home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Power Tools 1993 November - Disc 2
/
Power Tools Plus (Disc 2 of 2)(November 1993)(HP).iso
/
hotlines
/
cnhl
/
ltaylor.txt
< prev
next >
Wrap
Text File
|
1992-08-10
|
18KB
|
474 lines
Bill Murphy and Lloyd Taylor
Radio HP February 1992
BILL:
Well thanks, Tom, and welcome again to Radio HP. You know we've been
talking for some time about HP's success in very large companies in
establishing real leadership in these firms as they adopt open systems.
We've really been making progress. Big companies are now looking to
Hewlett-Packard to help them adapt and adopt this new paradigm of
computing.
Now the question often comes up however, what about our favorite big
company. Is Hewlett-Packard adopting open systems? Is HP utilizing a
client/server architecture? Does HP use it's own stuff in information
processing?
Well, I thought the best way to address that particular question is to
go right to the horse's mouth. HP's Director of Information
Processing, or Chief Information Officer, if you will, Lloyd Taylor.
And fortunately we have Lloyd here today with us. Lloyd, really glad
to have you here.
LLOYD:
Great to be here, Bill.
BILL:
Now, before we get into issues that I just raised, briefly describe the
IT environment you have within Hewlett-Packard. How many systems do
you have and what does it look like?
LLOYD:
Well, not surprisingly, why it's large. We have 1100 CPUs in our
Business environment, be they HP 3000s or 9000s.
BILL:
So you have 1100 3000s and 9000s.
LLOYD:
That's right. In the business environment, why if you added to that
R&D or on the lower levels of the manufacturing floor, there'd even be
a lot more.
BILL:
What about devices on the desktop?
LLOYD:
Well if you address that generally, we have 97,000 varieties of desktop
computing devices that includes anything ranging from terminals to
workstations and PCs.
BILL:
97,000. So more than one for every man, woman and child in the
corporation.
LLOYD:
That's right.
BILL:
What about network traffic?
LLOYD:
Well right now, we're running about 700 gigabytes a month. And that
number continues to grow. During 1991 that volume nearly doubled.
BILL:
Wow, its a large operation. How many employees are involved in IT
within HP?
LLOYD:
Last count, we had 5500.
BILL:
OK gang, this is a big account we're dealing with here. They have 1100
systems and almost 100,000 desktop devices. Now, Lloyd, as Chief
Information Officer of a large company, or a fortune 29 Chief
Information Officer, what are the key issues as you see it? As you
move ahead, what are the things that are on the top of your list?
LLOYD:
Ok, well Bill, in no particular priority sequence, why not surprisingly
we have a very conventional applications architecture. Sets of
applications that have been developed over the last 10 or even 20
years. And like most other companies, we have very high maintenance
costs.
Secondly, why our business needs, our entire business environment is
changing at a very rapid pace, while at the same time we have very long
application development cycles.
Then thirdly, why we spent the last decade or two accumulating the
warehouses of data on HP equipment and have the natural problem of
getting broader access to that data for decision support.
Then finally like everyone else in HP and I suspect elsewhere, why,
we're under alot of cost pressure.
BILL:
Well, you know, I tell you, I've talked to alot of large companies,
and I'm sure our folks in the field have as well, and those things
sound very familiar. I think alot of people have the same kinds of
problems.
OK, those are the issues that you face, what are you doing about them?
LLOYD:
Well, let me take the four issues that I mentioned then I'll take
those in sequence and make a few comments about each.
In order to address the first two issues that I mentioned, that is, our
relatively high maintenance costs and conventional applications
architecture, and being able to respond more quickly to our rapidly
changing business needs, why, we're migrating as quickly as we can to a
client/server type of architecture.
BILL:
OK
LLOYD:
Let's first address the issue of shorter development cycles. We have
proven through our initial efforts that a client/server architecture,
particularly in our experimentation so far, in a transaction processing
environment, results in shorter development cycles. Why is that true?
Well, first of all there's more structure in that type of architecture.
We're able to develop applications with smaller development teams often
splitting the teams between the client and the server. And finally the
process of system integration is less complex resulting in higher
quality software.
BILL:
Yeah, so you can get the things developed much more quickly in this
kind of environment, and you can also reuse alot of the code, can't
you?
LLOYD:
That's right. We think there's a definite opportunity both now as wel
las in the longer term to have alot of the code modules in our
architecture reusable.
Next, let me say a word about our high maintenance costs. The dream of
all application developers has always been to find someway to separate
the application from the underlying data. And the client/server
architecture naturally provides that long sought after opportunity.
And therefore, why, we're able to perform maintenance on either the
client or the server without impacting the counterpart.
BILL:
You can separate the two, operate on one and it doesn't affect the
other.
LLOYD:
That's right.
BILL:
That's going to be a huge advantage.
LLOYD:
Let me say, by the way we had made certain choices of technology within
our architecture inside HP and those choices arise from the kind of
general notion that I often quote that open systems does not mean open
season. Meaning that we had to make choices inside of HP for our
architecture and not surprisingly why our customers are going to want
to make choices on their own as well. With at least today's
technology, it just isn't practical, even in an open systems
environment, to have one of literally everything.
BILL:
Yeah, so you had to make certain choices on databases, and tools,
application development environments.
LLOYD:
That's absolutely right, and the choices that we made for HP, does not
necessarily mean that those choices are the right choices for our
customers.
BILL:
But I think the key point is, you've got to make some decisions on
these particular issues to adopt this kind of environment.
LLOYD:
Absolutely, I agree
BILL:
So, you've seen some significant advantages in developing new
applications in this environment. I assume that you are developing
many new client/server applications.
LLOYD:
Yeah, Bill. Let me give you some quick examples of that. First of
all, the pioneer in the implementation of client/server architectures
inside of HP has been our customer support organization. With the most
notable example of a deployed or in use application is one that's
called Track X, that's used in all of our response centers, worldwide.
Secondly, we're in the middle of a development process right now to
replace all of our human resources and payroll systems in the United
States, using purchased software that is of a client/server
architecture.
BILL:
So, you're redoing the payroll system?
LLOYD:
That's right.
BILL:
Don't mess that one up, huh?
LLOYD:
That definitely is a bet your career opportunity.
BILL:
OK, what else?
LLOYD:
Then finally, why, we've completely redone our quality management
systems, and it as well is in a client/server architecture and is in
the process of being deployed as we speak.
BILL:
How much of your application development is being done in a
client/server environment today?
LLOYD:
Well, all application development of any significance is done in
client/server environment.
BILL:
So you've really move all new application development into a
client/server environment?
LLOYD:
That's right.
BILL:
Boy, that's a major step. One of the things I've found in talking with
a lot of large companies is they understand the same benefits that
we're articulating here, but they're really concerned about the
implementation. How do you get it done? Do you have to go to the
outside to get it done? To get some help? Can you hire people, can
you retrain people, what's your experience been?
LLOYD:
Well, first of all I think you can do it all internally if you choose
to do so. Our experience so far has been with our people that are a
little newer to HP, let's say people that's graduated from a University
within the last 5 years or so, and have, if you will, therefore a more
recent computer science background, then it's no problem really at all
for them to migrate to understand a client/server type of architecture.
Or even learn new programming languages such as C, for example.
On the other hand, people that's, lets say, have been in the business
for 20 or 25 years and that they're having a tougher time make the
transition.
BILL:
So you've been able to do all this transition to a client/server
environment and develop all these new applications in the client/server
environment really in-house. You've acquired some skills and hired
some people.
LLOYD:
That's right. With the exception of this one project where we
purchased the software.
BILL:
You purchased the software but you're still implementing it in-house.
LLOYD:
That's right, yes.
BILL:
That's interesting. OK back to your third point about having lots of
data.
LLOYD:
Yeah, that's really a tough problem. It's not a technology problem.
There are many tools on the market, whether they come from third
parties or from HP that provide means of accessing data and then
subsequently manipulating it in, lets say a decision support system.
That really isn't the issue. The really tough thing to grapple with is
the underlying data, and like I suspect, in fact, know with other
customers, why we have a data problem internally. Meaning that we have
the same piece of data that carries a different label on different
systems, we have a data integrity issue and so on and it's the cleanup
of that data and putting it into usable databases that can be accessed
by these tools is a real challenge.
BILL:
I would imagine that's a major undertaking that the whole organization
has.
LLOYD:
Yeah, its a real problem, no question about it.
BILL:
I see exactly the same thing with a lot of other large companies where
they have to get standard data models and just move toward that. What
about the fourth point, Cost?
LLOYD:
Well, there's kind of two dimensions to that. First of all there's no
question with advancing technology that we're getting into less
expensive mips, and that certainly has a very positive impact on our
cost structures. But in addition to that, we have undertaken a very
ambitious program of consolidating our data centers over the next two
or three years. In the US for example, we're trying to get from 40
data centers, at least at our last count, down to a number that
probably will turn out to be less than 8. In Europe, we're trying to
get from 20 to 6, and in Intercon, and I mean particularly Asia
Pacific, from today's environment of 7 down to 3 or 4.
BILL:
So you're able to take advantage of the more powerful machines, the
less expensive mips and consolidate data centers, and obviously save
the space, people in different locations, and so on and so forth.
LLOYD:
That's right. And I think we're all familiar with a lot of the
dynamics as these more and more powerful machines have smaller and
smaller footprints and take up less space and so forth.
BILL:
Right. I've talked to a lot of people about that very same notion and
everybody again likes that because it appears to be some very large
savings that are possible but I also hear that yeah, you consolidate
all these data centers but still you have to distribute the information
and access the information so it's possible to loose many of the
benefits in telecommunications costs. Did you take a look at that?
LLOYD:
Well, first of all, I completely agree. That's not a big issue for us
in the US, first of all because the US telecom market is very
competitive and second of all we have a very robust network inside of
HP that's already been deployed. Outside of the US, however, one has
to be careful. By way of an example, a T1 circuit in Europe, in
general cost about 10 times more than a comparable bandwidth in the
United States. So one has to be very careful in these consolidations
that careful evaluations of the undoubtedly incremental and increased
telecommunication expense isn't going to eat up all of the operation
benefits.
BILL:
Yeah, but you still saw enough benefit to consolidate, you said 20 to 6
in Europe as an example.
LLOYD:
That's right, and let me give you an example of that. Just recently I
returned from Europe and while I was there I inaugurated our new
consolidated data center in Brussels that serves all of the
Scandinavian and Benelux countries and still generates positive results
in spite of very significantly increased telecom costs.
BILL:
And I would also anticipate that those telecom cost will come down over
time.
LLOYD:
Oh, I think that's inevitable with 1992 coming upon us in Europe and
what I think would be eventual deregulation of the PTTs in Europe, much
like occurred in US a decade or so.
BILL:
Obviously a major announcement for us over the last couple of weeks
have been the new workstations allowing us to get below the $10,000
price point, actually below the $5000 price point for these new
workstations. You've talked about alot of applications that are using,
obviously desktop devices as clients,have you been able to take
advantage of the new workstation technology in developing these
applications?
LLOYD:
Well, absolutely. By way of an example, and I'm not entirely sure of
the numbers but I don't think I would be exaggerating by saying that we
deployed already about 2000 workstations in a business transaction
processing environment. And continue to do so very aggressively. Why
do we want to do that? Well, there's really four reasons for that.
Our experiments have clearly shown because of the multitasking
capability of the workstation and the larger screen that we're able to
consistently get productivity improvements in the 15% range. But of
equal importance is that the investment to migrate the very traditional
application using the workstation is a very low investment. And then
in addition to that, why having justified the workstation on nearer
term productivity improvements, why we have the platform then already
deployed for our future client/server environment.
BILL:
So you actually can get the benefits today with that environment that
you want to get to tomorrow.
LLOYD:
That's right. And so when the future applications are deployed, why
the major portion of the infrastructure will be already there and would
have been justified a long time ago.
BILL:
So you anticipate that more and more of these desktop devices, the
clients in this client/server environment would be workstations as
opposed to PCs or terminals?
LLOYD:
Well, I think it's going to be mixed. Theres no question about it.
BILL:
Yeah, PCs are certainly not going to go away.
LLOYD:
No question, they're not going to go away.
BILL:
But the lower price point of the workstations and the advantages you
mentioned, the multitasking, the large screens, good graphical user
interfaces will allow you to use those devices in more commercial
applications in addition to technical applications.
LLOYD:
Certainly that's true with todays economics. That always can change
like anything else in the world. But today it looks like a very very
attractive opportunity for us.
BILL:
OK Lloyd, this has been outstanding, and alot of very valuable
information but let me wrap up with a question I think may be on the
minds of alot of folks that we have, they would like to get to talk to
your counterparts, I mean a fortune 29 CIO that has 1100 systems is
somebody we want to be calling on on a regular basis. Sometimes you
and your counterparts are fairly difficult to get to so here's the
scenario: You're in the elevator with a sales rep. and he's got you
for 2,3 or maybe 5 minutes, what would get your attention, what do you
think he should bring up and he should say that would allow you to say
"OK, I'll give you an appointment, come in to see me because it sounds
like you can really help me"
LLOYD:
Well, I think theres two opportunities here, to set the hook, so to
speak. First of all, is the obvious answer. And that is the lower
cost per mip equipment from Hewlett-Packard today. Like I said, that's
the obvious answer and certainly our competitors often use the same
type of argument.
BILL:
A lot of people say that today.
LLOYD:
What I think is the surprising thing is that we've already proven
inside of HP is that the client/server architecture inherently and
independently of case tools results in reduced development times. And
that's a really important issue on the mind of every CIO at least that
I know, and might be a very very interesting way to get his attention
in addition to the lower cost.
BILL:
So, a properly designed and implemented client/server environment will
result in a much more rapid application development environment.
LLOYD:
That certainly has been our experience so far and theres absolutely no
indication that that's going to change in the future.
BILL:
OK gang, if you get a fortune 29 CIO in the elevator, talk about
reduced application development time in addition to reduced cost.
Lloyd, really, thanks alot for taking time out of your busy schedule to
join us. This has been extremely valuable and I'm sure everybody in
the field appreciates it very much.
LLOYD:
Well, it's great to be here and glad to help.
BILL:
Thanks a lot. so there you have it, a lot of good tips from a fortune
29 CIO who's exactly the kind of guy you want to call on. I hope you
found this useful and informative and I look forward to seeing you
again on the next Radio HP.