border="0">Dr Paul Horn, Sr VP,
IBM Research, and Dr Caroline Kovac, VP, IBM Solutions Research, were in India recently to
inaugurate the IBM India Solutions Research Center at IIT, Delhi. Horn has devoted his
life to semiconductor research, while Kovac has been in user organizations before, and,
hence, is heading the solutions research of Big Blue. Being in IBM research is like being
at the cutting edge of technology. Horn was basking in the glory of having attained the 1
GHz mark two years earlier than projected, and having revolutionized the microprocessor
technology with the copper wire technology. And Kovac was excited about the setting up of
the research center in India, whose focus is customer solutions. Horn and Kovac took some
time off from their hectic official visit to India and spoke to DATAQUEST about issues
ranging from ERP and supply chain management to NC and Merced. Also, Horn elaborated on
the focus of research at IBM, which has definitely changed from just technology to
application of technology to real-life customer problems. Excerpts:
IBM is the first one to reach the 1
GHz mark. And it was sooner than expected. What do you attribute the success to? face="Arial" size="2" color="#000000">
Horn: Well, you know that work was done in IBM’s Austin Research Lab. We are
very proud about the accomplishment. I would say that part of this exercise is that we
have been working on not just developing the next generation. But technology...we are
trying to understand how you could go well into the future, to leapfrog–what the rest
of the community doesn’t do. And to push the limits. Which is something what we do in
research. That was an accomplishment which came from our labs and our development
Kovac: I think another
aspect to that is that we have put a tremendous amount of focus on linking our own
research efforts with our development organizations. Not just particularly in the area of
microelectronics, but also in the area of software and solutions; creating linkages
between our research and development teams in IBM; bringing technologies to market;
innovate it further by working jointly with customers, especially in the area of
solutions. Thus, creating applications out of bedrock technologies from our
There are two aspects to
technology. One is the research part and the other is the implementation part. How does
IBM go about implementing the technologies that it develops?
Horn: Let me talk in general.
This is very important problem for us. We can’t be the best research organization, if
we are just good in research and not good in implementation. Then you cannot really
succeed in impacting the world. So we have put in place a number of things, especially in
the last few years. When Lou Gerstner came to IBM, he was a customer of ours and he
remembered as a customer that we always had great research in our laboratories. But he
could never quite get his hands on it as a customer. So, his first priority was ‘how
are you guys going to get this job implemented faster?’ And that led to a whole bunch
of different things, including a much bigger focus on solutions that we are talking about.
Because if you work with the customer, then you work on real problems. Not what the
researcher thinks are problems. The real problems are highlighted by the outside world and
not in the research labs. And, therefore, when you build something, you will build it
exactly matching the real world needs, not just the needs perceived by the researcher. We
believe that we are better at implementing things now, than we were 10 years ago. And now
the focus is on the whole thing, all the way from innovation to how is it going to be
developed, to how do you get it into the hands of the customer. What I would call is
market pull as well as the technology push. We have always been great in technology push,
new ideas bubbling up with new technology. The customer focus or the solutions focus gives
you the market pull. So you really earn the money when you build them.
align="right" hspace="2" border="0">Kovac:
It helps being seen. Because you know that it is the right product that you are making.
IBM has developed so many
cutting-edge technologies, but when it came to championing them, it wasn’t IBM.... face="Arial" size="2" color="#000000">
Horn: Right. Our history has been exactly what you just said. There have been
wonderful innovations–like RISC architecture for example. But in the last couple of
years, we have been much better at it–the copper technology in the back end of the
chip. That has been in our research laboratories for 15 years. Now, we have a two-,
three-year lead on Intel–very strong technology company because it was able to flow
the technology much faster from its organization to the marketplace. This time, we are
well ahead of the rest of the world...we are able to store the amount of information on a
computer chip, enhancing the magnitude of resistance–thus getting the information out
of research to the marketplace almost three years ahead of our nearest competitor has
demonstrated our expertise. These technologies will flow into our products, well ahead of
others. Because we have the key to the flow fairly well. And if we don’t, I
wouldn’t have a job. I am very happy bringing new technology to the marketplace.
Can you describe the difference
between the two perspectives of pure Supply Chain Management (SCM) and SCM with ecommerce? face="Arial" size="2" color="#000000">
Kovac: I think SCM and ecommerce are related. For any organization there are
various stages. Stage one is when it organizes the data and creates an information
repository. This is when it pulls the organization together and has a common repository to
utilities. This is stage of the ERP. Stage two is when it ties up the distribution and
supplier chain. Stage three is of intelligent enterprise, create relationships etc. Then
start commencing transactions–start making orders over the chain. That’s when
ecommerce comes into play. This is quite an opportunity. This is when you start doing
things electronically. We are working with a lot of organizations on this in the US. And
now with the IBM Research Center here in India, we would be doing the same things here as
well. Like for example, EDI on a web-based environment.
But which one comes
first—ERP or SCM? Is there a sequence? color="#000000">
Kovac: You can’t do SCM in isolation. SCM looks at the external side, assuming
that production is taken care to provide for the customer. ERP is pretty much necessary.
Currently, it is the happening area. Coming to the point of EDI, SCM, and ERP...there is
this concept of the network computer...
But which one comes
Horn:The NC, the
NetPC, the Windows terminal...
...as the head of research of IBM,
how do you foresee these things?
Horn: I think the focus really
is network computing. Which is a model for how you develop and deploy applications over
the network to a device, which could be the NC, the traditional PC, a Windows terminal. It
could also be a handheld device, more and more handheld devices are being used, which can
run the applications. So it is about a new model of the way you manage
computing—where you download the key applications over the Internet, intranet, or the
network, and then do the work locally. In some cases, the local device could be an NC. In
some it could be a PC. I think, we are quite hung up in the world, about what the device
is. For all you know, it could be my Palm Pilot. The real question is how you do it.
Whether to hold the applications, the data, the way a traditional PC does, or keep all the
key information somewhere on the network and download it when you need it. You just
replace one end thing with another end thing, and all the stuff comes down from the
network. No one doubts that model. I think that model is very exciting.
If you see history, lot of bubbles
or concepts have come up, for example, it was artificial intelligence 10 years back. It
just disappeared over a period of time. Now it is the network computing, as you define it.
Don’t you think the same would happen to the NC? size="2" color="#000000">
Horn: I don’t think this could possibly happen to network computing. The issue
is not whether it is a PC or it is an NC. What we should be looking at is what the
customers see in this new model. And right now they have a costly, complicated
environment. They have the PC, the servers, and then they have legacy systems. So they are
developing applications in different languages. They got to deal with this very
complicated environment. Instead, you can create a very simple environment in
Java—enterprise-level applications. Download them over the network, and they can be
run on a PC, which has a Java layer or an NC. Java runs on simple clients, which lower the
cost of ownership. You could argue there wouldn’t be a lot of NCs. But the model is
Kovac: Look at the
mainframe environment. It had its own advantages. But it didn’t have the richness of
applications. It certainly has security, connectivity, manageability, and the ability to
control in a centralized way. The distributed model brought with it a lot of great
things—the richness of applications. But it also brought with it a great deal of
complexity. So there is this new model.
But if you see, as systems are
getting more and more complex, building them is getting easier and easier.... face="Arial" size="2" color="#000000">
Horn: Absolutely. They are getting better. But if you see we are getting more
productive. The complexity is in the heterogeneous nature of clients. I don’t know
what the operating systems might be in digital forms, embedded systems...managing these
kind of systems over the network.... Now how are we gonna deal with that? This can be done
in the Java environment. And this is the only one emerging right now to manage this kind
of a heterogeneous environment. Or, in the other case everything is going to be Microsoft,
which is probably the other alternative. There will be Microsoft in the car, Microsoft in
the TV, Microsoft in the handheld etc. I think it is rational to think that everything
won’t be Microsoft. Then the only other alternative for simplicity is Java.
So, Java plays a crucial role
in IBM research?
Horn: It does. I think it plays
a crucial role in providing some level of standards that we can build around in the IT
What do you think of the Intel-HP
combine for the Merced chip, which they are developing?
Horn: Merced, they are claiming,
is a combination of RISC and CISC. They have speculative research done on the concept of
binary translation, which I don’t know will work or not. If it works, it will create
a revolution in the microprocessor world. You can have any microelectronic architecture
you want, but if you want to have a very fine translation layer, that allows you to use
any instruction set, you could go back and get your legacy instruction set like your x86.
Now, if that had existed, then Intel would not have to build Merced. If you had binary
translation, then you could have had any x86 processor and any operating system could have
worked on it. But it might be the next revolution. What they are doing, is that the chip
has got two pieces—one that deals with the old system and another that deals with the
new code. So, the Merced chip is a hybrid. If this had existed, then they could have just
used the x86 chip long back.