The average Supply Corps officer can tell you a lot—perhaps more than you would care to know—about the Uniform Automatic Data Processing System for Navy Stock Points, hereinafter referred to as UADPS. Ask him, for example, what the system is and he might describe it as “a flexible but uniform set of Department of Defense constraints, within which the Navy has successfully devised and executed a plan that uses medium and small-scale computers at each of the nine major Stock Points.” By the time you have thoughtfully considered the answer you may have forgotten what the question was.
Stripped of all esoteric verbiage, what is the system? It is a method of assuring that the U. S. Navy does not suffocate beneath what threatened to become an ever-growing mountain of its own unfilled requisitions.
The method was conceived and its specifications were drawn up in the spring of 1961. By 1964, the Navy’s nine major Stock Points —four Supply Centers on each coast, one at Pearl Harbor—were using a high-speed electronic digital computer together with data storage devices and remote terminal equipment to service their customers.
Having said that it is a method, let us editorialize and add that it was and remains, over-all, an excellent method. To be sure, it is an overtaxed method; and there are other apt adjectives—“pioneer” and “limited” are two.
Those who conceived the present UADPS anticipated many if not most of its future limitations and, to the extent that the state of the Automatic Data Processing art permitted, forestalled a number of seemingly insurmountable problems both before and during the implementation period, 1961-1964. But ADP, as was noted by Mr. Keith G. Taylor in the July 1965 Supply Newsletter, “is a field that is constantly being eroded and refashioned by change. New hardware is obsolete by the time it is installed, and new systems are over-taxed as soon as they are placed in operation.”
Four flaws seem to have crept into the present UADPS system. The system has no:
• Back-up computer capacity in the event of a massive surge of demand on the system, as in the event of a national emergency.
• Back-up methods for processing transactions in the event of on-site computer inoperability, due to national disaster, fire, sabotage, or armed conflict.
• Immediate access to pertinent management information for both the Naval Supply Systems Command and the cognizant stock point.
• Awareness of the computer’s potential to compute, correlate, predict, and display. Stated another way, there seems to be little recognition of the computer’s latent ability to do tasks other than bookkeeping.
In peacetime, the present system might have limped along and “made do.” But this is not peacetime and the pell-mell pace of a runaway technology is making hitherto undreamed-of demands. And these demands now exceed the capability of the installed hardware. As a result, we are forced into more costly and less efficient use of the machines to find solutions to this problem on the one hand and are constrained on the other hand by the number of hours in any given day.
Obviously, the present UADPS is infinitely superior to a system wherein the stock points are manned by masses of people massaging papers and making routine decisions. But an indication of how much more remains to be done can be seen by noting how little we presently dare ask of it. The basic and rigid information retrieval scheme (basically this is manual or punch card processing, only faster) allows only a small set of predetermined requests for information to be submitted to the computer for processing.
How, then, can the computer enhance the support of stock point logistics? Of the many ways, let us isolate one proposal and, since it is our own personal dream, let us not stint. Let us think in terms of one of the biggest and best of the computers. For our proposal would not be possible without the recent successful advent of the very large, very high-speed computer. And, in borrowing this piece of hardware, we must also borrow its attendant “software” of sophisticated control and executive programs (see glossary, page 52).
To be sure, there may be bugs in our computer; but we shall put up with them to get its advanced input/output capability, its very high internal speed, and its ability to work concurrently on many disjointed problems. These machines, which measure their instruction execution speed in billionths of a second, and which are capable of simultaneous execution of many programs, have the capacity to far outstrip the collective capabilities of the present nine uniform installations. It appears reasonable, then, to:
• Install a very high-speed computer configuration, with extensive files which can be addressed directly at any point (random access files), at a site geographically selected both for physical protection and communication facility.
• Provide each stock point with a small- scale computer to process one-time or locally determined requirements of a small-scale nature.
• Provide high-speed communication from each stock point to the large computer complex for the Uniform System requirements, and also for processing large-scale unique requirements.
• Provide small-scale cathode ray tube display devices for both stock point and command-level interrogation of the management and operational information on file at the central complex. These devices are similar to some airline reservation inquiry devices.
• Provide redundancy in the central complex and alternatives in the communications links as a basic requirement. Redundancy, as applied at vital computer installations, is the duplication of both hardware and processing so that there is no instance in which the system is down due to equipment failure.
This proposal obviously would be very similar to many command and control and time-sharing configurations. A logistics counterpart, however, does not appear to be infeasible and, in fact, now appears to be a pressing requirement. A basic tenet must be emphasized in the evaluation of this proposal: Not only must we question any automation that interferes with or delays an operational mission, but also automation that does not actively support a basic mission.
Preliminary estimates indicate that the cost of the configuration may well be below the hardware cost of the present Uniform System, even with redundancy as a prerequisite. Initial fixed cost will be high, with site preparation and communication links among the considerations. But in view of the period of product stability which the computer industry appears to be entering and since this large- scale type of equipment is typically not quickly outmoded, a longer amortization period should prove attractive if purchase is to be considered.
A system of this type has tremendous reserve capacity, and can smooth the fluctuating levels of activity occasioned by unpredictable conditions; i.e., the effects of the Vietnam crisis on West Coast activities, and the influence of the Dominican Republic and Cuban crises upon East Coast activities. For both normal and abnormal operating conditions, time zone fluctuations will handle themselves quite nicely.
For example, there will be some two or three hours when East Coast, West Coast, and Hawaii all will be making demands during their respective normal working hours, as a high point in activity, and three hours when only Hawaii will be on, as a low activity point. Future advance base use of this type of complex will reinforce this interesting characteristic and make its usefulness even more apparent. In the unlikely event that a very large processing problem with a high priority commitment is of such a magnitude that it taxes the system, priority processing can subordinate or defer low priority tasks and thereby bring the full force of the complex to bear on this one urgent problem. Automatic priority scheduling by computer executive programs is a well-documented and proven capability in large systems. In a minor sense, priority pre-emption is available in the present system whereby a limited priority handling of transactions is automated, but this is at most a primitive arrangement at present and useable only under certain strict operational installation constraints.
This consolidation of processing at one location will facilitate new and extended remote operations as a system concept; i.e., area support agreements, or the addition of new activities to the basic Uniform System. In the latter case, the capitalization of new activities entering the Uniform System will be drastically reduced from the present cost of installing and maintaining individual computer systems on site. In the former case, an extension of an already operational, remote net work will be a logically well-defined and manageable problem. This is not to say that there will be no complications in programming for this new concept of uniform remote applications, but the problems will be at least confined within physically and administratively manageable limits. There is considerable merit in this line of reasoning, since not only are system changes then within the confines of a central organization, but also the accountability for the response and the operability of the system can be isolated. The question of whether a local processing problem is generated by operator error, equipment error, or program error will now be a function of one location and one organization. Also, heretofore unavailable information on how well the individual activity uses the computer will be measurable and available as one indication of efficient use and management of resources. In addition, the availability of previously unknown and undefined system statistics will serve both to identify and correct irregularities before they reach a crisis proportion.
A very significant advantage of this centralized processing scheme is in relation to the stock point—Inventory Control Point (ICP) dialogue; that is, the retail-wholesale communication and control. There will now be a focal point and a better facility for system- wide improvement, such as the gradual trend toward shifting functions out of stock points and into inventory control points. The immense capacity of such a complex is well adaptable to absorbing new functions and new tasks in its stride, thereby providing an initial giant step forward for any future consolidation-of-function plans and eventual evolvement towards integrating Navy-wide supply and fiscal operations into a total logistics management concept. For the present, a single point of contact for the ICPs appears to have some very beneficial overtones, especially in regard to decreasing the investment in stock levels in the system while at the same time increasing responsiveness.
There are a number of other system benefits to be gained in this new environment— some large, some not so large; but in sum total they represent a significant favorable argument. Some of these benefits are technical and are in such areas as programming, hardware operation, and program maintenance. Some of them are measurable, such as the cost advantage of processing on the main computer at magnetic tape or telecommunication line transmission rates versus punch-card speed, which presently limits many of the existent programs, i.e., electronic speeds would be substituted for the much slower electro-mechanical speeds. The cost and convenience of reduced punch card handling also cannot be overlooked.
Some benefits are measurable only in part, such as the advantage of each stock point having virtually limitless use of a very large computer for the price of a medium-scale computer on site. Further, the door is opened for many other interests, such as the ability to prepare, uniformly and centrally, many of the routine reports which plague stock points, and to have all such reports prepared with the same data base and the same correlation or calculation procedure.
The welfare of the logistics system as a whole cannot be subordinated to the requirements of an individual component of the system; however, there should be a recognition of the local problems generated (or solved) by this central complex. The objections generated in the interest of command prerogative are, of course, a serious first question of any such centralization of operations or functions which traditionally have come under the direct control of a commanding officer. The psychological and physical degradation of the individual Data Processing Departments is one of transitory nature since the removal of the present equipment will be replaced by immediate on-line access to a much larger capacity system, but is nonetheless of concern. In addition to these two objections there could be added the question of system and stock point response during transition to the proposed central complex and the attendant frustration of having the operator of the equipment out of physical reach of each local command. Some of the above points reflect policy considerations, some are concerned with operational considerations—and all deserve a full and serious discussion; however, they do not imply a peremptory rejection of the proposed schema.
Tangible returns to the stock points for this investment—and, incidentally, spin-off returns to the supply system itself—accrue in the area of stock point transaction processing and in the area of local management information. Lost processing time at the stock points for hardware failures will be virtually nonexistent due to the redundancy at the central complex and due to the inherent ability of these very large machines to operate in what is known as “crippled mode”; that is, with one or more of the modular units inoperative. Of further advantage is the decreased dependency of the equipment on the computer console operator. Operator error is presently a measurable contributor to processing time at the stock points. With a very large complex, many operator functions are handled by a programmed executive routine, and the operators themselves are of a high technician caliber. This in itself should contribute materially to increased throughput for each stock point.
Of no less significance is the local flexibility available. Desk-size display and inquiry units will provide, on request, instantaneous status presentations of a wide variety. Within a sophisticated information retrieval framework, the activity is not only allowed to ask the expected, but also those questions which are not expected. That is to say, there is no rigid set of predetermined requests for information, although the available data pool itself may be restricted. The data available for display of management information and trends will be continuously updated for each stock point as transactions are processed, trend predictions and comparisons with previous period made, and bottlenecks located. This is a recognition of the long-neglected subject area of internal control at the local command level. Heretofore, the automated system emphasis has been very heavily weighted in favor of stock processing (and rightly so), but to the detriment of feedback and internal control at the stock point. Availability and ease of access to pertinent data will materially assist in the control of work in process at each activity. Data on manpower, backlogs, and work scheduling are vital, of course, to responsible and responsive administrators and can be denied them no longer, whatever the rationale.
Any effective information processing system should improve the level of performance of the user.
Local, small-scale, unique applications are well within the scope of the computer envisioned for on-site use, and large-scale unique applications (such as the advanced and complicated Automated Material Handling System at NSC Oakland) will be much more manageable on a very large machine than on the present medium-scale business computer. All things considered, this wedding of advanced computer concepts and stock point requirements can be expected to produce recognizable benefits within the stock points and to provide a significant beginning for a truly integrated supply system in which the administrators can react quickly to changing local requirements. In this way, the individuality of purpose and independence of action at the respective stock point commands are enhanced by, rather than subordinated to, centralization.
The presently installed equipment is of a different configuration at each stock point. This was necessitated by differences in the volume of transactions and by the operational commitments at each stock point, and further required by the general upgrading of the system as it was developed and installed at various locations. From a digital computer programmer’s point of view, this presents a great number of technical problems since any program which he writes must run on all configurations in the Uniform System. This, of course, leads to the obvious comment that a single computer installation would be much easier to program for, the program would be easier, to test and “debug,” and, by implication, it would be easier both to find a problem once one develops and to solve it once it is found. Also, in the proposed consolidated site scheme, these same Uniform System programmers would not interfere with the stock point Data Processing Department schedule when a program is ready for testing or debugging, as is the case at present. At those locations which are processing at near capacity, this is a serious consideration. These programmers would have their own input/ output device, at the present central maintenance locations, for the same immediate access and response that the stock point itself receives.
At present, a serious escalation in the volume of transactions at any one stock point may place it in the position of having to decide what not to process due to the capacity of the machine being exceeded. This is particularly true at the large stock points. However, a 100 per cent increase in the number of transactions at NSC Oakland, for example, under the proposed remote system would represent a possible 20 per cent increase of total demand on the central complex; and this would be smoothed over inactive time periods at other locations, as mentioned previously.
As always, the redundant computer would be standing by to handle the unprecedented happening, the complete failure of the primary computer, or just to take over while preventive maintenance is being performed on the primary computer. This maintenance, by the way, can be expected to be of very highest quality and resident at the site, in contradistinction to the various degrees of maintenance expertise and response which are experienced by the present nine activities.
With this one location, there is an equipment and programming facility for switchover to future generation equipment for interfaces with DOD and, more importantly, for centralized processing of stock point and inventory control point programs at one location. Supply Systems Command decisions which are equipment oriented now need only to reference this one location and this one configuration. A final consideration in the consolidation of major equipment to one site is that the physical security of our vital logistics control network will be greatly enhanced, provided, of course, that the previously mentioned redundancies and their ancillary protective devices (such as separate power supplies) are also provided.
The presumption that this complex be located at one physical site (as opposed to multiple installations) also presumes that this site must be physically detached from any operating stock point. The exact location is not suggested here, but it may well be governed by available communications, security, and electric power facilities. In the event of a national emergency, there is the possibility that destruction or sabotage in the area within, or contiguous to, any stock point may well damage or destroy its computer, its stored files, and, consequently, much of the possibility for recovery or back-up. In the also-likely event of a natural disaster (an earthquake at Oakland or a hurricane at Norfolk come to mind), or a fire or destruction of another origin, the local computer installation is just as vulnerable as it is in the case of willful destruction. Other types of breakdowns, such as massive equipment failure on the computer itself or on the stored files, or a power failure, or an annoying and disruptive air conditioning shutdown can and do stop processing for a matter of hours or days at various locations in the present system. It is apparent that the possibility of such occurrences at a redundant central complex is very unlikely. If such an unhappy event were to transpire at a stock point under the proposed central system, the files would, of course, remain intact at the central complex; and, at most, a remote station would be inoperative until it could be replaced or until the stock point was again operational. There is an intuitive appeal, and indeed a sound reason, in the belief that it is not only easier but less expensive to replace remote devices than it is to replace entire computers. Centralized switching of demands from inoperative points to operative locations would be a relatively simple matter in this instance and pre-programmed for this contingency.
Figure 1 shows a time-phased implementation of the proposed central computer complex. This is a general schematic and is not intended to show all interdependencies. There are two features which are considered essential to the successful implementation of this complex. The first requirement is the formation of an independent, nucleus study group composed of approximately five persons of unusual and recognized ability in the following areas:
• commissioned officer, group leader and specialist in Uniform Automatic Data Processing, programming, analysis and hardware problems and techniques.
• specialist in software, mathematical optimization techniques and hardware.
• specialist in applied programming for the Uniform System.
• specialist in Uniform System analysis for stock points
• specialist in inventory control points.
In addition, the usual secretarial and administrative facilities must be available as needed. The group mission is to act as a professional data base for reference in later phases and to ensure proper consideration and co-ordination of all technical aspects of the schema. This group is to study command and control, university, commercial, and scientific installations where remote processing, timesharing, or very large scale computers are installed and working, and where the professional stature of the installation would make it of interest from any point of view to a uniform logistics system. Equipment, processing techniques, software, communications and site would be of special interest. While all available indications point to the technical feasibility of the project—from the hardware, the system design, and the communications standpoint—these aspects must have a thorough verification by the group during this phase. Subsequent to a six-month preliminary distillation of this very complicated subject area, site and equipment specifications and selection could begin, using this nucleus group as a point of reference. It should be pointed out that this nucleus group must be given the widest latitude in confidence and travel support in order not to penalize the project from the outset.
At a point one year from the formation of the nucleus group, the second essential feature is apparent from Figure 1. Nothing has been pushed so far that retreat is impossible. The cost up to this point is minimal to the Navy and to the present Uniform System. If the site cost and the equipment cost are found to be prohibitive or the project determined to be technically not feasible for some previously undiscovered reason, very little is lost by abandoning the project, and very much is gained for the Navy by the study and recommendations of the nucleus group. The question of whether or not to commit after this point becomes more and more irreversible as time and resources cascade during the implementation of the programs. Up to this point, however, premature specification has been carefully avoided, which will materially reduce the possibility of costly late changes.
The task is obviously long-range and cannot be implemented with the idea of converting within a year or whenever the impending glut on the stock point computer installations becomes a crisis. This implies a certain stability in organization, planning, and career rotation (where military personnel are involved), and a purposeful and scientific approach befitting the complexity of the electronics and the processing which are involved.
Certain implications in the present organization of centralized programming, analysis, and maintenance of Uniform System programs are apparent. First, the formation of the nucleus group will, and should, drain at least four of the very best people from the present central maintenance organization. The site and equipment selection process will do so also, but to a lesser extent, as this is clearly a Supply Systems Command level responsibility. Secondly, at some point after the analysis and programming for the new system get under way, the present Uniform System will have to go into a purely maintenance status as the analysts and programmers devote full time to the subject of program and system revision in order to take full advantage of the capabilities of the machine and of this different concept of data processing. The solution to the question of supply system response just prior to and during the transition could well be a small increase in staffing and a large amount of higher level co-ordination and planning for this period. All system changes must be known and planned for in advance. DOD revisions (MILSTRIP, MILSTEP, MILSTAMP, et at.) are one point of conflict here, but intra-Navy co-ordination will also be in order so that within house we do not have another concurrent system revision waiting in the wings to make an appearance.
At this point, the subject of direct benefit to the Supply Systems Command should be mentioned. Direct access inquiry to the central files and the bank of management data for each stock point becomes a reality for the Command and brings it back sharply into focus as the administrator and controller of this vast and sprawling network of stock points and, eventually, of inventory control points and beyond. Information which is accurate, immediate, and instantly updated is a sine qua non of effective management. In an era of instantaneous response, be it weapons or logistics, administrators who are not armed with this tool are in effect not administrating or, at best, are reacting with information which is weeks or even months old. The answer to a pertinent question can be only a touch away from the display inquiry device but much further away when information must be staffed and correlated. A device in each flag office, keyed to that information most needed at the office, would cost little more than the modest rental of the device itself. The ability to assess the continuously updated data in the distant computer is to be available for individual Stock Point interrogation in any case. This display is quite similar to the type of device used in recently innovated airline reservation and banking inquiry systems and is an accepted and proven tool in the management of remote locations. Each display device may be restricted or unrestricted as to the files to which it has access and as to the type of computation it can command of the computer, and can be configured to present graphic data as well as letters and numbers. Initial implementation with a basic set of data available would be almost routine with presently known techniques. The more exciting and demanding applications can be phased in as the system is shaken down. Project status, predicted workload and stock or financial positions should all be well within calling distance. Accurate participation in war gaming exercises through logistics system simulation is a likely possibility in this proposed configuration. An independent assessment of the work of an activity or the progress of an undertaking based on complete and accurate retrieval of information could be one of the most important management uses of the proposed system.
Both local command and Supply Systems Command administrators stand to benefit from this use of the computer in advanced management concepts, wherein centralization does not subordinate action to reaction.
Configuration—the specific pieces of individual equipment (hardware) combined to make a data processing system in a particular installation.
Control Program—a program by which the computer directs operations or processes, or automatically holds or makes changes in operations or processes on the basis of prescribed rules, also referred to as a monitor or executive program.
Interrupt—To break into a program being performed on the computer in order to perform a different operation or to inquire of the computer’s stored records.
Random Access Storage—Storage by means of techniques and devices which make it possible to obtain information from all locations in essentially the same amount of time. The term is relative. For example, drum storage is more random than magnetic tape but less than internal storage.
Real-Time System—A data processing system having the capability to access and act very rapidly upon any item or group of items of information, to receive and transmit data from and to remote points in extremely short periods of time, and to assign priorities to different tasks so that routine operations can be interrupted in favor of more urgent ones. Applied to computers capable of keeping pace with events which follow each other with great rapidity or even occur simultaneously.
Software—A term applied to all programs and routines used to extend the capabilities of computers, such as compilers, assemblers, control and executive programs. Contrasted with hardware.
MILSTRAP—Military Standard Transaction Reporting and Accounting Procedure.
MILSTRIP—Military Standard Requisitioning and Issue Procedure.
MILSTAMP—Military Standard Transportation And Movement Procedure.
This glossary was adapted from a more comprehensive listing which appeared in the July 1965 Supply Newsletter.