Re. ATM Comes and Goes

[Sam Ghandchi <[e-mail]>]

Hi,

I think what is happening to ATM is that it is
cart before the horse.  In networking it is
usual that before even a protocol is out of draft
form, there are even products based on the 
secondary outcome of the implementation of the 
protocol developed.  I think many of us 
remember all the products that were based
on the primary and secondary  and ... outcomes
of OSI, but in telephony, people had telephone system 
a long time before having local intercoms or voicemail.
 
I think the whole notion of creating ATM LAN or ATM WAN 
is what has messed up ATM.  I think ATM should have focussed
on a new voice/video/data infrastructure, the thing that
has been commercialized by the term Information Highway.
 
Unfortunately instead, there has been all kinds of carts
before horses that now has ended in the current fiasco
and despair in the ATM community, which is only 
comparable to what happened at the end of OSI 
carts-before-horse protocols and products ...
 
I still think that ATM network infrastructure is a viable 
infrastructure and will have to be built and it is equivalent
to the worldwide telephone system which took a long time to 
be built.  I do not know in the absence of a giant AT&T 
in our times, who will take this task and whether it
will be a cableco or a telco or what, but I do not think that any 
other alternative technologies can even remotely be comparable 
for such a basic infrastructe.
 
But I think one good result of all the ATM works has been 
that networking today after the first upsurge of ATM is 
different from the past.  Today's world of networking
is the world of SWITCHED NETWORKS, whether LAN
or WAN.  This is a new thing.  Nobody thought that 
switching would become so prevalent in Ethernet,
Token Ring ... networks.  I think ATM has a headstart
in recognizing that the INTEROPERABILITY is central 
to success once switching architecture is chosen for
basic network structures.
 
What I think this means in the real world is that 
an UNIVERSAL SWITCH should easily sense the
network technology (Ethernet, Token Ring, ATM,
FDDI, Fiber Channel, Frame Relay, ISDN, etc.).  In other 
words, I think the next generation of switches should be 
multi-network technology switches rather than separate 
interfaces for ATM or Ethernet or Token Ring ....  Just 
like the movement from single-protocol routers to 
multiprotocol routers.
 
I think in such a switching world of networking, it 
would not matter what the world-wide connectivity 
infrastructure is.  But as I said I think that
fabric will be ATM, as I am not convinced any
other technology to be able to offer better scalability 
for such a worldwide infrastructure.
 
My $.03: -)
 
Best Regards,
- Sam

-- 
---------------------------------------------------------------
Sam Ghandchi
[e-mail]
---------------------------------------------------------------