On Tue, 5 Mar 2002, Howard C. Berkowitz wrote:
> 3.11 Multi-homing
> Is multihoming really the right term here? The problem as seen by
> users is fault tolerance, and its solution may properly involve
> sub-IP, network layer, and higher layers. Should this definition be
> constrained to focus on multiple routed paths to all elements,
> without prejudicing mechanisms above and below the routing layer?
Fault tolerance here involves not only different layers, but also the
business relationship between domains. Multihoming works only for domains
with customer and provider relationship. However, for peering
relationship, users need multiple routed paths as well. Therefore the
definition shouldn't prejudice business relationship.
Multihoming is a very vague term and at least implies mechanisms as well.
There is a need to extract what is really the requirement.
/Yong
>
>
> see both multi6 and draft-berkowitz-multreq-02.txt for possible terminology.
>
>
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:04 EDT