Re: [Isis-wg] Fwd: New Version Notification for draft-franke-isis-over-ipv6-00.txt

[prz <[e-mail]>]

 On Sun, 22 Nov 2015 17:30:50 -0800, Marc Binderberger <[e-mail]> 
 wrote:
> Hello Tony,
>
> one question about the IPv4 draft "draft-ietf-isis-wg-over-ip": what
> happened
> to it?  Why was the work stopped?
>
> It got to workgroup level, so must have passed some checks and tough
> questions :-) - and then R.I.P ?
>

 The major driver was then (we're talking about Internet Bronze Age ;-)
 I think MCI wanting to play with ISIS without all the SNAP stuff to get
 data onto some PCs. They
 were using early Linux or FreeBSD and the SNAP was a problem.

 We had an implementation for them AFAIR but then Cisco did not support
 the work with the beating 'security through obscurity with ISO' 
 argument that
 "someone could send packet with dst address of your router from 
 'outside
 your domain'".

 Now, don't forget, this was routing area IETF where 'you don't have
 2 major vendor inter'oping in field with real implementation means
 you can't WG LC'  (as it should be IMO). So we had de-facto some Fore
 implementation, CSCO wasn't on it, people like Bay were all in
 corporate where ISIS wasn't used, smaller players didn't have ISIS,
 people changed jobs ;-)
 So, it all got shelved.

 Couple of years down the road all major vendors had proper LLC/SNAP
 support in their kernels/stacks and the issue was not much of an issue
 anymore ...

 To be fair, Tony & Dino (AFAIR again) wrote GRE down (2784) and that
 got very good traction until today and with that the issue was kind of
 moot for major vendors since e'one had to implement GRE and with that
 you could encaps ISIS to anywhere. Irony being of course that people
 started to talk about 'auto-termination of GRE tunnels' and with that
 allow the 'attacks into the middle of network' ;-)

 This is all reconstructing from memory after 20 years so I may have
 gotten some stuff factually wrong, other oldtimers may correct me.

 Having said all that, I still think that ISIS-o-ip4 makes sense for
 its simplicity, my concerns about the ip6 attempts described in the 
 draft
 I already kind of sketched out albeit HomeNet is all about IPv6 
 obviously ...

 Speaking of which, maybe running one-hop auto-GRE tunnel (or some other
 little tunneling monstrosity ;-) & putting ISIS over
 it would be an option for HomeNet (don't think I ever saw that done). 
 Just
 a random thought ...

 -- tony