Изображения страниц
PDF
EPUB

Mr. GOLDWATER. Thank you, Mr. Chairman.

Just very briefly. I look forward also to our witnesses today, and I think we need as much input as possible on this new program, this replacement program, to find out if the direction that we are going in is the right direction, and the equipment and the system, the whole program, is the right one.

And so, I think it is important that we hear as many folks as possible. The projected $2.8 billion purchase price-this is no small undertaking, and obviously the efforts of what we are trying to accomplish-the magnitude seems to grow with time. Maybe that is something we need to address: Whether we are trying to do too much, and how can we do the same thing with less.

So, I think it is important that these hearings are being held. To review where we are at—in a way, I was disappointed that the Administrator yesterday did not say more in the direction that the FAA was headed in; but at least he was frank enough to tell us that he needed more time. That point, I think we should follow up. Mr. Chairman, I would join with you in welcoming our witnesses today.

Mr. GLICKMAN. Thank you, Mr. Goldwater.

We have four witnesses today. Mr. Harry Davis had indicated a time problem. I think you need to be somewhere. So if the other witnesses don't mind, we'll try to get you on and off fairly quickly.

Mr. Davis, why don't you proceed to the witness table. Your entire testimony will appear in the record. It appears it is in outline form.

Mr. DAVIS. Yes; I thought it was easier for you to read, in outline form.

STATEMENT OF HARRY I. DAVIS, PRESIDENT, SYSTEMS

REVIEW ASSOCIATES

Mr. DAVIS. First of all, I did answer the questions sent to me by you, but I added another question 8, which shows my background, in which I listed a number of similar systems, since I had the good fortune or misfortune, as the case may be-of witnessing or participating in them. I have participated as a project engineer; sometimes I was a director of many of these systems listed in section 8.

Mr. GLICKMAN. You have testified to the subcommittee this year, before, but I wonder if you would mind just giving a two- or threesentence résumé of yourself for the record.

Mr. DAVIS. Yes; at present, I am a consultant to the aerospace industry on electronics, electronic systems. I have been in the office of the Secretary of the Air Force, Assistant Secretary of Research and Development, also Deputy Secretary of the Air Force.

I was also director of the Rome Air Center development laboratory, one of the leading laboratories engaged in radar, radar systems, traffic control, and air defense, technical command control, intelligence, processing systems—all of which are very computerintensive, software-intensive.

We've done a lot of paving the way for many systems that are going on now.

Mr. GLICKMAN. That's fine.

Mr. DAVIS. Well, the testimony I have just follows the questions.

e justification of the need? I think you have heard ard that yesterday: The obsolescence of the hardware. ers me, perhaps, more than the hardware, is the software gets older, it gets more and more expensive to more and more areas. And that is a very worrisome ustry, it is very difficult to obtain 10 years' life on a stem-each generation of a computer family lasts

s.

er of fact, if you were to buy automatic test equiphat was one of the subjects I last investigated-from vendors today-and they are very software-oriented, er-oriented—you can only get a guarantee that they the software for 4 or 5 years.

. Mr. Chairman?

MAN. Yes.

. Could I interrupt just a minute? . Yes.

N. Could you explain to me briefly-when it comes to I understand generally-between software and hardapplies to computers, and why software maintenance go up? I don't understand that.

is it not-▬

s. We have to realize there are two types of software. re-so-called application program software-has to do do you want the computer to do. And then there is a h of software called support software, and that is big. IBM 360 system-and I cannot quote this authoritativeis all secondhand. IBM spent $3 billion developing the ystem, which is one part of the support software. There ugs" in it which have never been discovered.

NASA's Shuttle. The Shuttle software program-rememuttle was delayed for 2 days-that program was tested ged, you know for months. Finally another bug showed t keeps happening. There are always errors in these nich you have to keep refining and getting out.

intenance of software is getting to be a nightmare, and of the software is so-called operating systems, compilers, itors, debuggers, stuff like that.

NN. But you're saying as they get older, their mainteis proportionately higher than the hardware? IS. Higher; yes.

if one starts a hard technological program, you will find percent of the costs-that is, for the first system-are Of that 70 percent-and you see that in my notes-about t is so-called support software, not the operating systems. you buy a computer from a company, they will furnish oftware, and will maintain it and keep it up to date, give atest versions, for so many years; and then they drop it. am saying is that the IBM 360 system-the support costs be paid. The 9020 computers, which the FAA has, is one of BM 360 system, which is about two or three generations

ystem will not be supported. FAA will have to pay for the of that.

Mr. DUNN. You're talking about companies and guarantees? Is that what you're talking about, basically?

Mr. DAVIS. They will give you guarantees for a limited period of time, 10 years of whatever. Obsolescence is that rapid.

Mr. DUNN. Thank you.

Mr. GLICKMAN. Thank you.

Mr. DAVIS. OK. I think you can scan down the-paragraph 1, the justification of need. That has been presented so many times, I doubt if I should go over it again.

Mr. GLICKMAN. But you'd buy that? You think we do need to substantially modernize?

Mr. DAVIS. They can stretch what they've got. I agree with them. They would stretch it. They have time; but it is going to get increasingly difficult to get spare parts and support software, which they will pay for themselves. This is the obsolescence argument.

Second of all, they do plan to introduce some new capabilities, which I think we have gone over before, the DABS and things like that.

Some day we may get so-called the AERA concept-which I guess you've heard briefed. I'm not sure you have—which would require more computer capacity than they have.

They do have a problem of trying to get more productivity from their controllers. They will not have traffic controllers. They will not be given more traffic controllers, so they will need more computer help to make sure that productivity increases as the traffic increases.

So, it's the obsolescence problems, the growth of traffic problems that requires attention.

Underneath, the new capabilities-which I think are well justified-also drives for replacement.

In the second paragraph, I threw in as a bonus what you don't have to have. It is not necessary they be smaller. It's nice to make them smaller. They don't have to be the latest technological whiz. It's nice to have that, but don't spend money for that. You don't have to have too fancy displays.

Mr. GLICKMAN. You don't have to have Atari design traffic control.

Mr. DAVIS. We can overdo these things.

As far as the minimum systems requirements listed those, too, are quite obvious. The smooth transition is very important. As I say later, I looked at how they planned to do it. It looks sensible. They were aware of the problem. We have had many, many problems introducing new generations of software and hardware when old hardware and software-and transition is always difficult.

I was pleased with what I read. If they execute it as they saidvery good. I think they would have a good chance of not having bad problems. But that is a worry.

Safety, of course, is paramount. A proper distribution of responsibility is considered by the agency. That is always a worrisome thing, and I think they've paid attention to that.

You've asked for alternate ATC system concepts. I volunteered a few things, which I'm not sure you need:

st-you might think of satellite-based air traffic control sysThis has been proposed many times before. It is very attracThey will be expensive to the users. It says "government". is an error in my notes-"all users" should be substituted for ernment".

ansition to a satellite-based system would be difficult, but I hat eventually we may have a transition to a satellite-based

m.

e second type is like the present system, with all of the compus—ground-based systems in a large central computer. There everal different main type-the large main frame type, which have now. They can supplement them with small computers. is called hierarchical architecture.

ere are various forms of redundant architecture, which I bethey probably considered. They have taken a look at what is d distributed architecture, which means that no one computer got the whole load. Any one could break down, and the rest of ystem can fill in for it.

at distributed architecture is a capability that is difficult to eve, but is being worked on and might be available in 2 or 3

5.

section 3, I gave three ideas. I broke down question 3 into two s; also volunteered some comments on satellite-based systems. would be very revolutionary, but I think it would be necesin the distant future.

e large main frame has reliability problems. You have softe problems. You have to be careful about structuring software mat you don't get into difficulty. There are various forms of ndant architecture, and there are pros and cons of various cates, and subtle differences. And you'll hear this afternoon, e pros and cons of various approaches.

nd distributed-which is very esthetically pleasing, but probpremature at this time.

ou asked about interim or fallback measures. I mentioned there some growth left in the computers. You could offload some tions to the minicomputers-that is satellite computers. If you that, you would extend the life of the system only marginally, use of hardware and software obsolescence. And costs would bably make it unattractive.

ou can buy huge "number crunchers" and emulate old maes. I'm not sure you have had the term “emulation". In emulayou can take a big, new, fast machine and use the software for older machine on this machine, by making it look and behave the old. That is a very inefficient way to do things, but it does e the software transition, if you get a better machine, emulate older machine on it, and gradually update the efficiency of the

ware.

is an inefficient way to use the big machine-to emulate an one. I think you will hear more about emulation-I am sure will-this afternoon.

Not all emulation attempts succeed, although there have been he successes. When the first B-1 was built, they ordered a one-kind computer; and when they tried to build three more, that mputer wasn't available. So they bought a bigger computer, emu

82-375 0 81 8

lated the old, so all the software written for the first one is now used on B-1 Nos. 2, 3, and 4, with a completely different computer. That is an example of a successful emulation.

I say that these huge "number crunchers" will avoid the hardware scarce parts problem, but not support software obsolescence problems.

I think this is a very important fact-none of these interim measures face the fact that computers become obsolete in about 10 years. Stretching beyond 10 years is possible, but increasingly difficult and costly. And we know that.

The adequacy of the acquisition plans-I thought they were―― Mr. GLICKMAN. If I may stop you for a minute. There was some discussion yesterday about the propriety of having interim solutions. The Administrator seemed to not be very enthusiastic about that approach. While there may need to be some short-term, interim solutions, he was not very excited about an interim solution to this problem. And from what I gather, you also-

Mr. DAVIS. I don't blame him. There is a lot of hodgepodge right now in the system; and if we had more hodgepodge, we could lose control very easily.

Mr. GLICKMAN. OK.

But you just said-all computers should be phased out in 10 years.

So even if we bought the newest computer equipment tomorrow, in 10 years you'd be saying it's no good?

Mr. DAVIS. That's right.

Let me talk about the strategies. There are two strategies taking that into account. One is to say: "I want 20 years, therefore I'll buy everything, and I buy and pay, through the nose, for someone to keep that software happy for 20 years." That means you will have, the last 10 years, a set of software unique in the world-nobody else will have it-that will be used, and you will be responsiblethat is, FAA-to maintain it. And you'll be paying those costs. Now, the other strategists say: "No. I'll replace the hardware frequently." And since hardware that you buy is generally supported by a company, with the support software-which, as I said, is the biggest cost-you'll be getting the software support for a relatively low cost.

Now, it possible that the second strategy is better than the first, because as time goes on, hardware costs are going down like mad, and software costs are going up like mad.

At this time, it is very difficult for me to say what is the best strategy. Right now, FAA wants to go on the first strategy, and buy one set of computers, and try to stretch it for 20 years. What I am saying to you is, it could very possibly be that it is cheaper to buy new hardware than to preserve software. Mr. GLICKMAN. Go ahead. This is critical.

Mr. DUNN. Run that by me again. One, you buy new hardware every 10 years?

Mr. DAVIS. No. No. 1 is, you say, "I'm going to buy it once and

maintain it for 20 years."

Mr. DUNN. The hardware or the software?

Mr. DAVIS. Both.

Mr. GLICKMAN. The whole package.

[merged small][merged small][merged small][merged small][merged small][merged small][ocr errors][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][ocr errors][merged small][merged small][merged small][merged small][merged small]
« ПредыдущаяПродолжить »