Yes, the edge router may, but does not have to, perform IPREF mapping.
It can pass to any router withing the private network.
On 2/15/23 11:04, Evan Pratten wrote:
Ya, I guess using non-ip-addresses for the refs is a good idea for
networks that involve non IP-based hops.
Would it be possible to have a router do reference pass-through? I'm
thinking of a kind of double-NAT situation where I might want router 1
to delegate the routing of refs to router 2.
WAN <--> R1 <--> R2 <--> Clients
On Tue, Feb 14, 2023 at 10:08 PM waldemar <walde...@wdmsys.com> wrote:
I was not thinking of chaining, this sounds like source routing, I am
not sure. Cascading is certainly possible. The destination may rewrite
one IPREF address into another IPREF address. This could be done
multiple times.
I wanted to avoid any sort of negotiations, any kind of time dependency,
and I was trying to minimize amount of information shared between peers.
The peers don't trust each other that much except to agree to
communicate. Using real addresses leads to negotiations and requires
knowledge of peers address spaces and protocols, so that was not a good
option. I was thinking of peer networks (multiplayer games, NAT
traversal), high delay networks (space networks), and highly secure
networks (financial, military). I thought avoiding negotiations would be
the key. I was also convinced we'll be dealing with more than one
network protocol for a while, hence no dependency on a single protocol.
IPREF might speed up unification, especially IPv6 in the Internet, but
it could also make it easier to develop specialized network protocols.
Maybe for high delay networks, maybe for highly secure networks, or
maybe for simplified networks.
On 2/14/23 12:25, Evan Pratten wrote:
I find this very interesting.
Would it be possible to chain references? for example
10.0.0.1+700+800? I can't think of a use case for this, but I'm sure
it would cross someone's mind to try.
The way I see this, IPREF is essentially encoding some or all of the
route to the final host in the address. Why not use real IPs all the
way down? For example: 10.0.0.1+10.0.0.4. This wouldn't require any
translation of reference numbers. Although, would make things less
dynamic.
---
Evan Pratten (VA3ZZA)
https://ewpratten.com
On Tue, Feb 14, 2023 at 11:10 AM waldemar <walde...@wdmsys.com> wrote:
Hello,
I have submitted a new -00 draft,
https://datatracker.ietf.org/doc/draft-augustyn-intarea-ipref/. I am new
to this, although I worked on an RFC some 15 years ago. I have contacted
ADs for the area who advised me to seek feedback on this list. Please,
provide your thoughts. I will be also submitting proper declarations in
compliance with BCP 79. I need more time for this.
Thank you
Waldemar Augustyn
_______________________________________________
Int-area mailing list
Int-area@ietf.org
https://www.ietf.org/mailman/listinfo/int-area
_______________________________________________
Int-area mailing list
Int-area@ietf.org
https://www.ietf.org/mailman/listinfo/int-area
