🚀 go-pugleaf

brunkhorst@mayo.edu (Geoff Brunkhorst) wrote about attending Interop:

> Overall, it appears that NeXT will seriously have to look at it's high end  
> networking in the coming year.  A lot of discussion about FDDI (and it's copper  
> brethren CDDI) will be in the <$500 per desktop connection by 1994.

I would suggest that all computer vendors head straight for 
ATM to the desktop. ATM is the first true multimedia networking 
technology. FDDI and 100 Mbs Ethernet are still contention-based 
and allow no mechanism for bandwidth reservation. I just heard
from someone else at Interop that several people were doing live
videoconferencing from Sun workstations hooked up to ATM switches 
from MPR, Fore, and Adaptive. They're using Parallax boards for
video input, compression, and decompression.

Some day ATM is going to provide the backbone of the public switched 
network (though that may take longer than everyone would like). ATM is 
fully scalable from 155 Mb/s up through gigabit speeds. ATM over unshielded 
twisted pair should be possible within the next year as well. When that 
happens, institutions will be able to use existing copper to the desktop, 
with ATM switches hidden in their wiring closets instead of Ethernet hubs. 
These switches will give 155 Mb dedicated to every desktop in a workgroup or
department. These switches will then be interconnected at 622 Mb or higher 
with fiber, and may eventually connect directly to the public network. 

Right now wide area high-speed networking is too crowded with SMDS, 
Frame Relay, and a dozen other technologies that are in various stages
of field trials. The LAN is where ATM will be deployed first. Why have
a bunch of 100 or 200 MIP machines choking each other to death on a 
10 Mb Ethernet? Or fighting over tokens or slots on FDDI? The choice 
becomes even clearer when multimedia (isochronous) traffic is added
to the equation. 

I say, ATM or bust.

--Russ
rs05@gte.com