The Routing Over Large Clouds Mailing List Archive by date

[Telford001@aol.com]

In a message dated 96-05-31 16:15:01 EDT, myjak@tridis.ist.ucf.edu (Michael
Myjak) writes:

>Paraphrasing, Steve said:	
>
>	ATM is just packet shredding with overheard. If you need
>    [mp-to-mp] connectivity, why not just use IP? its sufficient to do
>    the job, and there's already a protocol defined *and field
>    tested*. Besides, there's an entire Internet worth of "rooters"
>    out there to support it.

In the data arena, high performance low cost packet switching
is sine qua non to build a high quality, high performance internet.

As packets most commonly are delivered to the same 
communications subnet from which they are received,
my partner and I concluded in 1986 that by Amdahl's Law, 
the key to building a high performance computer network 
was the optimization of common LAN communications 
subnet technologies by building cheap full-bandwith LAN 
switches (fully-powered bridges).  

These LAN switches would serve in the construction  of
high performance switching fabric communications subnets, 
which would consist of collections of fully powered LAN 
switches and LAN segments

Routers whose network interfaces were themselves
fully powered LAN switches would then serve to interconnect
these LAN switching fabric communications subnets
into high-performance internets.

I architected such a router in 1986-1987.  I designed such 
a VLAN router in 1989-1990 (I called it the LSB [Logical 
SubBridge] Router -- I describe my reasoning in gory detail in 
ftp://hsdndev.harvard.edu/pub/ndtl/doc/{rtebridg.ps,rtebridg.ps.Z}).  

My VLAN router was PC magazine editor's choice in March 1993. 
Currently Xylan and Cisco are marketing and selling such routers.

Today, ATM is a communications subnet technology that has 
a lot to recommend it as well as some drawbacks.  If ATM becomes
the communications subnet technology of choice, then the goal
should be to build high performance low cost ATM communications
subnets.  

To achieve this goal, high performance low cost ATM cell switches 
need to hit the market.  The right way to interconnect
ATM cell-switching communications subnets is to use routers
whose network interfaces are themselves high performance,
inexpensive ATM switches, i.e. the CSR (Cell-Switching Router).

The CSR is one natural evolutionary path of the VLAN
(LSB) router.  An ATM SubSwitch replaces the Logical
SubBridge or VLAN.  The TTI VLAN router offers three
possible mechanisms of path selection through the
LAN switching fabric.  In the CSR, ATM path selection
replaces these LAN path selection mechanisms.  

In an internet constructed with well-architected CSRs,
there are no possible conflicts between network layer
and communications subnet path selection mechanisms.
(I discuss this issue from the standpoint of VLAN
routers in my document).  NHRP is not needed.

Joachim Martillo
Father of the VLAN Router
Grandfather of the CSR
President 
Telford Tools, Inc. (TTI)
"Garage Mechanics for the I-Way"
17 Pleasant Hill Ave.
Dorchester/Mattapan, MA 02126-2813
V:  617-298-1617
F:  617-298-1745
E:  Telford001@aol.com