[Olsr-users] [manet] OLSRv2 - RouterID

[Henning Rogge]

Am Friday 03 April 2009 13:28:54 schrieb Dearlove, Christopher (UK):
> Incidentally Thomas just called me about something else.
> He's got email problems, so nothing personal he hasn't
> contributed to this. (I had some yesterday. Hope it's not
> catching!)
Ahh okay, thank you for telling me.

> Options do not go down well with the IESG. (See also comment
> below.) And it still would need to be tested etc. (We actually
> try to follow the "running code" pattern.)
We are in the process of producing running OLSRv1 code for it ;)

> I have no problems with that.
Okay :)

> I think that you might also be interested in an idea Thomas
> and I had that also relates to this area. We had planned to
> write it up as an extension/informational at some point.
> While I don't speak for Thomas, I would be happy, and I
> suspect he would be too, to bring you in on it if so doing
> got it written sooner than we might get to it.
>
> It arises from the basic question "what is an address?".
I talked with Thomas a lot about the idea tu use OLSRv2 with 802.11s or 
another MAC-layer based network. Theoretically it should run with anything 
that has unicast and broadcast addresses.

> In NHDP/OLSRv2 it's an IP address. But 5444 allows other sorts
> of addresses. Now we can't simply write NHDP/OLSRv2 to use
> any sort of address because of two issues
>
> - The NHDP simplification of "you've only got one address,
>   use the IP sending address rather than put it in the HELLO
>   message".
>
> - The OLSRv2 issues of that we use sending IP addresses to
>   determine where a TC message came from (and match it to
>   the neighbours we know the HELLO-advertised addresses of).
As long as the network below packetbb has the concept of source and 
destination address this might still work... but of course it's a little bit 
more difficult to describe in the RFC, that's right.

> But we could consider using addresses in HELLO/TC messages
> that aren't IP addresses as long as they are convertable
> to/from IP addresses. (Thomas took notes of the discussion
> where we thrashed this out together in SF, and right now I
> forget the two conversion direction issues - you need
> convertability in the two directions for different reasons
> and with different consequences, and different subsets of
> addresses - you don't need to be able to convert an address
> you never see, but we may need extra error conditions for
> what happens if presented with one you shouldn't see which
> is one of the complications.)
Broadcast/multicast addresses might complicate the conversion too, but I see 
the idea. We talked about something similar for OLSR.org in the past, 
exchanging out IPv4/IPv6 unions in the code with pure IPv6 and do the 
conversion during packet parsing/generation.

> How might the conversion happen? Sometimes it may be well
> defined (IPv4 to IPv6) and sometimes it may be a configuration
> issue (if we are agreed that all addresses are in 10.0.0.0/24,
> why send the 10.0.0 part even once?). What other cases are there?
> (That's an open question. A key point is that router A needs
> to be able to convert router B's HELLO/TC addresses to/from
> IP addresses, not just its own.)
And they have to agree how to convert it, which can be a little bit difficult 
if you have multiple OLSR networks running.

> Why didn't we put this into NHDP/OLSRv2? First note that we try
> to apply the same standards to our own ideas as to anyone else's.
> (We're not perfect, we may not always succeed, but we try.)
> - Time and related matters. I think that one I spelled out in
>   my last email.
> - IESG. We have had our fingers burned by getting 5444 through
>   the IESG, it was a lot more painful than we expected, and we
>   are trying not to offer any more hostages to fortune. Addresses
>   that may or may be not IP addresses, and complicated conversion
>   issues sounded our warning bells. But as a separate draft, that
>   at worst just fails and doesn't take NHDP down with it, fine.

I know, Thomas told me about it...

Henning

*************************************************
Diplom Informatiker Henning Rogge
Forschungsgesellschaft für
Angewandte Naturwissenschaften e. V. (FGAN) 
Neuenahrer Str. 20, 53343 Wachtberg, Germany
Tel.: 0049 (0)228 9435-961
Fax: 0049 (0)228 9435-685
E-Mail: (spam-protected)
Web: www.fgan.de
************************************************
Sitz der Gesellschaft: Bonn
Registergericht: Amtsgericht Bonn VR 2530
Vorstand: Prof. Dr. rer. nat. Maurus Tacke (komm. Vors.), Prof. Dr.-Ing. 
Joachim Ender (Stellv.)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.olsr.org/pipermail/olsr-users/attachments/20090403/b50aeba3/attachment.sig>