Right now we have an RDE process stub that receives updates from session and listener processes and does nothing.
We can do an initial implementation of the RDE based on RFC4271 section 9 that:
- performs preference calculation for received routes.
- performs route selection and maintains Adj-RIB-In, Loc-Rib, Adjs-RIB-Out in ETS (could be one table?).
- performs route dissemination to peers after processing updates.
The RDE should also be notified when a peer listener/session BGP session goes down to invalidate the routes and monitor those processes too to invalidate peer routes in case of a crash.
We can postpone route aggregation/reduction, next hop validation and modifying the routing table for an initial version.
Right now we have an RDE process stub that receives updates from session and listener processes and does nothing.
We can do an initial implementation of the RDE based on RFC4271 section 9 that:
The RDE should also be notified when a peer listener/session BGP session goes down to invalidate the routes and monitor those processes too to invalidate peer routes in case of a crash.
We can postpone route aggregation/reduction, next hop validation and modifying the routing table for an initial version.