> Hi Dino, Thanks for the comment Sharon.
> Im not sure if it makes sense, but when we were in the satellite IP session > in IETF, > it was explained that: > > - as satellite routers move around in orbit > - each assumes RLOC of sky sector its in Yes, those are addresses for the satellite nodes. They are not RLOCs stored in the mapping system. The ground-station xTRs are the ones with the RLOCs that are stored in the mapping system and what the satellite underlay delivers packets to. The spec does not propose to have LISP encapsulate to satellite nodes. The satellite network is a pure underlay. Potentially a satellite node could be an RTR (but quite likely not reachable from a GS-xTR). But that is for future study. > So it can be simple for a ground station XTR, > to source plot the shortest sky grid path. Well a GS-xTR prepends an outer header with a destination-GS-xTR RLOC. That is the instruction for where the satellite nodes should deliver the packet. This is no different than a terrestrial underlay. > Each such path is good for a short while, > as these satellites keep moving. The source-route references in the spec requires a lot more design and thought. And IMO, is complex. I would say lets get the satellite network to deliver packets hop-by-hop at first pass. > But its relatively simple to track and predict - > if a sector is occupied and reachable from neighbor grid sector. If you say so. ;-) > So naturally hierarchical hexagonal grid comes to mind :) hierarchy > reflecting transceiver range. > > Just a thought. The problem with using a source route of satellite node hops, is by the time you find the path, the links or sats are gone. There can be a lot of entropy in the system if you try to manage this on the ground-stations (head-end). Dino _______________________________________________ lisp mailing list [email protected] https://www.ietf.org/mailman/listinfo/lisp
