Click for next page ( 24


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright Β© National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 23
Chapter III TRANSITION INTRODUCTION The transition implied by the IRS equipment replacement program is a massive project involving the expenditure of $218,000,000 over a period of eight years. Computer equipment will be changed at 10 Service Center locations and eventually at the National Computer Center as well. There will be corresponding modifications to buildings and facilities, not to mention a very large software effort to exploit modern computer programming languages. The transition process will involve as many as 500 technical and professional people at its peak. The documentation necessary merely to coordinate such an effort will itself be very extensive. The Committee could realistically examine only the overall strategy and orga- nization of the transition and assess the adequacy of the available resources. At the more detailed level, it was able only to spot check the formats and scopes of individual plans, the care and electiveness of planning, the skills of the personnel, and the capability of the management team. It did not have the resources to examine the quality of training or the effectiveness of development tools. Much of the planning detail evolved while the study was in progress, and the Committee was forced to make judgments about future events that would result from ongoing {RS planning. The Committee could and did consider alternative strategies and organizations for effecting the transition. It met with {RS managers to discuss the plans and the responsible organizations; it reviewed alternatives with them, discussed how the project fits into the overall {RS management structure, and followed the evolution of the management team. The Issues Two broad issues arose in the Committee's assessment of the transition plan- ning. The first is the overall strategic approach how the system will be changed, the sequence of major steps, and the overall plan for managing the transition. The second is the organization and management of the transition. The questions here are how the strategic plan wit! be executed, who will do it, what controls must be applied, what techniques will be used, and what resources will be available. Perspective Technology, particularly new technology, imposes no limitations on what the IRS intends for the future; therefore, it is not a pivotal issue in the present plan. On the other hand, the computer field has advanced rapidly; the new equipment and its associated operating system software will be dramatically different from that now installed in the Service Centers and the National Computer Center. Thus, technology does become a problem area simply because inherent in its installation are several important concerns. 23

OCR for page 23
24 The IRS intends to keep the general computer processes unchanged to mini- mize risk. While this is admittedly proper, the effort cannot be regarded as simply an exchange of one computer for another on a business-as-usual basis. The so-called Equipment Replacement Program is a big undertaking by any measure. There are risks and pitfalls that must be appreciated in advance and reflected in planning. Significant detailed system changes may be necessary simply to exploit or accom- modate new equipment or system software. Unfortunately, many such details cannot be known until vendor selection is made. However, there must at least be an awareness that the effort is in every way completely different from the customary daily operational environment of the IRS in scale of effort, in nimbleness of management and responsiveness of deci- sionmaking, in scope of documentation, in completeness and tightness of control, and in reaction to unanticipated problems. The Committee recommendations below can lessen the risks, but the Equip- ment Replacement Program must be seen by the fRS as the large engineering development project that it is. Assumptions Crucial assumptions include the following: . . During the transition the IRS must fully accommodate any changes that may occur in tax law. Because there is no need for a radical change in the general structure of the information processes supporting tax administration, a long-range strategy of evolutionary growth is appropriate. Since the {RS cannot pre- dict legislative changes, any such long-range approach must embody the flexibility to accommodate new requirements. A program to replace equipment only, without redesigning information processes in detail, is satisfactory. Indeed, in an effort of such size it would be far riskier to redesign the software processes and change the hardware simultaneously. While system design changes generally should be deferred until the transi- tion is complete, some flexibility can be allowed under prudent and tight management review and control. Although some detailed changes in software will be dictated by the adoption of COBOL, The Committee recommends that the existing general structure of pro- grams and files not be altered to change the processing flow, the functions provided, or the overall system architecture. It is obviously impossible to avoid some changes in software when it is moved from one computer to another. At a minimum, the present assembly language programs must be modified to accommodate the characteristics of the COBOL language. It may even be necessary to restructure the files in the database. At the other extreme, one could envision a complete redesign ofthe information flows that collectively perform tax administration. Naturally, this would imply a wholesale redesign of software; the IRS would be starting ab initio on its software.

OCR for page 23
25 Our view is that the latter is unwise because of the magnitude of the job. The former is unavoidable, and how much redesign beyond it is a matter of judgment as the details of software conversion develop. To keep the software task to a manageable size, the project management must enforce a high level of deterrence to change. There will be a myriad of technical details in deciding how much soft- . . . . . . . ... . . . ... .. . . ware change Is appropriate, some ot~wh~ch will not be known until the equipment has been selected. Thus, our recommendation and advice is to deny all proposed change unless the case for it is extremely strong and the payoff from it sufficiently large to justify the additional commitment of resources, impact on the schedule, and risk. A crucial point in any software effort is the status of the documentation, those descriptive items or program listings that specify what a program does and how it is done. The Committee raised the point with the IRS and was assured that either the documentation is already adequate for the purpose of conversion, or where it is not, appropriate documentation will be created before conversion begins. Some of the latter effort is already underway. We did not have an opportunity to examine samples of the documentation but, given the age of the present software, it is inevitable that it has been patched and repatched to accommodate essential change over the years. Clearly, inadequate or incomplete documentation will magnify the already large task of software conversion and disrupt schedules and funding alloca- tions. The {RS decision to examine the status of documentation first is sound; such resources as are needed to do the task thoroughly must be committed. Avoiding Delays During 1981 transition costs will reach $1 million per month; they will rise to $2 million by 1985. This is a crude but useful measure of the maximum cost of delay. On a project of such complexity, the potential for delay is ever present, and while some delays will arise from pressure to alter the basic way in which processing is done, others will arise from incomplete planning, mistakes in implementation, or unforeseen events. Additional expenditures to minimize delays can be thought of as an important form of insurance. Therefore, some contingency arrangements should be incorporated into the total plan. To summarize, the essential first step in minimizing delay is to select a strategy that minimizes risks. The second is to invest in good planning; the third. to seek out and prepare contingency plans for potential problems. These factors should and do dominate the discussion and recommendations. STRATEGY OF TRANSITION Overall Strategy The transition is to be carried out under an overall strategy of equipment replacement. This is not to say, however, that each item of new equipment merely replaces an old one; the situation is much more intricate. Existing systems must continue to operate as they now do while they are gradually replaced by new hardware with converted software. Some detailed changes will inevitably occur; for

OCR for page 23
26 example, new file layouts or better structuring of programs may be needed, and interim interface software modules will surely be needed. Piece-by-piece change is a desirable way to minimize trauma and delays while upgrading a large system that must continue to accommodate its operational Toad in the meantime. Cut-Over Strategy To understand the process of transition, consider the extremely simplified but basically accurate description of the system shown in Figure I. There are two locations to consider: the National Computer Center (NCC) in Martinsburg, West Virginia, and ten Service Centers (SC) throughout the country. Each Service Center receives taxpayer returns and related documents, batches and processes them for editing and data encoding, and enters the data into a local computer system for correction of errors. After transcription, magnetic tapes are shipped to the National Computer Center for posting to the master file. For any of a variety of reasons an account may fait to post, requiring its return to the Service Center for resolution of the problem. Service Centers are responsible also for complete revenue accountin~reporting monies deposited, assessments, abate- ments, and other control information—as well as printing and sending notices or bills to taxpayers. At the National Computer Center the centralized business master file (BMF) and individual taxpayer master file (IMP) are posted and analyzed; prescribed outputs are produced on magnetic tape. The BMF and IMP are maintained in taxpayer-account sequence- the BMF by the employer identification number as- signed by the TRS, and the IMP by taxpayer identification number. Each account contains taxpayer identifying data such as number, rename, address; returns liability and settlement data; and results of return examination. On a regular weekly production cycle, input tapes from each Service Center are received at the National Computer Center by air freight. In turn, refund tapes are certified to Treasury Disbursing Centers, and other output tapes are sent to Service Centers for printing of notices, bills, and registers. Information on accounts is produced on microfilm, which is distributed to Service Centers to aid research on accounts; the National Computer Center handles no taxpayer documents. Except for special printed reports and extracts from the two master files scheduled for monthly, quarterly, semiannual, or annual production, input and output are chiefly magnetic tape. Conceptually, the overall processing sequence is simple. Tax returns arrive by mail at Service Centers and are transcribed to magnetic tape by a system that will not be replaced or altered (System T).2 Transcribed returns are processed for arithmetic accuracy; when errors are resolved, updated information is sent on magnetic tape to the National Computer Center (System P). A copy is also used to update the local Service Center file on a "pending" basis (System H). ~ "Cut-over" is technical jargon, but commonly used to describe the events of conversion. Further- more, it is descriptive of the transition process in that a long sequence of coordinated actions must take place, generally over a lengthy period of time. Cut-over is not an instantaneous event, such as would be suggested by throwing a switch from one position to another. 2 Refer to Figure 1 for the overall flow of events.

OCR for page 23
27 Processing of Tax Returns Within the Internal Revenue Service SERVICE CENTER PROCESSING L Tax Returns U.S. Mail Transcribe Returns _ Magnetic Tape ~ Prelin inary ~ Magnetic Tape | Processing ; Integrated Data Retrieval System ( (IDRS) ~ \ H Hold in Pending Data Base Control Q Query and Report Facilities , 1 '7 Link Network Terminal Control /// / // Terminals in Service Centers and District Offices FIGURE 1 / NATIONAL COMPUTER CENTER PROCESSING a; / / M Master Files Maintenance Issue Selected Copies to Service Centers D Duplicate Selected Copies on Microfilm O Other Processes Outputs to Other Systems

OCR for page 23
28 . Transcribed returns are processed at the National Computer Center to update the master files (System M). Once a week, magnetic tape copies of selected records and parts of records are sent to each Service Center (System I). Each Service Center uses the material from the National Computer Cen- ter as its local database, supplementing it with pending data from the local preliminary processing and local files used to support ancillary activities (System D). Each Service Center provides the same set of query, processing, and up- date capabilities to itself and the associated District Offices (System Q). A separate system manages the telecommunications network and termi- nals connected to each Service Center (System N. which is not being replaced or altered). Systems H. D, and Q together comprise the Integrated Data Retrieval System (IDRS), which was developed to provide fast access to up-to-date taxpayer account information. With the IDRS, Service Centers and District Offices are connected to a specially created database whose content is based on the probability that an inquiry for a particular account will occur. It typically contains about 10 percent of the taxpayer accounts for which a Service Center is responsible. IDES also provides operational management information for examination and collection ac- tivities. It has significantly improved the ability of the {RS to provide quick re- sponse to taxpayer's inquiries and to monitor delinquent accounts. The files at the National Computer Center are processed weekly to make full copies of selected records available to Service Centers on microfilm (System D). All other processing at the National Computer Center is self-contained or provides outputs to systems other than those at Service Centers (System 0~. The replacement program does not change any of the flow structure shown in Figure 1. The basic change is to four sets of hardware, with concomitant software changes. They are: System P in the Service Centers Systems M, I' and O in the National Computer Center Systems H. D, and Q in the Service Centers . System D at the National Computer Center The overall program is divided into three largely independent conversions: Phase 1: Convert the files and programs at Service Centers (Systems P. H. D, Q) and use temporary interface programs to keep them operating with the National Computer Center. This phase is called the Service Cen- ter Replacement System (SCRS). Phase 2: Convert the National Computer Center (Systems M, I, O). . Phase 3: Convert the microfilm operation (System D). The general concept of separate, noninteracting phases is clearly sound. It reduces risks and minimizes the complexity of the events to be controlled; both are essential in a project the size of that being undertaken by the TRS.

OCR for page 23
29 As of the date of the Committee's work, only the Service Center Replacement System had been developed in detail. It is the most complex, and this report there- fore discusses only Phase 1 in detail. Service Center Replacement Strategy The general strategy for Service Centers is to use the Memphis Service Center as a pilot project, on a schedule that permits thorough testing and operational certification. The remaining Service Centers will be upgraded, one at a time, by installation teams. A schedule and general plan for this has been prepared, and it is a sensible and sound approach. The general approach for the pilot effort is gradually to move files and their associated processing operations concurrently from the old system to the new one. The procedure can tolerate some schedule uncertainties and delays, and it avoids the trauma of a tightly scheduled cut-over, which can be extremely troublesome when delays occur and deadlines are missed. Use of a Higher Level Language During the transition, the TRS will continue training its stab in the use of structured program design and the programming language COBOL. A major part of the transition will be converting the current assembly language programs to COBOL. The decision to move to a high-level language is a good one, since it will ease future changes and make support of the software more convenient. The choice of the language COBOL is also sound. Other possibilities raise difficulties of various kinds; e.g., some languages are not appropriate to the nature of the processing the {RS does; others are not provided by all vendors who might bid so that choosing one of them could inadvertently restrain competition; still others do not provide enough additional capability to justify their additional complexity and difficulty of use. Making the change to a high-level language during transition is also a sound decision, allowing learning costs to be absorbed in the transition effort. ORGANIZATION, MANAGEMENT, AND RESOURCES As explained earlier, the Service Center Replacement System portion of the Equipment Replacement Program will take about five years, will occupy a stab of 500 people during its peak year, and will involve not only some aspects of the National Computer Center, but also all Service Centers and most of the ]:RS organi- zation at one time or another. A comprehensive program of such magnitude clearly demands an organization to manage and control it. Such an organization will represent the chain of command through which activities at the various echelons are planned, approved, assigned, supervised, and executed. The particular details of the organizational structure selected to manage and control a major effort de- pend not only on the nature of the undertaking itself, but also very much on the characteristics of the parent organization in which the task is to be carried out.

OCR for page 23
30 In the course of its examination, the Committee pinpointed many specific issues that have significantly influenced the success or failure of other efforts in a wide variety of organizations. As issues were recognized, each was reviewed and dis- cussed with the relevant fRS project staff. The discussions and review actions satisfied many of the Committee's concerns, but others, highlighted under the next few headings, deserve continued review by the IRS staff. The Task The Service Center Replacement System consists basically of replacing the obsolescent computers in the 10 Service Centers with modern processors, plus converting all the existing programs from assembly language to COBOL. The task is straightforward in principle, but in practice it becomes complex because of its magnitude 1.5 million lines of code, 1040 programs, 10 Service Centers- and the need to carry it out over a period of five years without disrupting normal operation. The Issue of Organization To plan and execute a large undertaking like the one the {RS intends, some organizations would establish an entirely separate entity, often referred to as a "project office," with its own manager and staff; it would become a self-contained entity with complete responsibility. Other organizations execute large projects within existing organizational structures. The {RS has chosen the latter alternative. It will augment the stabs of existing sections and add staffas needed in appropriate branches of the Systems Design and Programming Division (Figure 2~. Conversion of ah programs will therefore be done centrally at the {RS headquarters office in Washington, D.C., under the responsibility of the Director, Systems Design and Programming Division. The advantages of such an arrangement are as follows: Computer program conversion will be carried out in organizational units that are already responsible for, and therefore familiar with, the functions and structures of the programs in detail. There is a continuing need to maintain the existing computer programs and incorporate changes, such as those mandated by legislation, while simultaneously writing and supporting the new computer programs. Tt~ makes sense to retain a close connection between the two efforts. New additions to the stab an increase of perhaps 200 above the present level of about 625 will be required to carry out the conversion. It will be simpler and less disruptive to integrate new and relatively inexperienced staid into existing units. It is probably not possible within the organizational tradition of the Inter- nal Revenue Service to establish an autonomous and powerful project once to accomplish the tasks of the Service Center Replacement System program. The Committee agrees with the advantages cited above. We endorse the organi- zational plan developed for the Service Center Replacement System but note that such an organization usually has difficulty in uncovering and resolving the many

OCR for page 23
31 ~4 - - c - C 5,} — o ~ C ~ E ~ c ~ ~ o — ~ <: <: ~ ~ ~ ~ sit So ~ c ~ ~ Go C) ·~ Cal ·E Cal et Go o ._ · .~ o e a ~ ~ .~= ~ O ~ E _ ~ C ~ a 0.0 ~ ~ · - ._ Cal circa 6 ·Crq q e" Cal o Ct V o . - es ·~ o Cal . . _ C 'a , ~~ =_ a . . C U. ~ a Vat :t C C.) - ' ° V ~ C ~ ~ ~4 o - a ea _ :t _ ~SV - ~ - S :5 ~ ~ V7 V a m-S ~ V — ~ s 5 — ~ . _ _ ~ U. V . s ~ :t C —~ . ~, C, _ = C~ V ~ ~ ~ . . _ _ C~ E ~ _ — ,, ~s C~ ~ :t C _ ~ o C ~ · _ _ _ — s ~ ·, c,, ~ _ C.} _ C~ o, ., ~C — — o~ _ _ <,o,~,e _ V, U. C~ . . _ _ ~t _ .~ _ s _ ~ C — ~ & . _ . ~ S r~ V: ~ .. · . a - ~ ~ _ _ ~ ~ V ~ ' Y V— ~ _ 2 e =_ _ —4 _ 0L C ~ V' .O C~ Vs' — CL O ~i C (~ _ _ S (n V~ r~ [_ ~ . . . m 1 1 & 0~ — C., rS _ & ;.) _ V-O-c e, _ . ., ~ _ _ ~ m .' V} C .O C O :~-~ _ ._ ~ 0 C~ C~ () ~ U.— _,, . _ _ m 1 1 1 1 1 1 '1 . _ . s— 1 ~ a i'~ ~ m~ ~ - .~ ° — s C~ V . . . · V ~ ~ ~ V C) s :~ 2~ C — J _ '_ ~ _ S L;) V7 'V E- =,, . ~ _ _ ~ m V, ._ ~ C) >, . :t r _ C ~ :e _ ~ s C ~~(_~- _, . . 0~ _ C~ _ ~ ° .O _ s 0 _ ^~ V O t'C ~ _ ", s ._ V, V ~ > . . . ~ _ _ —} U) _ O . ~ c =._ C ~ ~ :) ._ . . . _ 1 ,_ 1 ~ I V C~ 1 c _ ~, 1 =._ 1 _ s 1 —~ u I ~ . . . 1 o——~D 1 ~ _ 1 ~ ~ ~ - 1 1 ~ U, U 1 ’) . . . 1 ~ — 1 — _ C~ — ·J S ~ V _ I eV'Vi~ ~ _ r~ ~ 1 :~ — 1 1~ 1- 1 ~ ~ 1 — .~ I C S I tV V — I I _ _ ' 1 1 ~ C ~ - 1 I ::S '= '— S 1 — ~ ~ '~ 1 I ~ V)'v ~ 1 Y . ~ ~ 1 1 v _ _ ~ 1 ~ — 1 1 1 1 1 1 1 1 ~ 1- 1 c~& I~a ~l | ";a_-~ ' _ · · · 1 ~——~D 1 1 ~ — 1 y v - = .~ S ~ 'V — C _ — — ~ ~ 'C C) _ — S C~ C) — [, C _ r~ a ~ . C _ _ S ~ , I ,, I ,, I ~'''- I I I , ... _ _ ~ =-,. 1 Y . . . 1 L~

OCR for page 23
32 minor conflicts that always arise. Decisions may be made at too low a level for management to see and understand their implications. Therefore, to maintain control and visibility, we strongly recommend that the Systems Development Office have sufficient staff and be organized so as to review the decisions and tradeoffs made at all management levels on the project. The Issue of Conflict . The new hardware and converted programs at the Service Centers will be Installed by two independent installation teams. The equipment team will be pro- vided by the System Support Division and the programming team by the Systems Design and Programming Division. The System Development Office will coordinate and closely monitor the work of the installation teams. AS the Service venter Replacement System program proceeds, conflicts will develop, particularly with regard to priorities and resource allocations, cluring both conversion of programs and installation in the field. Conflicts involving the conversion of programs can probably be resolved within the Systems Design and Programming Division, if care is taken not to set back the plans and schedules established by the Svstems Develooment Office or the field .. ~ .. . , . ^. operations. Otherwise these contacts may need resolution at the Assistant Commis- sioner level (Figure 3~. Conflicts during the installation phase in the field are mor likely to be the result of conflict between the schedule and priorities set by the Systems Development Office, on the one hand, and the day-to-day requirements of field operations on the other. These also may require resolution at the Assistant Commissioner level (Figure 3~. The {RS now has a Policy Resources Board for automatic data processing; it Includes all Assistant Commissioners and is chaired by the Assistant Commissioner (Data Services). The role of the Board is to decide on system development priorities and to allocate resources. We endorse the concept of a separate Priority Committee, chaired by the head of the Systems Development Office, to deal with the issues of conflict during the development and implementation of the Service Center Replacement System. The members should be at levels close to those at which operational problems will arise. The committee should include senior representatives of the divisions within the Office of Data Services plus representatives from the regions. A committee, with its scope of authority defined in writing by the Assistant Commissioner, will, in the opinion of the Committee, greatly expedite the cooperative and timely resolution of conflicts. Installation and Cut-Over The Committee views the interval during installation of new computer hard- ware and the cut-over from old software systems to new ones to represent a special risk. Previous experience with similar transitions of large systems indicates that great care is required. The Committee recommends that specific detailed plans be prepared for the cut-over phase, including acceptance criteria, and that prior preparations be made for emergency back-up actions.

OCR for page 23
33 Ct ~ ~ A: C V ~ ·E Cal ;^ o C) E ' ~ a fi ~ cq ~ do en o . o C) ~ ~ U. hi C ~ Cal Ct . . o ~ ~ ~ ~ ~ 3 C ~ o ~ ~ hi LL1 o ~ ~ it; ’:L— a ~ ~ 3 CD c C: U. ~ Pt .O d~ l l l 1 1 3 _ o.; O Cal CL, I_ O it, US C) me, —~ V, ~ ~ At,, ~ O ~ ~ 0, =3 C:.- o ~ . I- - :- ~ ~ ~ ~ =' ~ ~ ~ ~ I ~ ~ C) C Cal .O .' ~ o 6 c: ~ ~ecL _ v] o 0 3 ~ => c ~ O - o~ C ,_ Cd ~ 1 C) ,_ 3 ~ o 4., ~C o o . C) ~ ~ m ~ C) ·— C~ E -~ ~, ~ :~` ~ ~ V, , 1 L= _ ~ — ~Oa _ ~ ·~ E . _ . r c ~ ~ ~ = 3 C Z o ~ o o ·=C.> == :5~ .. ~ ~ V, o o o .o . ~ . . . . ., >, ~ ;>, ~ O4= ~ Ct_ ~ h~ C~ —t ~ =, ~3 E ~ E~ o o~._ 3 3 o ~ ~ . _ . Z ~ 1 1 C) ~o ~ ~ := .= ~ . E C ~ ~ _% (., — .' ~ o C.'a ~ E 6 - .V) C) ' =.' '" E ° E~ ~ _ ~ o ~ ~ c V: _ _ ~ ~ i 6~- V, ~> C o .o l 1 ~ r~^ O C~ E Ct ~o _ CO~ _ l l 1 1 1° 1 1 ~ 1 I ~ 1 l& 1 1 - 1 1= 1 - ._—1 1 °~ - 1 , ~ ~, ~ ·- ~ ~ n-~' la ~ C~ C) o . C~ C Cd V' C~ C . ° ·— =) ~ ~ C) O L~ C)

OCR for page 23
34 The risk of system cut-over can be lessened by a number of early preparatory ac- tions. We recommend that the test-and-evaluation program contain precisely defined criteria for cut-over qualification. Following successful completion of testing, complete cut-over is ready to occur. Two safeguards will be necessary at this stage: . % . The old and new systems should be operated concurrently until all person- nel have become accustomed to the new system, the required operational results have been obtained from the new system, and satisfactory overall reliability has been demonstrated. Only then can the old system be dis- mantlec3. A readily available back-up system and/or a responsive arrangement for system recovery-and-restart must be available. No cut-over steps can be considered until testing clearly indicates that a successful transition can be completed, and that all problems or delays can be handled within the time limits required by Service Center operations. It is our opinion that the final decisions related to cut-over during the actual installation of the new systems and programs in the field should be given to the officer with operational responsibility the Regional Commissioner or his designee. There are two major reasons for this: . . He is closer to and more sensitive to the operational needs in the region, When conflicts arise between the schedules and priorities set by the Sys- tems Development Office and the day-to-day requirements of the field operations, operations should be properly represented and have priority. Systems Development Office Planning and scheduling, coordination, and fiscal control of the Service Center Replacement System program will require constant attention to detail. A Systems Development Office, reporting directly to the Assistant Commissioner (Data Ser- vices), has been established for the purpose. The Office will coordinate and maintain the "project plan" for such diverse activities as fiscal oversight, design, develop- ment, implementation, procurement, facilities, training, and public relations. It will also coordinate and maintain the overall program budget. The nature and extent of the authority vested in the Systems Development Office must depend entirely on delegation from the Assistant Commissioner (Data Services) and/or the Commissioner himself. However, it is clear from the nature of the technical and management effort to be performed in the Systems Design and Programming Division, and from the limited number of personnel currently planned to staffthe Systems Development Office (approximately 15) that it can play only a general planning, coordinating, and advisory role. It could not be a true project office. This is not to demean its role, but simply to clarify it. We believe that such an arrangement offers an appropriate rode for the Systems Development Office, given the nature ofthe Internal Revenue Service as an organi- zation. Even so, the Systems Development Office, to carry out its responsibilities in the detail and in the timely fashion required, will have to possess strong project-

OCR for page 23
35 office-like authority and be in close and constant communication with all the units involved in the project. These include, at a minimum, the analysis and program- ming teams, recruiters and trainers, the installation teams, the budget office, contractors, and field office personnel. The information to be analyzed and reported for control must be extensive. We doubt that the presently planned staffing level for the Systems Develop- ment Office is adequate to ensure monitoring in sufficient depth. We therefore recommend that the staff of the Systems Development Office be approximately doubled from the planned 15; that the Systems Development Office obtain expert advice on the adequacy of its project management system at the earliest date; and that representatives of the Systems Development Office routinely attend plan- ning and design review meetings where program issues are discussed and decided, to develop their own sense of the information being provided by various task teams and also to maintain the automated project control system. Project Funds Management i! Funds management is essential to achieving the schedule and staying within the budget. The Committee considers the area of effective funds management to be extraordinarily critical, and the key to minimizing conflict and delays. The Committee recommends that the {RS require close and careful control of project funds by the System Development Office. Management of project funds is a task that spans the complete project life. It s directly related to all phases from planning to development, testing, installation, cut-over, and system operation. Funds management is also a vital too] for the project manager to use in tracking progress against schedule, relating expenditures to planned task completions, and controlling the budget. No project member can be allowed to commit or expend funds except in accordance with the approved overall plan and budget. In the project budget the total effort must be divided and subdivided until all tasks are defined at a level sufficiently detailed to be readily understood and man- ageable. In such a process, the Work Breakdown Structure will identify al] major project areas from the grossest to the most detailed level, until all task and expendi- ture activities are completely identified. At the lowest levels of activity, estimates will be made of resource requirements (manpower, money, material) and budgets will be assigned. As budgets are aggregated from the most detailed Work Break- down Structure levels to each major Work Breakdown Structure area, a total project budget will evolve. During the implementation period, all activities must operate within the ap- proved budget. In this period, work packages will be developed to correspond with Work Breakdown Structure identified tasks. Personnel will be assigned tasks, equipment will be purchased or leased, and all activities will be geared to the budget. Performance and cost measurement reports and controls will be needed throughout the implementation period. All staff personnel authorized funds under a budget and assigned to tasks should report to project managers on a regular basis.

OCR for page 23
36 . Regularity in reporting provides early warning of forthcoming or anticipated problems in schedules or costs. In contrast, after-the-fact reporting simply clocu- ments history and is of correspondingly less value in project management. Reports must clearly measure progress made against that planned and explain in adequate detail any variance in progress or cost. The project plan and budget ~~u~u ';uver ail direct expenditures, all support elements, and as required under IRS fiscal control procedures, training, construction, communications, contractor sup- port, and so on. ~~~ ~~ ~ I I —.~d ~..~ AL _ 11 ~ 1 _~ 1_ _ 1 ~ The Committee discussed the topic of funds management with the TRS staff on several occasions. We believe the area of overall TRS project management needs further attention. We understand, for example, that TRS funding for the project will be commingled with other IRS funds; and that the normal {RS work assignment and funds management controls will be applied to work performed for the project.3 The Committee strongly recommends that the {RS hold separate all project funds. It further recommends that the Director of the Systems Develop- ment Office be given budget assignment and control authority for project funds. These recommendations do not require handling of funds outside current {RS financial management organization. However, they do require establishment of project budget and fund control measures that allow the project manager to oversee the expenditure of funds approved for the project. The collective experience of the Committee in managing large project tasks leads us to conclude that failure to follow the preceding recommendation will seriously jeopardize the optimal completion of the whole effort. Resources The Equipment Replacement Program is estimated to require an investment of 2000 sta~-years in the calendar years 1980 through 1987. The Service Center Replacement System portion is estimated to require 1500 of these sta~-years from 1980 to 1985 and is therefore the major part of the effort. To achieve such levels, the staff' of the Systems Design and Programming Division is expected to increase from the present level of approximately 630 to about 700 in 1982, and to peak at about 725 in 1983. Another 100 personnel may be needed in the field to carry out the implementation. The Committee is not in a position to comment on the accuracy of the estimates; to do so would require a detailed review of all the tasks required in the Service Center Replacement System, as well as a detailed examination of all computer programs and files to be converted. However, while stay additions at the rate of (say) 60 to 100 per year are not unheard of, such growth is Circuit. We conclude that the staking levels are attainable with a concerted effort but that the training requirements will be extremely burdensome. It is our understanding that the TRS is addressing these training requirements. It was mentioned previously that at the peak level of the program, the transi- 3 For example, computer programmers and analysts assigned to a work section may or may not work on a project task on any given day, and funds used for project construction are not under the control of the project manager.

OCR for page 23
37 talon costs may reach $2 million per month; thus, a three month slippage in the schedule can increase the project's costs by a maximum of $6 million. Given the program's complexity and duration, situations that may contribute to delays are almost certain to arise. We urge insuring against such an eventuality as suggested by the means discussed earlier. Each action will require an investment in additional resources, or at least the reallocation of funds. Project Master Schedule and Detailed Planning The Committee supports the concept of a single approved master schedule for the project. The master schedule would identify all key project activities and be supported by detailed schedules and planning in all project areas, but could not be altered without approval of the project manager. The master schedule would show relationships among project activities, allow concurrent actions as appropriate, and identify the project's critical path. The Committee recommends that the Systems Development Office main- tain and periodically review its master plan and the necessary project reporting procedures to ensure a steady flow of management information. Management information will be essential to support such a master schedule, so that progress can be monitored. The Committee recommends that contingency plans be developed for the circumstances most likely to raise difficulties with the project system per- formance, schedule, or cost. Schedule delays can result from any of a number of events, and are normally accompanied by cost increases. The project manager should plan in advance for the most likely impacts by providing alternate courses of action, provision for the use of outside help, and financial management reserves. The Systems Development Office should consider retaining an outside contrac- tor, experienced in project management and in setting up and maintaining a project control system, to review the system that the IRS now has in place. Such a contrac- tor should provide staff who have actually managed a project control office. Their experience and their disinterested, unbiased assessment of the project's status and progress will be invaluable to the senior officers of the {RS. The Systems Design and Programming Division should contract on a retainer basis for the services of a large, experienced software contractor to augment its resources (within existing civil service regulations) on demand at short notice, in case of unanticipated delays in the program conversion. The Systems Design and Programming Division should assign two or three competent project leaders from the contractor's stab, as full-time members (not necessarily as project leaders) of certain key teams on the project, as may be consistent with civil service regulations. This will allow them to become fully familiar with the project and develop hands-on experience in key tasks. If and when problems develop and delays materialize, additional resources from the contrac- tor's larger pool can be called upon. The contractor's project managers will already be on the job and thus be able to augment the IRS's own resources with a minimum of delay.

OCR for page 23
38 Independent Testing and Evaluation In all major projects, testing and evaluating the products produced is a task of great importance and a possible source of conflict. In the Service Center Replace- ment System effort, testing and evaluation are particularly important, due to the large amount of software to be produced, and also to the fact that new systems must go immediately into key IRS operational functions. The Committee recommends that the IRS evaluate the desirability of con- tracting with a firm that is expert in test-and-evaluation operations, either to undertake the actual test-and-evaluation or to ensure the adequacy of the test-and-evaluation procedures developed by the {RS. Project Documentation Project documentation is frequently slighted in both planning and implementa- tion phases. Adequate project documentation is costly in money and in technical effort, but failure to provide it can prove even more costly in the long run, create substantial schedule delays, and introduce unnecessary added risks. The Committee recommends that a project documentation tree be specified by the Systems Development Office. The documentation tree should name all required specifications, test documents, manuals, handbooks, and re- ports. Each item required by the documentation tree should be described in form and content, and assigned to an organizational element for preparation and revision. The {RS Service Center Replacement System project is large and complex, and work on it will be conducted by different groups throughout the country. To pro- duce specifications for the new system, it may be necessary to update or even create the documentation for the old system. This could be a substantial and unrecognized task. Impact of a Higher Level Language The TRS computer systems are dynamic and are expected to undergo frequent changes in the future. The planned change from assembly language to the higher level language COBOL is necessary and technically feasible, but it is not without risk. The current IRS Service Center programs are in assembly language and can thus be expected to have advantages in terms of program size and processor re- sponse over corresponding programs in COBOL. The erects are not precisely known. However, in the Request for Proposal solicitation, the IRS has estimated that file size will increase by 50 percent arid that the burden on the central process- ing unit will increase by 300 percent. Changes in system response time may also occur as a result of the change from assembly language to a higher level language. Any slowing of system response due to language changes can be partially offset by newer, faster hardware. We recommend that the IRS staff continue to study the response time is- sue and estimate by mathematical analysis, simulation, or tests the likely effects on the performance of the computer system.

OCR for page 23
39 The Committee believes the change in language will prove to be a mixed blessing. The ability of the IRS to make fast changes to computer programs will be improved, but the need for faster hardware and additional memory (especially t maintain the response required by the {RS) may introduce significant hardware costs, especially in memory. Although analysis will provide some knowledge about the impact of COBOL, hard information will not be available until some tests have been conducted. As test data become plentiful, system design personnel can better identify system performance parameters and make design changes if necessary. Also, if necessary to ensure rapid system response, small portions of the programs could be retained in assembly language with little or no impact on program change issues. The Committee recommends that the IRS investigate the availability of a program translator to aid in the conversion of programs from assembly language to a higher level language. Some success has been obtained in the use of translators, and the TRS should examine a number of the approaches developed by industry.4 With the aid of experienced personnel and the use of translators, large portions of assembly pro- grams may be convertible to COBOL or near-COBOL by an automated operation. Success in the use of translators could result in worthwhile savings in manpower during the conversion. THE NATIONAL COMPUTER CENTER Although the Committee addressed only the Service Center Replacement Sys- tem, many of the same cautions and recommendations are clearly pertinent to equipment replacement and software conversion actions that will occur at the National Computer Center. Furthermore, there is a technological option available that seems not to have been considered by the TRS. Although networking as an overall nationwide system architecture is for various reasons not now acceptable, networking of the several computers within the National Computer Center is permissible, provided that electrical connections to the external 'world do not exist. As the various computer systems of the National Computer Center are replaced by more modern ones, the IRS should consider networking them to achieve such advantages as: Better and more automated work flow Better overall availability of computing power Less need for operator intervention and manual operation Ready backup among individual systems. . 4 For example, two large insurance companies have used such translators, one as early as 1976, for this purpose.

OCR for page 23