On Sat, Jan 13, 2007 at 11:32:53AM -0800, Christian Huitema wrote:
> > > I really do not understand you. Try in French :-) Because,
> obviously,
> > > if it is unicast, it is not multicast :-)
> >
> > unicast is a degenerative case of multicast.
>
> Bill, this is emphatically not true for high speed wireless links, such
> as the upcoming IEEE 802.11n standard.
>
> In high speed wireless networks, the physical layer gets tuned between
> sender and receiver, resulting in huge gains in transmission quality.
> You can only apply a fraction of these tunings to multicast traffic.
re wireless... unless you have tightly coupled directional
antennas, high speed wireless (one presumes that "high speed"
in this context is the 802.11n draft standard) there is
a "footprint" of signal propogation that -any- receiver
within that "footprint" may receive.
> There are also some nasty interactions between multicast and power
> saving. To save power, the stations sleep most of the time, wake up
> occasionally, and poll the server for any queued data. For multicast,
> you have to either guarantee that all stations wake up at the same time,
> which is hard, or accept to effectively replicate the multicast packet
> for each station.
for multicast, you can assume that all intended receivers are
"awake" for your application design, but that is not (to my
understanding)
an intrinsic component of multicast.
> Transmission tuning and power saving are both very desirable. The
> natural consequence is to try avoid multicast operation whenever
> possible.
The first i agree with, the second depends on design goals.
--bill
>
> -- Christian Huitema
>
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
