Re: Transport level multihoming

[Daniel Senie <dts@senie.com>]

Greg Maxwell wrote:
> 
> On Tue, 3 Apr 2001, Joe Abley wrote:
> 
> > On Tue, Apr 03, 2001 at 01:03:48PM -0400, Greg Maxwell wrote:
> > > > The requirement that no changes should be required in applications
> > > > in order to multi-home does not feature in our draft. Do you think
> > > > it should be there?
> > > >
> > > > If so, why?
> [snip]
> > > the above, why should backwards application compatibility for multihoming
> > > be a factor at all?
> >
> > So, is that a "no"? It looks like a "no" but I thought I'd check :)
> 
> Too busy with the "why" to answer the question. :)
> 
> No, I don't believe it should be.
> 
> The reasoning stated more clearly: Application compatibility for
> multihoming in IPv6 is not important because IPv6 itself breaks
> application compatability to a similar extent as multihoming changes
> would. IPv6 supporting application can be deployed with multihoming
> support thus making the marginal cost of multihoming support zero.
> 
> Additionaly, multihoming changes would likely signifantly lower backwards
> compatibility cost then IPv6 itself (for full IPv4 backwards compatiblity
> you practically have to run a second overlay network) while special
> multihoming support would only have an impact in some additional
> application code and the loss of multihoming for legacy applications (good
> reason to upgrade!).

OK. So today end users can multi-home using certain NAT boxes, having
zero cost to upgrade to IPv6 a fixed cost to get multihoming, and no
recurring costs. This permits users to comfortably ignore IPv6, new
protocols, and new applications. It means we'll be continuing to
encourage the use of NAT (path of least resistance).

Multihoming is important for anyone using the Internet for mission
critical work. This is already a given. These folks have applications
deployed, which are not going away. No amount of wishing will cause
these facts to change. This is, as Brian said, a problem we have to
solve, not one we find work-arounds for.

SCTP indeed looks promising for a variety of reasons. Replacing the
entire transport layer of the Internet, however, is not going to happen.

-- 
-----------------------------------------------------------------
Daniel Senie                                        dts@senie.com
Amaranth Networks Inc.                    http://www.amaranth.com