The Routing Over Large Clouds Mailing List Archive by date

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

> one problem with cell switchig routes is that they can't handle
> multipoint-to-point connections in any nice or standard way.

What is your problem?

> so they
> really are not transparent to ip as the authors claim.

As there is no terminology "multipoint-to-point" in the world
of IP, you can't see "multipoint-to-point", even though the
architecture is transparent.

In the world of QoS-less IP, functional equivalent of "multipoint-
to-point"  is packet relayed IP multicasting with multiple senders,
which is supported by IP routers including CSRs (note that CSR
works as a packet router for QoS-less communication).

In the world of RSVP IP, functional equivalent of "multipoint-
to-point" is a shared-explicit or wildcard filter, which can
be supported by having multiple VCs and switch them by PATH
messages or actual data flow.

> there is lots of
> hype about them, but the reality under the hood is not at all that
> clean:

Could you give me specific references of the hypes, unless you
are hyping us there are lots of hypes. :-)

I'm glad to dispell hypes on ATM, including but not limited to, NHRP.

> a better alternative would be frame switching routers switching
> frame relay or funi frames.  

As the only merit of ATM compared to other existing medium is that
it can possibly offer configurationless QoSed small cloud datalink
layer, we can have a better alternative with packet switching.

							Masataka Ohta