5/12/98 Jubak/Microsoft Investor article. [Very positive on ASND and FORE. QoS as main theme from H&Q conference}
On the road to Internet reliability
investor.msn.com
Posted 5/12/98.
The "pipeline" companies are still rolling.
But new opportunities are just around the
bend in quality-of-service companies like
Ascend, Fore and EMC.
By Jim Jubak
They should have posted warnings signs:
Danger, Metaphors at Work. After sitting
through a dozen presentations at the
recent Hambrecht & Quist technology
conference from companies selling the
gear that makes the Internet work, I
predict a switch from the pipeline image
to the QOS metaphor. (That's quality of
service, for English speakers.)
Translation: By next year we'll all be
talking a lot more about companies that
can provide efficient, reliable service on
the Internet than we will about those
building the supply lines. And I think that
means you should be thinking about
adding shares of companies like Ascend
Communications (ASND), Fore Systems
(FORE), and EMC (EMC) to your
portfolio.
The currently popular pipeline metaphor
works like this: Traffic on the Internet is
doubling about once every 3.6 months.
That means there's an insatiable demand
for capacity -- hence, buy companies that
are building new systems or that can add
capacity to a big, up-to-date Internet
network that they already own. Qwest
(QWST) is an example of the first type of
company, and WorldCom (WCOM),
especially after the MCI (MCIC) merger is
completed, is an example of the latter.
Thinking about the Internet this way,
investors have also argued the relative
merits of telephone and cable companies.
Which will own the pipeline bringing the
Internet to the home?
I'm not arguing that you should consign
that way of thinking to the slag heap of
history -- after all, the fact that Internet
traffic is growing about 1,000% a year is a
pretty strong foundation for an investment
thesis. But the growth of traffic through
that pipeline -- and especially the
increase in voice phone traffic -- has
started to reveal a new set of challenges
facing the Internet. And those challenges
spell opportunities for Internet equipment
makers and for investors.
Here's the problem: The Internet isn't very
reliable, especially in comparison to the
phone network, due to the basic character
of the network as built. If you use the
Internet now, you've experienced the
difference even if you don't know why it
exists.
We rightly assume that we'll get a dial
tone every time we pick up the phone.
Phone company equipment -- from local
loop to central switch -- is supposed to be
available 99.999% of the time. And it is.
The Internet, on the other hand, drives
users crazy with messages like "Server
not found" and unexplained busy signals.
That's because the Internet was never
intended for high-volume, real-time
communication, even though that's
exactly what we're now asking the
network to provide.
From its beginning as a Department of
Defense research project, the Internet
was designed not to eliminate mistakes,
but to correct for them. Messages were
broken into digital packets and then sent
over a reasonably reliable connection.
Error-detection software checked to see if
all the packets in a message were
received and if not, requested that the
information be repeated. I've seen
estimates that packet loss at major
network exchange nodes run between
10% and 30%.
Here's the problem:
The Internet isn't
very reliable,
especially in
comparison to the
phone network, due
to the basic
character of the
network as built.
The Internet's designers didn't worry too
much about a couple of other problems,
either, basically because they really
weren't problems as long as traffic on the
Internet was relatively low.
First, the Internet's basic design makes it
hard to prioritize transmissions -- all
packets try to crowd onto the network at
once. When the network is overloaded, it
simply bumps some packets, sending
them back for retransmission, no matter
what the content or the source or the
time-sensitivity. That, of course, actually
increases traffic.
Second, while the Internet's switches are
great at finding a route to get packets to
and from their proper destination, they're
not well suited to figuring out what the
least expensive source for information -- in
both distance and network time -- might
be. When I ask the Internet for a Web
page, the network sends me to the
specific server that I've asked for, even
though the server I specified might be
busy, for example. Contrast that to a
network that knows how to route my
request to the most efficient source of
that page -- perhaps a local data
warehouse that has duplicated the page
because so many users in my area are
asking for it.
Without a
guarantee that a
packet will arrive
on schedule, the
Internet is simply
not a viable tool for
many business
purposes.
All these characteristics of the Internet
make it extremely hard for a service
provider to give a customer any
meaningful guarantee of quality. You and I
may decide that it really doesn't matter if
that e-mail with the attached photos of the
kids gets delivered to grandma in
Spokane in the next hour or in the next
day. But businesses care. Without a
guarantee that a packet will arrive on
schedule, the Internet is simply not a
viable tool for many business purposes.
Delivering services of a specified and
guaranteed quality on the Internet is going
to be extremely difficult. We already know
that data traffic, the kind that dominates
the Internet, is much less predictable than
voice traffic. So the tools and formulas
that the phone companies have developed
over the decades to manage traffic on
their systems aren't going to be very
much help. For example, data traffic,
unlike phone traffic, doesn't smooth out
over time, it now appears. A traffic flow
that produces predictable peaks when
averaged over a day, week or month will
usually conceal random spikes big
enough to crash a system built to handle
"average" peaks.
That doesn't discourage me. In fact, as an
investor I like the combination of high
stakes -- business demand for
high-quality Internet service is immense --
and tough solutions. It should guarantee
high profits for any company that can
provide even a partial and temporary
solution. Right now I think there are two
approaches that attack the quality of
service problem and three companies that
you ought to consider as you build your
Internet portfolio.
Details
Company Facts
1-yr Chart
* Earnings Estimates
Company Facts
1-yr Chart
* Earnings Estimates
*Consensus EPS Trend
Approach #1: Add intelligence to the
system. That means adding software to
Internet switches so Internet service
providers can better track and manage
their systems. The hot product here is the
ATM (asynchronous transfer mode)
switch. How hot? Lucent Technologies
(LU) just paid $1 billion to buy Yurie
Systems, an ATM company with $51
million in sales. In another recent deal,
Qwest bought routers from Cisco
Systems (CSCO) and ATM switches from
Ascend that incorporate the company's
Navis customer network-management
system. The joint press release on the
deal noted that the technology would give
Qwest the ability to sell data services with
service-level guarantees to its customers.
Ascend is clearly one beneficiary of the
ATM trend. Listening to the conference
call with analysts after the company
announced quarterly earnings, I got the
strong impression that its ATM business
would pick up steam in the next few
quarters. I believe the strength of the ATM
line is the primarily reason that analysts
have steadily upgraded the company in
the last month. (Ascend is currently a
Jubak's Pick with a price target of $55.)
The problem is that since the ATM switch
business is less than half of Ascend's
revenues, the company isn't a pure play
on adding intelligence to the system. If
you want a lot more pop (and somewhat
more risk), take a look at Fore Systems,
a past highflier that crashed last June and
has been slowly recovering. Analysts
have recently started to upgrade the
company, and last quarter Fore brought
several new ATM products to market. The
stock has good momentum -- three-month
relative strength is 88, well above the
12-month reading of 65.
(Relative strength ranks a stock's price
appreciation against all other stocks in
the market. A relative strength of 88
means that the stock has outperformed
88% of all other equities in the period.)
Details
Company Facts
1-yr Chart
* Earnings Estimates
*Consensus EPS Trend
* Growth Rates
EMC is positioned to
benefit from two
very hot trends.
Approach #2: Add storage to the system.
One way to reduce traffic, to speed up
connections, and to deliver better service
is to duplicate frequently requested
material on local servers -- either those at
an Internet service provider or at a
customer company's own place of
business. A customer in Washington who
wants a video clip of a frequently
requested national event would download
that material from a local satellite site that
had itself downloaded the material in
response to local demand and then
cached it on a hard drive. Each customer
who wanted the clip would thus get it
more quickly from the local source
without having to wait on a remote server.
Building such a system obviously requires
sophisticated software to manage
requests and downloads -- and lots of
storage. Here I'd pick EMC as my horse
to ride. The company controls about 50%
of the market for storage devices
connected to mainframe computers and
also owns the biggest piece of the market
for company-wide computer storage. That
gives EMC plenty of experience with
creating and storing databases and then
distributing them to users over a network.
At the Hambrecht & Quist conference, the
company identified the Internet as one of
the biggest potential drivers of its future
growth. (And I certainly don't mind that
the company is getting an increasing
percentage of its revenues from software
that manages how information is stored
and distributed. Software margins are
substantially higher than the margins for
hardware.) EMC trades at 39 times
earnings of $1.10 a share for the last 12
months, but that multiple is just about
equal to the company's recent earnings
growth rate of about 35%.
Now, you might object that EMC isn't a
pure play on the Internet. Won't argue
with you there. The company has the
serious disadvantage of being the leader
in providing storage to run with application
databases from Baan (BAANF),
PeopleSoft (PSFT) and SAP (SAPHY).
So shoot me. The company's positioned
to benefit from two very hot trends. Sorry.
Updates
New Developments on Past Columns
Details
Company Facts
1-yr Chart
* Earnings Estimates
Catching the Cisco Express
On May 5, Cisco Systems (CSCO)
reported about as perfect a quarter as I
could have hoped for. Sales grew by 33%
over the same quarter last year. The
industry as a whole may be showing
slower growth, but the industry leader
isn't. Net income -- before a $419 million
write-off for two acquisitions -- climbed
35%. Gross margin actually climbed to
65.7% in the quarter, up from 65.3% a
year earlier. That margin increase is
especially impressive since Cisco's
product mix in the recent quarter included
relatively fewer high-margin routers and
relatively more lower-margin switches.
Credit CEO John Chambers, who keeps
managing to find ways to wring costs out
of the business. I'm setting a December
target price for Cisco of $88 a share. I
added the stock to Jubak's Picks on
September 12, 1997, at $48.
Terms of Use, and Privacy Policy. c 1996-98 Microsoft Corporation and/or its
suppliers. All rights reserved.
Quotes supplied by Standard & Poor's ComStock, Inc. and are
delayed at least 20 minutes.
NYSE, AMEX, and NASDAQ index data are provided real time.
Investor's editorial goal is to provide a forum for investment ideas. Our articles,
columns, and other features should not be construed as investment advice, nor
does their appearance imply an endorsement by Microsoft of any specific security
or trading strategy. An investor's best course of action must be based on individual
circumstances. |
