LANTimes article. Q&A with UUNet, PSINet and GTE CEOs (Part II)
wcmh.com./lantimes/98/98apr/804b022a.html
LT: I've seen estimates that Internet traffic may increase
tenfold in the next year. How will ISPs keep up?
Sidgmore: We've got to build more facilities, lay more fiber
infrastructure, and deploy more switches and routers. We're
going to continue to build out that superstructure, and the
way to build that out is to pour more capital into it. [And]
QoS is going to be more important to large corporations as
they put more important applications on the Internet. Over
the last few years, Internet growth has been from adding
new subscribers and doing fairly simple applications. We
haven't seen the big explosion that I think is still to come
from putting the basic corporate applications [on the
Internet]. Those are just starting to migrate to the Internet.
As that happens, I think you're going to see more and more
QoS guarantees and other applications and products from
ISPs to address all of the concerns.
Gudonis: We are building a new Global Network
Infrastructure (GNI) starting in the United States with a
high-speed fiber network. We are building this new
15,000-mile backbone network in the United States. We
are building a lot of backbone capacity to stay ahead of the
capacity crunch. We acquired dark fiber from Qwest
[Communications Corp.], we bought Nortel [Northern
Telecommunications Inc.] optical electronics, and we are
deploying a network that starts out at OC-192 (2.5Gbps).
That gives us plenty of capacity. Then, with dense wave
division multiplexing, we are going to add even more
capacity.
We are also deploying more Web hosting operation centers
around the world, distributing hosting centers close to
customer concentrations. If you [as a corporate Web
hosting customer] are doing 7x24 global commerce, you
may have customers in Asia and customers in Europe.
Rather than having to deal with traffic bottlenecks [on
overseas Internet connections], we will deploy mirrored sites
for you in data centers in Europe and Asia. So you have a
very short path for your customers to get to your information
and you are going to get higher performance.
Schrader: We buy fiber and instead of adding T-3s
[45Mbps connections], we add OC-48s [622Mbps
connections]. That's a fiftyfold increase, and because of the
efficiencies of the packets, we get a lot more than fifty times
[the bandwidth]. ISPs could probably get two hundred times
more delivery effectiveness by building gigaPOPs, massive
points of presence with large numbers of modems and large
numbers of interconnection equipment.
What you want to think about here is each two or three
years, you see a new innovation that goes on the Internet.
PSINet is blending the computer expertise and the telephony
expertise into one business. So, every two to three years,
you will see a major enhancement in the routing capability
and switching capability, or the use of fiber. Each time you
will see this leapfrog effect of one time, three times, ten
times, a hundred times. Then everything else has to balance
and improve and you get this instability and then it becomes
more stable. Then it gets totally stable and you get this new
enhancement and it gets unstable again. That's part of life
and that's what the Internet is about. That's why the
telephony industry cannot deal with this directly. It must
become an ISP. Have the kind of expertise that PSINet has.
Have the kind of facilities that we have. Well they already
have it--they're telephone companies. What we're doing is
adding telephone facilities to the computer expertise.
LT: What are the trade-offs of being independently
operated as opposed to telco-owned?
Sidgmore: The primary pro [of being an ISP owned by a
telco] is that we've got access to the fiber infrastructure and
that means a lot. It lets us control [network and service]
quality. For example, if you have ownership of the circuitry
all the way through, from local access through the Internet
and out to other cities, you can control the quality better if
you own all of those facilities. You can also package these
services together in more clever ways to create more
innovative products because you can manipulate the facilities
some. The other advantage, of course, is that most
telephone companies have more money than most ISPs, and
money is important to scaling the network.
The downside isn't so much the telephone company; it's
being part of a large company. There is the risk of losing
your entrepreneurial spirit. That's something we worry about
a lot. But we would argue that WorldCom is the world's
largest small company. WorldCom wasn't even around ten
years ago, so it's really a new company. We think we have a
good shot at remaining agile and entrepreneurial because of
that.
If you're independent, you're not going to have the same
kind of access to capital. That isn't so bad when you're a
small ISP-- when you're [worth] ten, fifteen, or twenty
billion dollars. You can always get enough capital to get to
the next generation by doing another secondary offering or
raising venture capital. But when larger independent ISPs
need to scale their network, you're talking about major
amounts of capital. The advantage is you don't get distracted
by other businesses, other markets, or regulatory issues or
things like that. You're instead, very focused on a singular
item. I'm sure that's what Schrader would say. That's what
he tells me every time I talk to him.
Schrader: The pro of being independent is that we control
our destiny. Being independent means that we have freedom
of action. We can go after any market that makes profit for
our shareholders. There's nobody who says, "Schrader,
you've got to stop selling because you're cannibalizing your
parent's revenue." What if we were owned by AT&T and,
after working hard for two years, their revenues drop by 30
percent because we did a good job [selling IP telephony
services] and their fax business is gone because we put it on
the Internet at one one-hundredth the price?
What is the con [of being independent]? Living with the
analysts who wonder, randomly it seems, why we don't sell
the company. There's no reason to sell the company. I think
we're going to be wildly profitable in the future. Why should
we? There's no reason to. That's the con.
Gudonis: [When BBN was independent] we had more
management flexibility because we were a smaller company.
We had less coordination with other operating units because
we had no other operating units to deal with. But we also
had a lack of network facilities, insufficient distribution, and a
smaller R&D budget. [At that time] some of our customers
expressed concern about our staying power and did not
trust more critical, ever larger global IP networking services
to us. They bought a T-1 [1.5Mbps circuit] from us but also
wanted a global virtual private network and managed
content distribution. They were concerned whether we could
step up to that. Now that we are owned by GTE, they trust
that we will be there in the future.
In getting married to a telco, we sought three things:
network, distribution, and capital. GTE has provided all that
and more. On the same day that GTE announced the
acquisition of BBN, they also announced a billion dollar
investment in a new national fiber network. As to
distribution, we got the brand-equity of GTE as a much
larger brand name than BBN had. GTE brought a lot of
voice-telephony skill to mix with our IP and data skill, so
now we are unleashing a new set of IP and PSTN [public
switched telephone network] services--like IP fax, IP
telephony, unified messaging services--that we could not
have done if we were independent.
LT: How will businesses first start using IP telephony
services?
Sidgmore: There are a lot of application-based services that
IP telephony will bring to the market. For example, you're
on somebody's Web site looking at product information and
now you want to talk to a product representative. It would
be very convenient if you could just hit a button and talk to
them, even if the quality was slightly less [than what you
would get on the PSTN]. Another example: Fax traffic is a
huge percentage of what is currently counted as voice traffic.
It comprises something like 50 percent of all international
traffic. Fax [over the Internet] is very easy to do. There are
lots of applications like that, such as voice mail and
messaging and so forth.
Internet voice has become much better, and I think it will be
good enough for a number of applications. But the reality is
that circuit-switched networks are better equipped today to
handle perfect quality interactive voice. If that's what you
want, that will probably be that way for some period of time.
Schrader: There won't be a single traditional telephone
company around in ten years. They will all look like ISPs.
Eighty percent of all telephone calls in five years will be
made over IP and the LAN manager is going to be
responsible for delivery of telephone services. The LAN and
telephony infrastructure teams are going to merge. The
legacy fax and voice and VPN environments are going to be
on the Internet. They're going to be on extranets. They're
going to be on VPNs that are all IP.
Gudonis: What you will see is reduction in long-distance
calling costs. But it should be fairly invisible to corporate
customers because they will buy service from us, and if we
deliver it across our Internet backbone as IP telephony, that
will be invisible to corporate customers. You won't have the
VocalTec [Corp.] method of booting up a PC to have an
online phone call. We are going to add additional enhanced
services such as unified messaging, which is E-mail, voice
mail, and a fax mail box all in one--resident in the network.
You don't have that sort of thing today, but with IP
telephony you will be able to do it.
LT: What hurdles still need to be overcome to make IP
telephony appealing to the vast majority of corporations?
Sidgmore: Last year IP voice was basically people sitting
around shouting into their computers. It was a very poor
situation from a quality standpoint. Today it's much better in
most cases and, yes, occasionally there will be differences in
voice quality. I do not believe circuit-switched voice
networks are going to go away. I think they work just fine
for the majority of core voice traffic and probably will for a
long period of time. I do think quality and the simplicity
needs to improve.
Gudonis: We are one of the few ISPs that has a strong
engineering history with QoS technologies. You will see us
introduce those services that support, not only IP telephony,
but also videoconferencing and robust file-transfer services
over the Internet. But we need scaleable IP-to-PSTN
gateways, and we are working with Cisco in this regard.
It's easier for providers to deliver on-Net [IP voice]
services, where all traffic stays on my backbone or [for
example] the MCI backbone. But crossing that boundary,
the inter-ISP transfer of QoS packets, is something that the
industry has not worked on yet. There have been technical
discussions at the IETF, but there are no business
agreements yet.
Schrader: I don't see any technical limitations at all. Maybe
bandwidth. But we own the bandwidth.
The only place we're going to deliver high-quality voice and
fax and other value-added services, which are extremely
sensitive to latency or quality, is through ISPs who are
directly buying service from us. If you're thinking of using
multiple ISPs linked side-by-side and wondering how
they're going to get QoS across these various networks, it
ain't gonna happen. We're never going to deal with it.
By the way, I don't think simply reducing your [voice
service] costs by a factor of two is enough. You know,
executives talking to executives do not care about how much
phone calls cost. And if your company has ten thousand or a
hundred thousand employees across multiple continents,
then you already have a VPN in place and voice is probably
already costing you only about ten cents [per minute]. So it's
companies between five hundred employees and the Fortune
500. That's where we're going to be playing for multiple
offices doing interconnections for PBXes. |
