ge system-which is not a system of this magnitudes to develop, from concept to acquisition to beginning he system. So that whenever you are talking about major system, you're talking about a long time period. EY. OK. And the 7-year time period, you're saying one period. We can't tolerate using the same equipment. Is

ou're-

TLEBURY. We think it's risky, considering the age of the _nd

EY. Are you suggesting a parallel support to it, both of next on the 1990 today?

TLEBURY. Yes.

EY. And in the meantime change some of the equip

TLEBURY. That's right.

EY. And you can justify the changing of that equipment ngs and efficiencies for plugging in the new hardware? TLEBURY. We think that a cost comparison study of it made. We think it could very well turn out that way. EY. Why time frame do you think it would be necessary tudy, cost study?

IER. Well, they are under direction from the Senate in heir recommendation to have it done by July. They pproximately 3 or 4 years to do it. I think you have oming from Amdahl. We've seen earlier studies that id it is technologically feasible to make a direct replace

question is cost. The main factors are bringing it in without disrupting the major system acquisition or the erations. And those are management issues. And that I what the Senate appropriations and the conference comthe Congress really meant, to make a concrete study. I gest it's a 3- to 6-month study. I wouldn't think that Chat many management considerations that have to be o make a decision like that.

NEY. So, you think between 3 to 6 weeks we can have a done on taking your proposal, plugging in new hardware erim and, at the same time, start the new program for - seems to be the figure we're talking about would be the ach that we could take?

MER. I really would recommend it myself. I think it is a ach. It's good management control.

er point is that some of the information that would be to make that study in 3 to 6 months is already being They are going over the capacity and utilization of the , thus a major underlying study is already underway. CKMAN. Let me ask a question, if you'll yield.

any systems are on the market that will accept 9020

MER. Directly, without modification?

CKMAN. Yes.

MER. That, as I said, would depend on your total needed Amdahl has one line of equipment that goes anywhere o 30 times the size of the existing computers. IBM has two

different model lines, with various ranges of capacities. You ha Magnuson Systems and Control Data Corp. You have The Nation Semi-Conductor/ITEL combination. There are at least five, and I' heard of others who would or might demonstrate that their 32machine will replicate the IBM 360 instruction set with a minim addition of one to four instructions that are required for the app cation programs of the FAA.

So, I would say there's probably on the order of eight or nine. have not made a market survey. It's a simple procedure to make market survey, which is a normal preprocurement procedure. Yo just set up the situation, "I want a computer, of such and such capacity, that replicates the 360 instruction set, provides this much in terms of channel and IO functions," and go out and ask the industry, "How many of you guys would be willing to give me a proposal on that?" Maybe make them put up some money to make it an honest proposal.

I think that's a very straightforward procurement procedure. NASA would do that, in similar circumstances, that is go out and find out what systems capabilities were available.

Mr. GLICKMAN. Mr. Carney, do you have more questions?

Mr. CARNEY. No, Mr. Chairman.

Mr. GLICKMAN. I would just like to thank you.

I still have one concern, that is the issue is still not very well resolved in my mind whether the 9020 is capable of being extended for a period of time. You know, I think that is a key issue-two key issues:

Do we need an ultimate replacement? The answer is, universally, "Yes."

The second issue is: What do we do between now and then? The Administrator seems to believe that we should not go with a direct replacement or interim replacement. We should move more expeditiously on final product, relying, I guess the 9020, with some modifications and with some innovation, would be able to serve that purpose. You apparently do not agree with that, or at least you don't know.

Mr. PALMER. We don't think that we have the facts to really decide that issue.

Mr. SCANTLEBURY. The other thing I would say is that we have looked at a lot of systems. For instance, about 2 years ago or so, maybe 3, I was up testifying with regard to the District Government's accounting system and the fact that they were going to put that system in-in record time. And one of the things I said was, "Don't try to jam that system in fast, because every time you try to put in a system fast, you skip some of the acquisition steps and try to get the system operating without going through the normal, logical, procedural method of getting everything done every time you get a mess."

Well, if you've been reading the papers, I guess you know what happened to the District system. We have seen that so many times, that when you try to shorten up the acquisition process and jam a system in to make it work, you wind up with a mess.

And so, whenever somebody says, "Well, let's skip a few steps and we'll shortcut this procedure," I get very nervous.

KMAN. OK. We thank you for your testimony, and we d to continued contact with you on this subject. witness I don't think is here, Mr. Goldmuntz.

I would ask Mr. Bill Martincic, with Amdahl Corp. to rd and present his testimony.

KMAN. You might briefly let us know for the record, ahl, does, and feel free to summarize your statement, if t appropriate. I have a feeling we are going to get some y soon, so we may have to interrupt your testimony.

T OF BILL MARTINCIC, ACCOUNT EXECUTIVE, CORP., ACCOMPANED BY DENNIS TERRY, CONSULTLD ENGINEER, AND BILL EHRNMAN, PRINCIPAL

ENGINEER

TINCIC. Amdahl was started to directly compete with est scale computing equipment. It was formed by an IBM he name of Dr. Amdahl, who was the senior architect for e design group for designing large scale equipment. As a act, he was part of the design of the 360, 370 family that part of.

to go through our comments and read several of the s for emphasis and then skip over some. I do invite you ot me, if you like, in any one of those paragraphs. nid-1950's, the FAA embarked on implementing a comconcept unheard of at the time. In just 5 years, they accomplishing this innovative and creative system a matter of fact, at that time it was the second largest ackage ever written.

y's standards the hardware now seems archaic and the cumbersome. This is mostly a result of the evolution of y in the industry rather than any oversight in the work performed. Indeed, current ATC reliability is quite good, oday's standards, and the estimated half a billion dollar are does perform as specified.

ore than 10 years later, they are attempting to replace m. The procurement environment the FAA must conform the replacement effort to be long and arduous. The FAA t to deal with any deficiencies in the present system, would dilute the long-term effort.

e think that this represents a two-edged sword to the think that the agency would like to have interim equipcept for the fact that it would prolong the life of the which is something they definitely want to replace. We'll o show that there are short-term solutions to part of these and the interim soluution could lead to a system that rform all of the needed functions for the foreseeable

addressed the questions supplied to us by this subcommiteir order, the first being the need for ATC computer ation. We think that the ATC system needs modernization y to gain the benefits of present equipment. We are sure r testimony will try to justify the modernization based on comings of the presently installed hardware, and we will to that facet here.

Some of the benefits that the FAA could take advantage of as result of intervening industry advances; one, the packaging densi ty. As the package of componentry and computer gets smaller, a more and more circuits are packed on a chip, reliability increases Raw speed and capacity, computers are currently available which perform over 20 times the work per second of an ATC computer and can save internally over 4 times the data. Current available disk and tape devices can contain from 5 to 10 times the amount of data and transfer the data at 4 to 6 times the rate of ATC peripherals.

Current design of mainframes enables repair activity to be deferred to a period of low usage for most of the hardware errors which would require immediate repair on an ATC computer of today. Errors in the data content of main storage are now correctable by the equipment itself and the machine and input and output instructions are able to be retried without software intervention. This means that the need to possess many backup duplicate modules is considerably lessened.

The advances in maintenance technology, current hardware can be fixed using intelligent devices, that is, other computers, to aid the human. Diagnosis of the cause of failures can be performed remotely, thus affecting quicker repair time and mimimizing the need for flying in specialists. Remote diagnosis also means that maintenance centers can perform the more difficult troubleshooting, thus meaning that fewer people have to be trained in depth on the hardware.

Reduction in physical size, current computers occupy the floor space of just two of the ATC components of the 9020 system. This means that a replacement computer could be installed in an ATC center with no disruption to the current environment, leaving the ATC system functional as backup during cutover.

Compatability of instruction sets, many of the witnesses have talked about the 370 instruction set and the 9020 instruction set. In fact, the 9020 does not run a standard 370 instruction set. There are 13 to 15 instructions that are not available by any manufacturer today on standard off-the-shelf equipment.

The ATC computers used the machine instruction set of the IBM 360 and added about 10 percent more instructions. In the meantime, the base instruction set has become a de facto industry standard, with many vendors manufacturing mainframes that employ it and hundreds of billions of dollars invested by users in programs that depend on it.

My company estimates about $300 billion of software currently running on this industry standard. In our view, the 10-percent additional ATC instructions, only one instruction is commonly used, and that can be changed by recompiling the programs. Most of the rest of the ATC instruction additions are involved with ATC hardware and recovery schemes which are of questionable value using new hardware.

It is estimated that over 8,000 man-years have been expended in getting the first ATC system software running and in maintaining and enhancing it. Since the instructions have been so compatible, this running and proven software can be used as a bridge to the system of tomorrow with a minimum of conversion.

e

T

0

some questions: If, in fact, replacement is to occur 5 to n now, how can FAA avoid maintaining and enhancing pability without using an interim system? How would dress features such as DABS, conflict processors VFR dynamic simulation, et cetera, without an interim

se to the question on the minimum system requireook a tack to show how a method of going to an interim d be done. As stated in our last section we feel that the C software has a lot of utility left in it, provided that it by running on current hardware. And we feel that the on could occur in the following way:

software could be examined to see which of the incomcan be easily removed. A trap and emulate program written to perform the instructions and features which - and risky to remove.

hardware should be phased into the heavily loaded ATC un the "patched" ATC software, and the ATC instruc

or.

peripheral devices should be examined and replaced as

work should be done to the ATC software until it is mpatible. That is to say, get rid of those 15 extraordictions, and then the emulator could be discarded. lt of the above efforts will be the following benefits to

system will be in a position to upgrade any installation, vendor's hardware that runs the industry standard inet, at any time in the foreseeable future.

A will have gained all the advances that have occurred TC system's conception.

AA so chooses, they will have the option and capacity to the follow-on system on this standard hardware.

action is required before the follow-on system is availaAA will have the ability to upgrade any installation need is justified.

d and definition of fall back measures: Currently manuoffer families of computers from small to large. The ilosophy currently in use in ATC centers is to fall back in a fail-soft mode and then go back to broadband radar communication.

k this is sound, although in our conversations with cone find that only the very experienced controllers possess -to resort to manual control. We suggest that since many upport the 370 instruction set with families of equipment sized models, a system dedicated to back up could easily bed implementing only the functions critical to immediate aration using a smaller member of the same family. ey of cost estimates: Transferring ATC software to new can be implemented with minimal disruption in the prescenters, and no construction or remodeling will be reersonnel retraining is limited. In addition, the overall goal mple: to make software from a current ATC complex run ate-of-the-art hardware. With an assumed good working