[ih] When did "32" bits for IP register as "not enough"?

[Joe Touch]

> On Feb 13, 2019, at 6:12 PM, Jack Haverty <jack at 3kitty.org> wrote:
> 
> …
> 
> E.G., the "Multi-homed Host" problem - namely how do you deal with a
> user computer attached to more than one network.

See below. Unsurprisingly, it involves a router inside the end “host”.

>   Does it have two IP
> addresses? 

Yes, and more for the internal virtual router.

> Can it effectively use both network ports at the same time? 

Yes, with “policy routing”.

> Can it be constrained to not act as a gateway?

Yes.

> ...
> 
> I haven't been involved for a while, but I wonder if those outstanding
> problems have been solved, perhaps in IPV6.

We experienced this problem in the mid 90s when trying to deploy a somewhat unreliable prototype “gigabit LAN” for operational use. We wanted division-wide participation, but not to interfere with use if (when) the links failed. That included avoiding renumbering impact on long-lived TCP connections (notably telnet logins), NTP (which we couldn’t restart on new addresses without affecting running programs using disk-based files), etc.

Our solution is documented here, and became part of the basis of our understanding of the nuances of overlay networks in the X-Bone project:

Touch, J; Faber, T: Dynamic Host Routing for Production Use of Developmental Networks <>. In: ICNP, pp. 285-292, IEEE, 1997.

Joe


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://elists.isoc.org/pipermail/internet-history/attachments/20190213/33b2d0ba/attachment.htm>