Interview Dr. Eric Schmidt
Dr. Eric Schmidt, Sun's chief technology officer, has been
with Sun for 13 years. During that time, he notes, he has
worked with Sybase (and Powersoft) in several capacities.
"When Sun started, we recognized that we were going to
have to work directly with software companies that were
running on top of our platform. At that time, Sybase had
better IBM connectivity and better database-query
performance. So it was a natural fit. Sybase was the
leading vendor in high-performance database-query
applications, particularly in areas such as Wall Street. In
the late eighties, Sun decided to focus on the securities
industry, and we did that hand-in-hand with Sybase.
"Our relationship with Powersoft came much later.
Powersoft is much more PC-centric in approach, but we
were able to convince [Powersoft founder] Mitchell
Kertzman and his team to focus on Unix as a viable server
platform."
Today, Dr. Schmidt is heavily involved with Sun's Internet
efforts, so we focused on the Internet in this interview.
SYBASE MAGAZINE: Sun is obviously a major participant in the Internet phenomenon. If
you were talking to the CEO or the CIO of a company much less advanced in its use of
network computing, what kind of advice would you give?
ERIC SCHMIDT: There's a tendency in our industry to say, "Here's the new thing; let's ignore the
old thing." Customers can't do that. The most important thing to realize is that the Internet and in
particular the Internet protocol [IP] will become a part of the computing environment
information-technology companies offer to their customers. There will be cases where Web
protocols and even Java will become part of the computing solution companies offer. The only
question is how quickly this will happen. I believe that for companies with too many different
computing platforms, early adoption of Java lowers costs and increases customer satisfaction,
because it simplifies things.
Our Java strategy came out of discussions with chief information officers of our largest customers,
many of which are also Sybase and Powersoft users. Their jobs have been likened to changing the
nuts and bolts of an airplane while it's flying. They have to run the business and add technology at
the same time.
They are doing this by putting in intranet front ends on top of existing applications. That's the big
trend this year. I think next year you'll see much more consistent and coherent new applications in
this area.
SM: It sounds as though your advice generally is to start with an intranet front end.
ES: Yes. A lot of executives are asking, "When should I start building an intranet?" My answer is,
I'll bet that your key engineers have already started and they just haven't told you yet.
SM: We hear a lot about intranets. What about "extranets"?
ES: As far as I know, the term extranet was coined by Bob Metcalfe
[developer of Ethernet]. Bob observed that companies were taking
information that was on the intranet and projecting it outside into the
public Internet. In most instances, companies had an existing database
sitting somewhere in the company that was accessible only by a "wizard."
Then, instead of requiring you to call this wizard, they put in a Web front
end. This front end provides the essential answers that a database query
would have asked, but in a very easy-to-use format. It could be a simple
HTML form, for example.
So a customer somewhere in the world points her browser at this page
and types in the appropriate numbers, and an answer appears. Examples of extranet applications
are package tracking, inventory tracking, order status--ways of making customers feel like they
have a closer connection to their suppliers. We think there is going to be a huge explosion in this
type of application. Technically what you're doing is tunneling in HTML. You're letting HTML be
visible outside of the company, and almost all firewalls allow HTML in and out.
To link intranets to the public Internet, you must do it securely. The only way we know of doing this
is with digital signatures for authorized-user authentication.
SM: What's happening with the Web and the Internet is a true paradigm shift, rather than
just the latest evolution of client/server computing?
ES: It certainly is a paradigm shift for the industry. The best analogy I can think of is the arrival of
TCP/IP in its various forms inside companies. Think of it as a tornado. A tornado sweeps away
everything in its path and is destructive. I think TCP/IP is destructive in that it's coming to existing
client/server application users and forcing them to adapt quickly to the way in which TCP/IP
networks link together. Companies such as Sybase are moving quickly to make their applications be
fully TCP/IP- and Web-capable.
Every computer hardware and software company has an Internet strategy, but not all of these
strategies can succeed. I think that the key differentiator for companies will be their use of TCP/IP
as their underlying protocol.
SM: Could you discuss the role of middleware and the Internet?
ES: Middleware may be the last bastion of software that's not dominated by Microsoft, because
middleware is a very broad area of powerful business applications that can improve business
performance. I think we're going to see an explosion of Internet-based middleware. The underlying
assumption here is that the rigid client/server paradigm--essentially the old mainframe model with a
client that talks to the mainframe--is going out the window and being replaced by what is generally
known as n-tier computing. In the n-tier computing model, the data is stored in the appropriate
computer as opposed to the one you're constrained to have it in.
SM: Are there any limits on the number of these tiers?
ES: We're not aware of them. Eventually there are complexity limits.
We have customers that are building n-tier networked computing.
Think of n-tier networking as the next step in data warehousing. Data
warehousing grew up because it was cheaper to put a copy of the
data from the mainframe database into a format that was easily
accessible to users who needed access to the data.
The next step is to have several such places, appropriate to users,
where the data resides. Some data is very close to users, because
the bandwidth is strained. Other data is shipped overnight, because
most people don't need direct access to it during the day. This kind
of flexibility lowers computing costs and gives people more choices
to decide where the company data eventually goes. The downside of
this is complexity. The world was a lot easier when there was only one computer.
SM: Metcalfe has indicated that he thinks the Net is going to come to a halt a couple of
times before it matures. Do you believe that?
ES: He's raised a cry for action, and he's doing a great job in identifying the problems. I think the
problems are solvable before the genuine brownouts occur. He points out that because of various
technical and business reasons, there are times when network availability is seriously in question for
certain parts of the week. And regardless of how fast people try to avert this, it's difficult for them to
add enough capacity to overcome that problem.
There are two solutions. One is to give the network an infinite amount of capacity--more than you
think it will need--and then the hiccups can be dealt with. That's roughly the equivalent of designing
bridges. Bridges have enormous design tolerances to handle big winds and other shocks. That's why
most of them don't fall down. It's the same argument: You overbuild.
The other approach, which I favor, is to work on the payment and scheduling algorithms so that
they're more refined. And you do that by having protocols that can more accurately handle
bandwidth and then charge the users of the bandwidth accordingly. The Internet cannot remain
completely free for all users in the world for the next thirty years. It has underlying communications
costs that someone's going to have to pay for. And as the number of people using the Internet
increases, those costs will have to be allocated accurately.
SM: What about online commerce?
ES: Companies that are selling infrastructure are making money. Customers who are trying to offer
services such as entertainment or information services have not yet shown a different model that's
distinguishable, although many are trying. The debate is whether the right model is essentially a
broadcast model or a subscription model.
SM: Which is more right?
ES: I think it's too early to tell. I am skeptical of the broadcast advertising models. I am much more
comfortable with the narrowcasting models, which are very specific. I say that because I think the
underlying needs are personal. The Web is like everyone having their own radio station. I think that's
going to be the trend rather than mass homogenization.
SM: Sounds like one-to-one marketing. In fact, this issue has an article about this kind of
marketing that mentions Broadvision.
ES: Broadvision is an example of a company that is building a back end to sell to various specific
customers. This kind of technology remembers who you are. When you come back to a site, you
are connected to the correct database solutions in the enterprise to build a long-term customer
relationship. That approach is consistent with narrowcasting. Another example is Pointcast, which
can customize things you receive. Even though they are simulating a broadcast model over the Net,
in practice you can configure what you receive.
SM: What's the latest with Java?
ES: Well, first we announced at Java One last May a whole series of libraries. The momentum of
Java has not slowed; what's happening now is that people are beginning to use it. I think the next big
event will be shipment of Java-based network computers toward the end of this calendar year.
SM: Anything else you'd like to comment about?
ES: The Internet phenomenon isn't going to go away next year. We're in the beginning of a
technological infrastructure revolution involving TCP/IP and other protocols. It will go on for ten to
twenty years, becoming the dominant under-lying networking infrastructure that most of your
readers will use for the rest of their professional lives. So getting on it early gives you an advantage
in competitiveness down the road.
Sybase, Inc. Home | Search | Contact Us | Sybase Site | Powersoft Site | Financials | In The News | Partners
Copyright 1997 c Sybase, Inc. All Rights Reserved. |
