Re: Network layer reqt? [was Re: Transport level multihoming]

[Greg Maxwell <gmaxwell@martin.fl.us>]

On Thu, 5 Apr 2001, Masataka Ohta wrote:

> You miss the point.
> 
> > > Most TCP applications are so stateful that they can not accept a
> > > loss of bytes.
> 
> Neither SVC nor TCP reconnection restores lost data.

No. But today, connections with TCP/IPv4 and typical IP multi-homing get
dropped and data gets lost when the prefered path fails. Traffic is lost
until the routing reconverges which is usually long enough to back TCP off
into oblivion.

I believe it is acceptible to have a IPv6 multi-homing solution which
legacy multihoming support for older applications that causes sessions to
have to be restarted when a link fails as long as they can be restarted
right away and the same multi-homing solution allows for future enhanced
protocols (such as SCTP) to perform multihoming more intelgently.

Not that it wouldn't be nice for these legacy sessions to continue
operating during a path failure, but considering the current state of the
Internet, it should not be a basic requirement for the IPv6
Internet's multi-homing, in my opinion. 

I do believe the capability for future transports to impliment seemless
multi-homing a requirement. (i.e. We need to match the user performance of
IPv4 multi-homing today, and have the potential to exceed it if the users
upgrade applications).