Hi Scott,
From the May 1999 BCR Magazine, a good article by Eric Krapf on... a lot of things.
bcr.com
Visions of the New Public Network
from the May 1999 issue of Business Communications Review, p. 20–21
by Eric Krapf, managing editor of Business Communications
Review.
We may still be a long way from actually having a new public
network infrastructure, but a vision is starting to emerge about
what a PSTN-of-the-future might look like. The key elements, as
sketched out by entrepreneurs at BCR's recent NGN Ventures
conference, look something like this:
High-capacity access links.
An overlay of smart capabilities at the public network edge to
provision new enhanced broadband services.
A relatively dumb network core featuring a much simplified
service provider infrastructure. This infrastructure would be
powered at first by improved SONET-based technologies, and
eventually migrate to dense wave-division multiplexing (DWDM).
DSL Blues
The access piece is at once the closest to being realized and the
most problematic. On one hand, the technologies for delivering
multimegabits down the last mile are already being deployed:
digital subscriber line (DSL) on local exchange carrier (LEC)
copper lines, and cable modem service from CATV companies.
The problem—no secret anymore, especially in the case of
DSL—is that it is proving much more difficult to deliver the service
than many people imagined.
Rick Tinsley, founder, president and CEO of Turnstone Systems
(www.turnstone.com), who had outlined many of DSL's technical
problems at last fall's NGN conference (BCR, December 1998, p.
16), hadn't turned much more optimistic by spring. "Installation is
still very, very problematic," said Tinsley, whose company makes
a "copper cross-connect," which is intended to ease provisioning
and management of DSL.
Problems include delays in establishing service (and therefore,
delays in carriers earning revenues off DSL), loop quality,
troubleshooting, establishing SLAs and providing high availability.
Tinsley noted that the technology challenges don't lend themselves
to resolution via Moore's Law. "The infrastructure is optimized for
POTS and a regulated monopoly," he said. "These are problems
that are not solved by bigger ASICs and protocol stacks."
Tinsley doesn't consider the technical challenges insurmountable,
but he identified several business challenges for would-be DSL
providers. Though DSL service is often bundled with Internet
access, Tinsley argued that from a service provider's perspective,
these two services create problems of completely different
magnitude. He pointed out that while an ISP can become a
national provider with just 100 points of presence (POPs) in cities
around the country, to provide DSL requires a carrier create
physical locations in every single telephone central office (CO)
where it wants to offer service. Since this might amount to 10,000
sites to serve those 100 ISP POPs, Tinsley noted, "The logistical
challenge of this is very significant."
Moreover, he maintains that customer expectations have been
unrealistic. Customers started out expecting to get 6 Mbps for
$39.95 a month, but "[those] numbers really don't add up today....
It is technically feasible, but it's not economically attractive [to the
carriers], and that's why it's not happening faster."
The Smart Edge
While conference chairman John McQuillan suggested DSL had
already lost the residential market battle to cable modems—among
business customers, the story may be different. Businesses have
been willing to pay more than consumers for high-speed access,
and if a business model for DSL proves successful, it would boost
the concept of an all-IP public network.
That would almost certainly trigger a migration of intelligence from
the network core—as in the current PSTN—toward the edges of
what some panelists at NGN Ventures called the "PSDN"—public
switched data network. But getting from today's telcooriented
model to an intelligent edge won't be simple. James Dolce,
president, CEO and cofounder of Redstone Communications
(www.redstonecom.com), argued that the current infrastructure at
the edge doesn't scale well. Similarly, Stephen Collins, president
and cofounder of Spring Tide Networks
(www.springtidenetworks.com) characterized existing carrier edge
gear as "completely insufficient" to deploy advanced services.
Dolce cited four key areas where improvement is needed in edge
devices:
Port density.
Performance—must be wirespeed.
Reliability—must be telcoquality.
Quality of Service (QOS).
While vendors promised they could deliver these lower-layer
capabilities, the problem goes beyond beefing up the boxes:
Provisioning systems must scale along with the growth of the
edge. Collins offered Spring Tide's proposed solution: a "service
layer," in which intelligence is abstracted from the edge routers
and resides within servers. For example, a policy server would
handle provisioning, while an accounting server would be
responsible for billing. The major obstacle to such a deployment is
that there are no standard schemes for provisioning and billing of
IP services. (For more on billing, see Peter Sevcik's column in this
issue, pp. 10–12.)
A similar intelligent-edge vision was proposed by Anthony Alles,
founder and president of Shasta Networks,
(www.shastanets.com), who envisions "subscriber indexed
service profiles" that let providers deliver new services
cost-effectively. Quick, easy provisioning will be crucial, he said,
because these services will be sold to a large market of lower-end
customers with less-advanced skill sets, he noted.
It's impossible to tell whose exact vision will prevail, but right now,
Shasta has one key edge: Less than a week after the NGN
Ventures conference closed, Nortel Networks bought out Alles's
company for $340 million.
A Simpler Backbone
But if the edge and access portions of the network will have to
change, so will the network core. Panelists at NGN Ventures
discussed how new, integrated access devices would be
combined with optical networking at the core, resulting in networks
that are both higher-capacity and less expensive to build and
operate.
The stages of this migration can be seen in the recent acquisitions
by DWDM vendor Ciena (www.ciena.com) of startups Omnia
(www.omnia.com) and Lightera Networks (www.lightera.com),
which showcased their pieces of the solution at NGN Ventures.
Omnia, acquired for $429 million, calls its AXR 500 product a
"next-generation add/drop multiplexer (ADM)." The AXR 500, which
features an ATM switching core, is intended to replace a collection
of boxes—SONET ADMs, digital cross connects, routers and ATM
switches—in use today by the telcos.
The LECs, according to Omnia's president and CEO, Mike
Champa, "cannot continue to deploy multiple boxes for multiple
services. It's too expensive." He also said, "The number of
network elements has to decline in order for the network to scale."
Champa conceded there's a major obstacle to this goal, and it's
not technology-based. If and when the current tangle of "telco
plumbing" goes, "You're basically destroying the careers of 80
percent of the people that work for the ILECs," Champa noted.
Startups face other problems trying to sell to the ILECs, most
notably the long evaluation times that RBOCs impose. According
to Rob Newman, CEO of Atmosphere Networks
(www.atmospherenet.com), which builds next-generation SONET
gear, the carriers also tend to pressure new small vendors into
partnerships with the ILECs' large incumbent suppliers.
So the ILECs are likely to move cautiously in overhauling their
networks. However, there will be another, more ambitious stage of
network rehab and construction, as evidenced by Ciena's $552
million purchase of Lightera, which builds optical switching.
Lightera's CoreDirector product will reportedly deliver anywhere
from 640 Gbps to 48 Tbps (terabits per second) throughput.
Lightera announced the product last March, and says its first
customer shipment will be in 3Q99.
Today, though vendors use the phrase "optical switching,"
optoelectronic conversion still takes place. But within five years
equipment will switch lasers directly, predicted Desh Deshpande,
the Cascade founder who's entered the optical switching market
with startup Sycamore Networks (www.sycamorenet.com).
Deshpande believes that all-optical switching will open up such an
abundance of core-network capacity that the carriers will be able
to build mesh instead of ring networks. This would cut provisioning
times for big-bandwidth pipes like OC-48s from six months to six
minutes, he said.
Conclusion
John McQuillan honed in on billing and operational support
systems (OSSs) as the key to making the new public network a
reality. McQuillan noted that carriers simply won't use a technology
platform that doesn't allow them to issue bills from existing OSSs.
In other words, no matter how powerful the underlying
infrastructure, bandwidth will never be free—but carriers will
deliver abundant bandwidth only when they've figured out exactly
how to collect revenues from it. And that may be the monumental
task.
|
