Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
6 MANAGING TECHNOLOGY The Social Security Administration (SSA) has made significant progress in modernizing its systems, but the job can never be regarded as completely finished. The agency still faces inevitable technology-induced change, and managing it well will be both a great challenge and a great opportunity for improving service and effectiveness. Recognizing that this effort is a continuing process, the committee recommends that the SSA continue its transition to an on-line automated system, eliminating the most onerous paper-based manual tasks that still typic its overall operations. In doing so, the agency can continue to free up staff currently involved in repetitive, routine, procedural tasks, allowing them to devote more time to the more necessary and rewarding client advisory work that, when done well, brings credit to the agency. The committee further recommends that the SSA provide for technological change as an integrated part of its plans to automate all its critical, mandated functions. The committee also recommends that the SSA's computer operations evolve so that current second- generation systems are phased out by the year 2000. Successful systems transition involves careful planning of target systems, resource allocations, and strategies for innovative transition consistent with a coherent overall vision of the agency's future. 57 1
58 AUTOMATION AT THE PROGRAM SERVICE CENTERS The computer applications performed at the SSA's SLY program service centers (PSCs) differ in function and interaction from those performed centrally at the National Computer Center (NCC). Thus these two operations have significantly different requirements for both computer support systems and personnel. A major function at the PSCs is dealing with cases that are too complex for local offices to handle. These cases must be tracked so that actions and resolutions are handled correctly and records in the central files remain accurate, complete, and synchronized. PSC computer applications must therefore include tools for tracking cases, generating notices to clients' co~Tesoondin~ with local government agencies. and aiding decision making. - ~ ~ ~ ~ = ~ ~ O ~ ~0~ ~ A-- ~A- - -D ~- - - -- - - - -I- ~-~- The committee believes that such applications can be expanded to improve productivity and service. Furthermore, it is necessary that the PSC applications be served and supported by modern, efficient, cost-effective systems that can easily be expanded and operated and maintained economically. During its review, the committee found that the high-volume data storage and transaction processing required at the NCC are not required at the PSCs. However, with the introduction of more sophisticated systems to support decision making, requirements for processing power at the PSCs can be expected to increase more than requirements for handling large volumes of transactions and data. Given this scenario, the PSCs might best be served by networks of workstations and file servers rather than mainframes and "dumb" terminals. Software programs could then be maintained centrally and accessed from the file servers and executed on workstations. The workstations need only be sized for modest capacity but should offer fast execution speeds. The workstations could be diskless and networked for local file storage and data transfer to the NCC. The ubiquity of personal computers (PCs) and workstation equipment, the competitive pressures on software and hardware suppliers, and the capability for convenient networking of such equipment are reasons that networks of PCs are preferable to the use of mainframe computers for many applications. A1SO7 mainframe equipment does not support interaction on terminals as well and is not where new software developments involving decision support tools are making the greatest strides. Since it is more economical to buy processing power in small computers, this approach, along with being technically adequate, should offer a long-term cost advantage for both acquisitions and maintenance over a mainframe-based architecture. Software for PCs and workstations will undoubtedly be more economical, as will maintenance costs, when compared to those for mainframe-based systems. Labor costs for software development and conversion are also higher in a mainframe environment. i._
59 Rolldown Strategy The SSA recently initiated a plan, referred to as the rolldown strategy, to reuse surplus NCC mainframe processors by installing them at the PSCs to replace several aging and obsolete systems. As presented to the committee, these reinstalled mainframes will perform the same functions as the old machines--local administrative support such as controlling case work and issuing client notices. The committee recognizes that the rolldown strategy may reduce the high maintenance costs associated with the obsolete systems now in use and may improve overall system performance and availability as well--all at very little capital expense. However, as stated above, mainframe-based systems are not warranted for the PSCs, whose tasks are much better accomplished with networked PCs and file servers. The committee is concerned that over time, by virtue of the higher-capacity processors being relocated to the PSCs, applications will expand and the PSCs could evolve into costly and unnecessary regional processing centers. Furthermore, over time, the cost of implementing the rolldown plan is likely to exceed the cost of the SSA's acquiring new PC-based networked systems, because the software and maintenance costs for mainframe-based systems are greater. Hence the rolldown strategy should really be regarded as an interim tactic whose cost is justifiable only over the near term, about 3 years, at which time a longer-term architecture for the PSCs needs to be ready to put in place. Unfornazately, Lurches mainframe computers from the NCC ace not a good choice for reuse at the PSC~ The committee cautions the SSA not to rely on these PSC-based mainframes to support NCC backup and recovery or any of the other programmatic functions currently performed at the NCC. Also, all software for these systems must be controlled centrally at the NCC. MANAGING BACKllP AND RECOVERY The topic of NCC backup and recovery was addressed in the committee's Phase I report (National Research Council, 1990) and in a subsequent letter report to the commissioner (see Appendix B) dated June 15, 1990. Interim Data Capture Measures As an interim measure for backup and recovery, the SSA plans to enable field offices to perform some limited batch entry of data if computing power at the NCC is lost. Such data capture would be facilitated by and predicated on the installation of automated
\ 60 data-processing (ADP) systems at the PSCs--in other words, implementing the rolldown strategy. Although transferring captured data to the NCC's backup site for processing appears to be a good idea, it could potentially cause more harm than good if not carefully controlled. If this scheme is adopted, the SSA will have to be disciplined in capturing data during a disaster situation. Only data involving substantial changes in beneficiary payments should be collected and processed. Less critical changes and additions should be deferred until a postrecovery phase is entered and processing capacity is restored to the point that it can keep up with the transaction load. If noncritical business transactions are accepted from clients during an interruption of the NCC's ability to function in a crisis, the volume and loss of sequence of transactions may jeopardize the SSA's ability to recover comoletelv with its databases intact. The interim strategy based on data capture at the PSCs must be recognized for what it is: a minimally effective stopgap measure. As recommended in the committee's letter report (see Appendix B), the proper approach is to implement a second SSA data center. The committee is concerned that interim measures may create a false sense of security and dilute focusing on the real solution. Backup and Recovery Teams In connection with contemporary disaster recovery planning and as an organizational issue, the SSA must establish two groups as permanent elements of the agency's systems organization, namely, a backup team and a recovery team. Each team must be staffed with full-time dedicated personnel initially tasked to develop agency plans related to their backup and recovery responsibilities. The teams should also be responsible for assisting the field organizations in developing their own backup and recovery plans, and beyond that, for exercising and testing the plans regularly to identify and overcome any weaknesses that may become apparent when such plans are applied in real-world scenarios. AUTOMATION DIRECTIONS FOR SSI The SSA has concentrated its systems modernization efforts primarily on its Title IT programs, mainly Old Age and Survivors Insurance (OASI). Recently, the agency has begun to modernize its automated support for the Title XVI Supplemental Security Income (SSI) program. The early planning presented to the committee for modernizing SSI appears to be quite ambitious with respect to scheduling the new system's development. Although the committee favors seeing SSI benefit from modernized automation, such a major focus on a new area should not divert attention from the fact that much of the essential Title II automation has not yet been completed. In particular, the ability to
61 process complicated claims in a single uninterrupted session at a computer terminal is needed and will become increasingly important as a larger percentage of the U.S. population diverges from the traditional family structure. The evolution of Title IT automation, begun under the Systems Modernization Plan in 1982~ must continue. This is essential to assuring that a viable information-nrocessin~ · ~ ~ ~ _ _ ^ ~ 1~ · ^ ^ ~ 9 ~ · , a ~1 ~ e ~ , ~ ~ . . ~ liliraStI~UCtUre, DaSlC to Supporting 5bA,S clients, All be In place throughout the next decade. In the l990s, evolutionary changes will affect systems architecture, hardware, communications, and software and will require continuing investments, at levels comparable to those of the 1980s' to avoid declining levels of service or, worse, a repeat of the information systems crisis that SSA faced in 1982. The SSA should ensure that the initial steps toward automating Title XV! programs will bring an appropriate return on investment. The use of client-server architectures, graphical user interfaces, relational database management systems, expert systems, and other modern technologies and techniques should be explored. The committee recognizes that decentralizing some of SSA's processing will be difficult and may require maintaining duplicate code. However, by using a distributed systems architecture to achieve SST modernization, the SSA will prepare and position itself to evolve smoothly beyond its current second-generation systems that now support Title II functions. Since automated support for SSI is currently minimal, the SSA can take SSI modernization as an opportunity to begin its necessary and inevitable transition to a more distributed systems architecture. TOWARD ADVANCED APPLICATION SPECIFICATIONS To provide a foundation for technical decision making, the SSA must establish its data- and information-oriented structure and information needs. The present process- oriented structure of SSA's system development mechanisms is not based on such a foundation. Such a data- and information-oriented foundation will facilitate a matrix approach for planning service and system enhancements and consensus building. Starting with a business model that depicts how the SSA wants to do business and a subsequent data model, SSA must document at a detailed level what information its computer applications must provide, rather than specifying how those applications are to be structured or developed. The base set of advanced application specifications should cover all the programmatic services mandated for the SSA as well as administrative applications necessary to efficiently and accurately carry out the legislated mandates. The SSA must specify the information needs for each application. These include a description of the data needed, how current it must be, and the volume. For those data elements needed only for specific applications, the SSA must document the criteria and estimates of how frequently they will be used, if possible by measurements. These figures
62 must also be augmented with projections to the end of the decade. Documenting these projections independent of the applications allows more rational comparison of processing alternatives. Since the trend will be to distribute some processing to the- local offices, regional processing offices, teleprocessing centers, as well as the National Computer Center, which itself may become bicentralized, these estimates also provide a basis for deciding how to partition the system. Certain client interface tasks are appropriately done in field offices, and others involving the resolution of problems requiring state government interactions can be regionalized, but tasks that require consistency can be achieved best through centralization. Collated information needs will provide a blueprint for the data requirements of SSA's systems. Much of this data is already available in SSA's current systems but is incomplete. For the missing data, the SSA will need to identify the resources required to obtain the data and the process for putting the data into machine-processible form. From data resources to end-user applications, the process of developing advanced application specifications will provide a clear vision of what the SSA's systems can and will provide. The agency's databases will continue to manage and hold information until it is needed. Advanced application specifications will enable the SSA to easily identify where additional effort needs to be put forth to expand or refine the SSA's information resources. With a clear vision established, the SSA can evaluate alternative data-processing initiatives on a common basis. The costs of acquiring, storing, and merging source information can be estimated independently of a single application, and the SSA can assess ways to overcome barriers to improving applications, caused by the unavailability of requisite information. Data Streams and Partitions Most of the complexity in data processing comes from dealing with multiple data streams. Often data from different sources are not well synchronized. In the SSA's systems, because of the huge volumes, data are solit into eight regions. for which distinct -a ~ on- - --of ~ - 17-1 ~ ~ ~ ~ ~ ~ · , · ~ ~ ~ . · ~ · ~ · . ~ tiles ana processing scneaules are maintained. 1aently~ng wnen regional boundaries have to be crossed, and how frequently, is an important measure for assessing software complexity. A clear view of information-flow bottlenecks will help justify investment in areas that cannot be justified by a single end-user application. In estimating the necessary size of databases, SSA should consider resources and requirements and the length of time that data have to be retained for use by applications. Because the SSA's systems are large, some partitioning of software and data storage will be needed. An overall information diagram will help Claris the costs of alternate
63 partitioning in terms of reduced or more awkward communications among software modules. Effectively using software concepts such as modulal~zation, which are key to achieving distributed processing and improving postdisaster operations, requires a firm understanding of partitions and the information flow across them. Organization Considerations The committee believes that developing the type of specifications discussed above warrants establishing an internal group with a long-term assignment. It is equally important that the group remain small and not make actual decisions or design systems. Such a group would have to set milestones for its work and set specific goals for itself, including: documenting all information needs of existing automated Title II applications; documenting all information needs of applications involving disability evaluation without recourse to periodic medical reevaluations; and documenting all requirements for nonfederal information specifically, state certifications of birth, health, and death. ,# _ ·11 _ _ . It W1 ~1 not be easy to staff such a group. Its members must have a combination of SSA experience, the ability to work at a high level of abstraction, a commitment to the long-term goals of the SSA, and a willingness to isolate themselves from the day-to-day efforts that may be more exciting to creative and energetic people. At times their work may be assessed by outside committees or experts, but outsiders should not be asked to take on a responsibility that is so central to the future direction of the SSA's systems. It is understandable that some members of the group may be assigned for a limited time period, although not less than one year. This can have the benefit of transferring technology among different parts of the agency. TECHNOLOGICAL POSSIBILITIES FOR SSA Hardware and Storage The current devices used for storage and retrieval of data and images are somewhat
64 limited in scope at the SSA. Formatted data can be retrieved rapidly as long as there is relatively little variation in size. For image data, good sequential usage capability exists, but search abilities are poor. User need will drive the market and eventually lead to image-handling products that can effectively and efficiently search for patterns. Such economic forces are also likely to lower hardware costs for searching and storage devices. Advances in hardware will include specialized input-output devices and other hardware interfaces. These are likely to offer greater flexibility and ease of use for all classes of clients, including physically disabled persons and nontechnical personnel with minimal education or skills. Special devices for improved audio-visual-tactile input and output for both normal and physically impaired workers and clients should be carefully assessed in terms of their potential cost and benefits. Today's children, as adults, will have grown up with PC-based applications and video games, and this should provide insights for developing user interfaces for future SSA systems. Mass storage will be available for all forms of records. Retrieval and addition of information to individual client records may be provided through personal "smart card" devices. However, the ability of people to carry with them their own copy of SSA information on smart cards or other media implies the need for special stations for the updating and validating of the cards by the SSA or other information providers such as medical practitioners and hospitals. In the longer term even medical images will be storable on such media, and the SSA may find such applications useful. As the future makes more hardware and storage options both viable and necessary to satisfy expectations for client service, various system architectures will be possible, including a three-level hierarchy, a two-level hierarchy, and other configurations constrained only by one's imagination. Even the currently favored alternatives of a "bicentralized" or "distributed" system will eventually need to be reexamined in light of emerging technology and industry trends. Imaging It seems inevitable that in the future remotely located providers of high-volume information, and others such as courts, hospitals, physicians, and lawyers, will be able to provide and request information directly via an imaging system. Examples of such input might include medical X-rays and test data or other legal information needed as evidence to support disability claims. Medical or legal information submitted in voice and image form is likely to be the norm rather than the exception. This will make the collection, storage, and dissemination of data simpler from the user's perspective but will add to SSA's requirements for automation and associated facilities. Such facilities include communications systems as well ~ ~ · ~ ~ ,
65 as devices to store and access images on a centralized basis with assurances of authorization. The current state of the art in database and image technology still has not made it easy to deal with massive volumes of data unless the data are highly formatted. Thus, development of new software for handling different types of information is kev to the '~ , ~ · , , e do e e ~ ~ ~ ~ e ~ . e t;llt;~LlVC lIllU~r~llOI1 01 Images In ADA S 1mormatlon systems. The software must make it easy for all users, including those who are unfamiliar with the technology or who are even disabled, to find the information they are looking for by correctly "understanding" user requests. Software Systems software technology is expanding and evolving rapidly. Decisions to use a new operating system and environment must be considered when applications are being modernized on a large scale. Software system decisions must include due consideration for _ .~ .. ~ ~ .. . . . . tne ease ot updating and maintaining both the base systems environment and the applications. Updating of data, new versions of software, and specific hardware variations or problems should be monitored through a networked configuration management system. A method of keeping data, software, and hardware versions synchronized is an important feature of such a system. Movement toward more sophisticated distributed processing and distributed data- base management implies reuse and possible redundancy of software, and this requires better configuration management and careful consideration of the need for consisteru~y of data across network nodes. Therefore, the SSA must determine how long data can be out of synchronization at a node. With respect to the integrity of data, the SSA must know how much error is allowed in data for short- versus long-term operation; for curreru~y of data, the SSA must know how long data can be different across two nodes in the network. The SSA must also track the emergence of promising new methods and techniques. For example, the Object-Oriented database structure is rapidly gaining notice in software engineering applications; it should prove especially valuable in representing knowledge bases. Similarly, the development and practical use of expert systems and neural nets should be monitored and exploited once the technology is established and its application proven in noncritical areas. There are potential opportunities for using expert systems to guide clients and SSA personnel in the collection of information and to explain in lay terms input errors detected through the imposition of complex rules. Expert systems that explain decisions in complex claims adjustment and disability cases directly to clients without the I..
66 need for intercession by SSA personnel also are a possibility in the future." Other new technologies of potential value in the near term that should be tracked include picture archiving and communications systems (PACS), software management tools for validation and verification, and modern computer-aided software engineering tools. Specialized database management systems (such as PACS) are still undergoing research today and are possibly 10 or 15 years from being economically viable for the SSA. Owing to the possible needs (legal, medical, and so on) of information providers, imaging will increase and change loading on the mainframe and distributed networking services as well. Using this technology will require a great deal of resource for initial input of source images. The addition of voice input devices with the output of audio responses and language translation for SSA's clerical support staff, methods of controlling SSA's processing flow, and the use of expert systems for explanation of problems and special cases will add to the load and require sophisticated interfacing. Possible future additions include tactile devices (intelligent mice) and flexible programmable interfaces for different input and output modes. Telecommunications At present, most clients communicate with the SSA by telephone, rather than by mail or in person. In the foreseeable future, SSA could provide direct touch-tone response to caller inquiries. Inter- and intra-agency communications will grow more critical to the day-to-day operations of the SSA. The data needs of clients and SSA personnel, the increased level of service provided through the 800 number and other interfaces, and the integration with other technologies such as remote imaging all vie for telecommunications usage and exert pressure for additional capacity for both the telecommunications and processing infrastructures of the SSA. Transmitting hiah-resolution images from medical and legal offices will require ~ ~ ~ - -- - - ~ - - ~ greater communications capacity at low costs. Economic transmission of high-resolution images suggests that a "supertax" capability may be employed at the SSA. Such a capability can be accomplished but will require expanding the SSA's communications capacity in local area networks, wide area networks, microwave links, or satellite telecommunications. In an era of rapidly advancing information technologies, the SSA must have the skills to watch for new developments and plan for their use at the right time. Given that mixed-mode operation is likely to involve multiple facilities, the SSA should monitor the costs of tRefer to the Phase ~ report (National Research Council, 1990), pp. 52-55, section entitled 'Howard the Use of Expert Systems," for more information.
67 supporting different accesses (most of these by non-SSA personnel over SSA lines), since they may become large. Security Integrity of information, authentication of the provider, and long-term retention of data from remote providers are important issues being addressed government-wide, especially by the Department of Defense. Authorization for access to data may involve consideration of the types of users, types of access (read, add, update, or delete), and types of possible threat, such as deliberate alteration or deletion of data by a disgruntled user or "hacker" or a nonmalicious user having "fun" but inadvertently propagating errors. The SSA will need to evaluate ways to assure authenticity and confidentiality across a network. Encryption devices (end to end) and other protective mechanisms should be evaluated based on their applicability to SSA's functional requirements. Until now clients have interacted with the SSA only through its employees. This type of client interface reduces the potential threat to privacy. However, monitoring and automatic detection of unauthorized transactions will become increasingly important if clients are allowed any on-line interaction with SSA's systems. The probability of attempts at both internal and external penetration will increase with time. Expert systems may be used as a mechanism to prevent or frustrate such attempts. It is essential to design controls for access and privilege, especially if modifying different components of clients' files is involved. This concern becomes particularly acute if banks, employers, and other institutions are ever permitted access to selected aspects of clients' files. Technological security safeguards such as public-key encryption schemes and biometric devices to verify identity as a means of controlling access are becoming more economical. If smart cards come into use for storing client information, technological and other safeguards will have to be applied to protect against unauthorized access to and modification of client data. Maintaining operations under emergency conditions such as earthquakes, fires, and other civil disasters will become a more important consideration as automation increases, because the loss of access would be more obvious and a greater threat to the integrity of more highly automated SSA systems and data records. Retaining the public's perception of the system's reliability should be a primary objective. !
68 REFERENCE 1 National Research Council. 1990. Systems Modemization and the Strategic Plans of the Social Security Administration. Washington, D.C.: National Academy Press
