stubby area, or basically a dead-end in the topology. The ABR for the stubby area will re-write the LSAs it gets so that everything in the area uses them as the default route.
Problem, the PE is still technically an ASBR.
Solution: NSSA. a Not-So-Stubby Area seems to be the combination I'm looking for. ASBRs are still, well, accepted, and can send their routes to the ABR, and the ABR acts as a gateway for the entire area. Ergo, the PE router itself (ASBR) doesn't need an entire copy of the routing table, it just sends anything from the CE to the ABR, and that can figure out the correct path across the backbone to the other ASBR with a connection to the destination AS.
The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!