Hello Mike,
re: ATM in the local loop, here's an article that may avail itself to some degree of perspective on the importance and staying power of ATM in the last mile. It's from this week's (4/4/0) Sounding Board Magazine (copied below, for posterity):
soundingboardmag.com
[[BTW, I commend the entire issue of Sounding Board this week as being highly relevant to a number of topics which I hope to be commenting on both here and over in the Coluccio thread later today, or over the weekend. In particular, see the VoDSL and IP Telephony over Cable articles, and the effects of recent mergers on previous plans to introduce VoIP/IP Telephony by some of the major players.]]
On another note, and as relates to some of ALA's other last mile aspirations, be on the lookout for them to begin announcing some last mile Ethernet directions, possibly GbE, as would be supported in large part by their acquisition of Packet Engines, not too long ago. IMO.
FAC
---------ATM article begins:
"ATM: The Clear Choice for Local Access"
By Brough Turner
Until a few years ago, ATM
was promoted as the long-term
answer for voice and data
convergence--the ultimate
networking protocol. Until 1995, this made sense.
ATM is a scalable technology. It was designed from
the outset to support voice, video and data packets
on the same wire (or fiber, or wireless channel).
ATM has virtual circuits. It supports very high-speed
switching and provides QoS guarantees.
Unfortunately, development and deployment of
ATM took too long. First, it failed to take over the
desktop, succumbing to the lower cost and ubiquity
of Ethernet. Then, the web and IP took the world by
storm. IP became the dominant network protocol.
Now, all applications run on IP, and IP runs on
every kind of transport. So there is nothing on the
horizon to challenge IP as the dominant networking
technology.
If we forget ATM's earlier pretensions and just
consider ATM as a transport for IP, it's very good.
It's also very scalable. In fact, when the growth of
the Internet exploded in 1996 and 1997, ATM was
the only technology on the market that could provide
the optical transmission rates (155mbps and
622mbps) IP backbone providers wanted. There
are other choices for high-speed IP, but ATM
continues to grow in the backbone because it is the
only transport mechanism that can provide different
classes of service on the same fabric. With ATM, a
backbone provider can combine IP and legacy voice
traffic on the same network without interference.
Perhaps more significant for the coming all-IP world,
ATM's QoS guarantees can support differential IP
services, given there is networking equipment
intelligent enough to map the IP-based application
flow to the proper ATM class of service. This means
you can run VoIP traffic together with
mission-critical IP applications and bulk file transfers
on the same IP over ATM network, without
interference or loss of voice quality during file
transfers.
Many IP networks are built on top of ATM
networks. However, with DWDM and new fiber
networks providing vast amounts of bandwidth for
the backbone, a new generation of IP carriers is
bypassing ATM and providing guaranteed IP service
levels by overprovisioning. They are throwing
bandwidth at the problem, expecting the IP
differential services (DiffServ) architecture and other
emerging IP QoS technology to provide eventually
the capabilities that ATM provides today. These
new carriers' backbone IP routers connect directly
to the fiber network without the need for, or the
expense of, an ATM layer. While ATM backbones
are still growing, ATM is about to be eclipsed again.
There is one place where ATM is an excellent
match, and that is local access. Whatever happens to
other ATM markets, ATM has a strong future as the
transport protocol for the access loop--the last mile.
Here, there's a dearth of fiber, and the copper cables
are owned by entrenched monopolies. Local access
remains the most expensive part of any WAN. This
is where small and medium-sized businesses can
benefit by combining voice and mission-critical data
with e-mail and file transfers on a single T1 or xDSL
access line. ATM supports this, thus dramatically
reducing costs in the one area of the network where
bandwidth is most expensive.
Two years ago, T1 ATM access was hard to find,
but now ATM is available on both T1 and xDSL
from multiple service providers in North America.
The market is growing rapidly and will continue to
do so for many years. While there is an inexorable
trend to move everything onto IP, legacy services
never die. It is safe to predict 15 years from now
there still will be 9600 baud data circuits, X.25 and a
wide variety of private leased lines. The most
efficient way to provision these legacy services in the
access loop is ATM. In the long term, ATM will be
superceded by native IP mechanisms. Fiber will
reach the local loop and IP DiffServ will challenge
ATM for QoS. But it will be more than five years
before anything impacts the growth of ATM in the
access loop.
Brough Turner is senior vice president, chief
technology officer and co-founder of Natural
MicroSystems Corp. (www.nmss.com). He can be
reached at rbt@nmss.com. |
