.. --..l .I-_l..- ._..l. --. GAO St~plt~llllwr I!M) INFORMATION RESOURCES Management Improvements Essential for Key Agriculture Automated Systems RESTRICTED --Not RFLEASED to be released outside the General Accounting Offhe unless specifically approved by the OfTice of Congressional Relations. ---- I_--. (;AO/IM’I’14X :-W-H!? ~..--~..l__-l”~.--.~i”l-.- ---__.- -.._“..-------- United States General Accounting Office Washington, DC. 20548 Information Management and Technology Division B-240109 September 12,lQQO The Honorable Robert E. Wise, Jr. Chairman, Subcommitteeon Government Information, Justice, and Agriculture Committee on Government Operations Houseof Representatives Dear Mr. Chairman: After expressing interest in our ongoing work at the Department of Agriculture’s Agricultural Stabilization and Conservation Service (ASCS), on February 14,1990, you requested information on two major ASCS information resourcesmanagement(IRM) projects. You asked us to (1) explain escalating costs and schedule delays for ASCS’Grain Inventory ManagementSystem (GIMS) and ProcessedCommodity Inventory Man- agementSystem (PCIMS), and (2) evaluate how ASCS monitors and man- agesdisk capacity to determine upgrades for minicomputers in its approximately 2,800 county offices. Ineffective project managementand oversight contributed to cost Results in Brief growth, scheduledelays, and important user needsnot being met for the grain and processedcommodity inventory systems.Current cost esti- mates for these systems amount to approximately $62 million, which is almost 9 times the initial estimate of about $7 million. Onesystem was installed 2 years later than planned, and the other is scheduledto be installed in July 1991, which would be almost a 6-year delay. The pri- mary factors leading to cost increasesand schedule delays were poor initial estimates, ill-defined user requirements, an ambiguous systems concept design, and weak changeorder controls. Changesin legislation and policy also accountedfor someof the cost and scheduleoverruns. ASCS has drafted a request for about a $57-million increasein procure- ment authority, primarily to purchase additional minicomputer disk drives and computer equipment for county offices between July 1990 and September 1992. Our analysis, however, identified flaws in the method ASCS usesto justify these purchases.On average,about 12 per- cent of county offices’ disk capacity is wasted to maintain unnecessary files-files that should be deleted or offloaded to other storage devices. Including these files, as well as other deficiencies, distorts projections of future disk capacity requirements at county offices. We believe that cur- rently available funds can cover county office disk capacity needsuntil Page 1 GAO/IMTEGfN-S6 Agriculture Automated Systems i B-240109 September 1992, and that the $67-million procurement request is not justified. We are making a seriesof recommendationsto improve the way ASCS (1) managessystem development projects and (2) estimates and justifies computer needsat county offices. As an agencyof the Department of Agriculture, ASCS administers farm Background commodity, conservation, and emergencyassistanceprograms, including commodity loans and price support payments to farmers. To support this mission, ASCShas automated the acquisition, storage, movement, sale, and donation of grain and processedcommodities inventories. ASCS also has equipped its county offices with minicomputers to help process commodity loans, pay price supports, and enroll farmers in conservation programs. The GIMS, PCIMS, and network of county office minicomputers are critical componentsof MS’ information resourcesmanagement.These auto- mated information systems directly bear on how well ASCS performs its major program responsibilities, In fiscal year 1989, MS’ program and administrative costs amounted to some$13 billion. In addition, at the end of fiscal year 1989, ASCS’commodity inventories amounted to nearly $4 billion. ASCS awarded a contract in 1983 to develop the two major automated inventory systems at a combined cost of about $7 million. ASCS’esti- mates of scheduledcompletion and estimated costs for GIMS and PCIMS changedsubstantially between 1983 and 1990. About a 2-year delay occurred in GIMS installation, and numerous enhancementswere required to meet users’ needsimmediately after the system was installed in 1987. After the contractor completed the initial user requirements analysis and system design, ASCS terminated PCIMS’ system development in 1986, only to restart it in 1987. It is still in the acceptancetesting phase and is now scheduledfor phased installation that ASCS expects to complete by July 1991, nearly 6 years late. The cost of both systems has grown to a current estimate of about $62 million. Besidesthese two large inventory systems projects, MC%acquired and installed IBM System 36 minicomputers in each of its approximately 2,800 county offices to support program administration and office auto- mation. As of June 1990, ASCS had spent about $181 million to install, upgrade, and maintain these minicomputers. Page 2 GAO/IMTECOO-II5 A@hkure Automated Systems B240100 ASCS originally planned for a complete equipment replacement in its county offices by the end of fiscal year 1992. According to the Assistant Deputy Administrator for Management,planning for this replacement project was begun and later postponed by top AXX managementseveral times becauseof higher agency priorities. It now expects to completely replace all existing equipment by 1996 at an estimated cost of $960 mil- lion over its lo-year life, making it one of the largest ADP projects the agency will undertake in the next few years. Delays in starting this equipment replacement have made it necessaryto upgrade existing county office equipment on an interim basis until the replacement occurs.AX%is planning to request the General ServicesAdministration’s approval of about $268 million in additional procurement authority, most of which will be used for the interim upgrades to existing equipment. Federal Information ProcessingStandards Publications provide gui- Inadequate Project danceto federal agenciesfor effective managementof automated Management and system development projects1 The principal causeof the cost growth Oversight Contributed and scheduledelays was a lack of adherenceto these guidelines, resulting in inadequate project managementand oversight. Additional to Inventory Systems’ factors outside the agency’scontrol explain someof the cost and Cost Growth and scheduleoverruns, such as changesin legislation and agricultural policies. Schedule Delays ASCS’ Shortcomings in ASCS’initial attempts to determine both systems’ size and scopewere Determining Systems inadequate becauseit misjudged user requirements. A comparison of AN%’original user requirements and those initially produced by the con- Requirements tractor demonstrates the extent of this misjudgment. The contractor’s requirements showed about a 60-percent increasein the number of data entry screensand reports neededfor GIMS and about a 260-percent increasein those neededfor PCIMS. In other words, Ascssignificantly underestimated the size of the system it needed. One reason for this underestimate was that managementdid not involve key users in defining requirements. User representatives stated that (June 61983). Page 3 GAO/IMTJGG3046 A&wlture Aut~matcxl Systems B-240109 somekey users were not available becausethey were neededto manage an increasing work load. In addition, a division chief in ASCS’ Kansas City Commodity Office said someusers were not fully committed to helping identify requirements because,given past managementactions, they did not believe the new system would becomea reality. For example, ASCS managementhad terminated several previous develop- ment efforts becauseof conflicting demands for in-house resourcesand higher-than-anticipated bids for system development provided by contractors. Lack of managementcontrol over system design changesalso increased GIMS' cost. According to an internal agency document, contractor repre- sentatives stated that ASCS project managerspermitted users to go unchecked in submitting modifications affecting system design after the contractor began building the system. This agency document also noted that ASS’ lack of managementcontrol over system design modifications increasedsystem complexity and may have automated unnecessary functions with no cost-savingpotential. Poor Contract PCIMS’ cost increasesstem partially from ASCS' failure to thoroughly Management Contributed review the contractor’s system design document before approving and accepting it. Documentation prepared by the contracting officer noted to System Development that ASCS acceptedan ambiguous design that could only be corrected by Problems additional system changes.In someinstances,this may have led to the government’s paying extra for functions that should have been included in the original system design becausesystem changeslate in a develop- ment effort are generally more expensive than changesmade during design. ASCS’ decision to start, stop, and then restart development of PCIMS had further negative effects on the system’s cost and original completion schedule.In May 1986, after the contractor prepared the requirements and design for PRIMS, ASCS terminated this project segmentafter consid- ering several factors, including the contractor’s estimated development cost. However, in September 1987, ASCS senior managementreversed this decision and amendedthe contract to restart the project becausethe existing system technology was not meeting current needs.According to the contracting officer, the long delays between starting, stopping, and then restarting work on the contract resulted in additional costs. Page 4 GAO/EWEG9O-S6 Agricukure Automated System Interagency Project ASCS waited until after awarding the development contract for PCIMS Management Has before bringing two other Agriculture agencies-the Agricultural Mar- keting Service (AMS) and the Food and Nutrition Service (ms)-into the Contributed - . to System effort. According to ASCS officials, the three agenciesneed to share infor- uelays mation on processedcommodity inventories to accomplish their indi- vidual missions. However, the contractor noted that having three agenciesinvolved in the project increasedthe system’s complexity, size, and cost. The contractor further noted that the increase in the number of partici- pants and perspectivesthat had to be consideredin reaching key project decisionscontributed to scheduledelays and cost growth. As a result, the processused to coordinate and agreeon refinements to system requirements invited delays. For example, according to information pro- vided by project managementofficials, lack of an effective resolution mechanismslowed the agenciesin agreeingwhether FCIMS would be implemented in phasesor all at once. ASCSTop Management Another contributing factor to PCIMS’ delay was top ASCS management’s Has Not Provided Clear failure to provide clear and consistent direction. The Deputy Adminis- trators for Managementat AMSand FNSand the Director of ASCS’Kansas and Consistent Direction City Commodity Office, the principal user organization, identified this as a major problem. The ASCS official responsible for the project is the Deputy Administrator for Management.From 1983 to 1990, six people filled this position or served in an acting capacity. Decisionsto start, stop, and then restart the development of PCIMS are attributable, in part, to decisionsmade by three ASCS deputy administrators for management. Overall project direction and decision making suffered from this lack of continuity in a key senior managementpost. External Factors Have Finally, somefactors outside Ascs’control contributed to the difficulty Played Secondary Roles in in developing the two systems.Specifically, legislation such as the Food Security Act of 1986 changedGIMS’ requirements after ASCS acceptedthe Affecting - System system design. Similarly, legislative requirements such as those coming Uevelopment from the Prompt Payment Act and other agricultural policy changes, including commodity packaging and shipping period initiatives, affected PCIMS requirements after design acceptance.However, ASCS records and our discussionswith project managementofficials indicate that these Y legislative and policy changeswere not the principal causesin the two systems’ cost growth and scheduledelays. According to ASCS records, at Page 6 GAO/IMTEG@O-gS Agriculture Autimated 8yrtema B240109 most about $7 million of the nearly $65 million in contract cost growth for both systems was due to these new requirements2 is installed and operating, and several users told us that the new Problems Remain GIMS system is an improvement over its predecessor.Nevertheless,it does not Despite Significant meet some important inventory managers’ needs that were deleted in Investment 1986 to conserve contract funds. For instance, besides keying data into GIMS, users key some of the same data into separate personal computer- based systems to prepare management information reports. ASCSis also using in-house resources and another contractor to enhance GIMS to satisfy user needsthat were not met when the system was implemented in 1987. Through June 1990, the agency incurred about $5 million in direct costs for in-house programmer staff and had obligated about $1 million in funds for the other contractor to continue developing and modifying GIMS. The three user agencieshave no available meansto resolve potential disagreementsthat may arise during PCIMS acceptancetesting, installa- tion, and maintenance. According to the AMSDeputy Administrator for Management,each agency seeksits own interests rather than a common goal becausethe project managementstructure has offered no strong central control over project development. Representativesof ASCS and FNShad similar concerns. Having an effective project managementstructure in place to provide direction, control, and oversight is important during the rest of the sys- tems’ implementation phasesas well as the remaining phasesof opera- tion and maintenance. According to their representatives, each agency expects to correct or enhancethe computer programs during the system’s life cycle, but they have no final agreementon orderly and cost-effective software maintenance. Federal guidance states that estab- lishing a structured software maintenance policy is a vital step in cre- ating and supporting an orderly processin which all requested changes are formally submitted, reviewed, ranked, and scheduled.3Such a main- tenance policy is necessaryto ensure that all changesare reviewed for their effect on the whole system. 2ASCSrecords indicate that at most about $6.6 million of the nearly $13.6 million cost increase for GIMS and about $1.6 million of the approximately $41.6 million cost increase for PCIMS can be attrib- uted to new legislative requirements. 3Guidelines On Software Maintenance, Federal Information Processing Standards Publication 106 (June 16,1984). Page 6 GAO/IMTEG9O-S5 Agriculture Automated Systems Just as information managementweaknessescontributed to problems WeaknessesExist in with AS@inventory systems,they present problems for enhancements ASCS’County Office being planned for its county office minicomputers. In May 1990, ASCS Minicomputer Disk planned to request the General ServicesAdministration’s approval of about $268 million in additional procurement authority through Capacity Management December1996. This procurement authority is mainly for more mini- computer disk drives and equipment upgrades at its county offices. To determine whether these additional purchasesare justified, we examined ASCS’ procedures used to managedisk capacity for its county office minicomputers. We also evaluated disk file use for each county office. Wasteful Disk Utilization The ASCS KansasCity ManagementOffice (KCMO) is responsible for man- Found on County Office aging county office minicomputer disk capacity and procuring addi- tional disk capacity for the offices. ASCS county offices are staffed with Minicomputers employeeswho do not report directly to KCMO. Most of these offices do not have staff with highly technical computer skills. KCMO centrally managesADP support for the county offices, including procuring, writing, testing, and installing the computer software and related equip- ment used to perform administrative and operational functions, In this regard, KCMO managesdisk capacity at the county offices by peri- odically monitoring directories listing all computer programs and data files stored on the offices’ disks, KCMO staff purchase, develop, and revise the software run on the county minicomputers and maintain an official list of programs and associateddata files that should reside on the disks. KCMO updates this list quarterly and supplements it through discussionswith programmers, KCMO comparesthis list against county office directories, identifies files for deletion, and tells the counties to purge these files. In addition, storage of county-generated data files, such as word processingfiles, is monitored by KCMO. If they are unusu- ally large, KCMO recommendsthat they be off-loaded to diskettes or tape. KCMO performs capacity managementcheckswhen county offices pass a 70-percent disk utilization threshold. KCMO purchases additional disk storage upgrades only after checking that the county disk drives do not contain program or data files that either should be deleted or off-loaded to diskettes or tape. We examined file directories for the minicomputer disks in each ASCS county office to determine how efficiently disk spacewas being used. As of June 1990, computer programs and files meeting ASCS criteria for Page 7 GAO/lMTECQOI5 Agriculture Automated Systems deletion or off-loading to temporary storage devices(i.e., floppy disks or computer tape) accounted for about 12 percent of the total disk space used by all ASCS county offices.4On a county-by-county basis, we found that as little as 2 percent and as much as 36 percent of the disk space used was occupied by files identified for deletion or backup storage on an alternative medium. Table 1 presents a frequency distribution for specified ranges of disk spaceoccupied by these files. Table 1: Dlrtributlon of County Offlce Disk Space Occupied by Flies Identified Percentage of disk space occupied Number of counties for Deletion or Alternatlve Storage 5orless 155 6tolO 1,241 11to15 982 16to20 321 Over20 83 Total 2,782 In January 1990 KCMO took steps to improve disk capacity management procedures.For instance, it is planning to improve disk capacity man- agementprocedures by providing a computer program to assist county office staff in purging files from disks. In addition, KCMO is providing more comprehensivewritten guidance on methods that county employeescan use to help better use and managedisk capacity. These are positive steps, and the changescould improve the accuracy and timeliness of KCMO'S disk capacity monitoring. However, it is unclear whether they will result in more effective disk spaceuse in the county offices. The capacity monitoring processis not resulting in unneeded files (those identified for deletion or migration to other storage devices) being promptly purged on a regular basis. KCMO can tell the counties to purge unneededfiles, but it is the power of KCMO to deny purchasesof additional disk drives that appears to serve as the real incentive for the county offices to take disk managementactions. 4To identify the computer programs and files that could be deleted or temporarily removed from county office disks, we used AS3 criteria as of June 1990. (See app. III for more details.) At the completion of our audit work in August 1990, KCMO was evaluating the criteria to determine whether changes should be made. Page8 c * I B240109 ASCS’ Projections for ASCS’ method for projecting future county office minicomputer disk capacity requirements is basedupon current inefficient disk use. Fur- Interim Upgrades Are thermore, new farm legislation expected to be enacted soon may signifi- Questionable cantly changethose requirements. As a result, ASCS lacks reliable information to justify its request for interim computer equipment upgrades. ASCS uses a computer model to estimate its future disk storage, computer upgrade, and associatedmaintenance needs.6The model is basedon his- torical trends in disk capacity use and estimates of disk spacerequire- ments for major new software planned for the county office minicomputers. These trends are projected into the future to identify equipment needsand support requests for procurement authority. ASCS’ projections can result in questionable results for two reasons.First, ASS projects wasted disk spaceinto future periods. By not eliminating these files from ASCS’ projections, inaccurate estimates of disk capacity requirements and associatedequipment needsoccur. Second,ASCS does not use an overall file growth rate for an historical baseperiod in pro- jecting disk capacity for the future. Instead, AS(IS, uses assumptions and procedures-which are not fully documented and which we were unable to fully evaluate-that attempt to assessthe impact of major new software it plans to install. However, since AXS doesnot know its software plans beyond 1991, it would appear more reasonableto use an overall growth rate when projecting future capacity needs.This simple, straightforward solution assumesnew software will be introduced at rates determined by previous growth patterns. Additionally, the 1990 farm bill may require AS.Xto implement new farm programs, revise some,and discontinue others. This, in turn, may affect future minicomputer disk capacity needsbecauseof changesin number and size of files. Consequently, AK% future disk requirements may differ substantially from those indicated by the historical trends. To accommodateunpredictable requirements changessuch as legislative modifications, ASCS now builds a 25 percent additional growth param- eter into its projections at the suggestionof Agriculture’s Office of Infor- mation ResourcesManagement.This figure is based,in part, upon the agency’sobservation of past effects on capacity requirements resulting from legislative changes,and doesnot appear to be unreasonable. “Our ability to fully evaluate A!33 model was limited because ASCS never prepared complete ’ written documentation such as logic charts, detailed narrative descriptions, or flow charts to describe the methodology used to develop the model. Page 9 GAO/IMTEG99-S5 Agriculture Automated Systems -- . . -.- _____- .-_-..~.- --. Estimates for Interim As of June 1990, the agency had about $26 million remaining under its Minicomputer Upgrade existing procurement authority for minicomputer equipment upgrades, maintenance, and software in its county offices. This authorization Funding Through Fiscal extends through September 1992. In May 1990 ASCS drafted procure- Year 1992 Are Unreliable ment requests indicating the agency neededan additional $67 million in procurement authority for minicomputer equipment upgrades,mainte- nance, and software between July 1990 and September 1992.”This esti- mate includes about $32 million for additional disk capacity. The estimate consistsof (1) an additional $63 million of additional procure- ment authority under the existing contract, and (2) $4 million under a new contract to get larger computers for those county offices that require more disk storage than current systems can provide.7But these estimates are derived mainly from the model’s questionable projections. Given our concernswith ASCS’procedure for estimating future disk storage and equipment upgrade needs,we made our own estimate. We made calculations basedon more judicious disk usageand historical rates of growth for software and data files, and then used this to project future needs.(Seeapp. III for details.) Our estimates indicate that ASCS funding requirements for minicomputer equipment upgrades, mainte- nance, and software for the remainder of fiscal year 1990 through fiscal year 1992 would be approximately $20 million. Our projections assume that the average disk storage trend in growth for the past 2 years would continue through 1992. If we increaseour funding estimate by 26 per- cent to account for automation changesAXS may face after such events as the passageof the 1990 farm bill (the practice employed by ASCS for its own funding estimates), the figure would rise to about $25 million. This increase could be met within AX% remaining procurement authority, eliminating the need for the additional $57 million in procure- ment authority ASCS is requesting. In May 1990 ASCS also estimated that, for fiscal year 1993 through the first quarter of fiscal year 1997, it will need about $143 million for interim equipment upgrades, maintenance, software, and support ser- vices. However, we limited our estimate of AXX’ needsto the 1990 to 1992 period becausethe new farm bill may substantially changethe agency’s future equipment needs.Before the end of 1992, ASCS should be (‘We used A!X% actual expenditures through June 1990 and adjusted the procurement request to eliminate estimated funding needs before that date. ‘ASCY is pursuing alternatives to acquiring these larger computers. One alternative involves changes to the IBM System 36 processors to increase the maximum disk capacity the systems can support. Page 10 GAO/IMTEG9O-85 Agriculture Automated Systems c *” E240108 able to evaluate the farm bill’s impact and modify its request for pro- curement funds to more accurately estimate computer equipment upgrades and associatedmaintenance needsfor fiscal years 1993 through 1996. Conclusions Our review of two major automation projects raises considerableques- tion about ASCS’ managementof information systems development and enhancementprojects. The grain and processedcommodity inventory systems support an important ASCS responsibility involving the manage- ment of large commodity inventories worth billions of dollars. Poor planning, insufficient requirements analyses,and faulty contract man- agementhave kept ASCS from finishing the systems on time, staying within original cost estimates, and meeting important user needs.These problems are contributing in a major way to ASCS spending somenine times more in systems’ development costs for the two projects than was originally estimated. Unless these deficiencies in information resources managementare corrected, future information systems projects may suffer similar problems. The capacity managementprocessASCS usesfor its some2,800 minicom- puters housed in county offices acrossthe country likewise contains deficiencies. An improved disk capacity managementprogram would help ASCS better manageexisting disk spaceusageand more accurately identify requirements for future disk capacity needs.Improvements in its minicomputer capacity and performance monitoring could lessenthe need for the additional equipment purchasesplanned through 1992. Until ASCS corrects shortcomings in its disk managementprocedures and the methodology used to estimate future minicomputer equipment needs,risks increasethat additional ASCS funding requests will not be fully justified. More important, these problems raise concernsfor ASCS’ upcoming effort to replace all county office computer equipment by the end of 1996. This project will be one of the most significant automation undertakings the agency will face in the 1990s.The delay in starting this effort has resulted in the need for millions to be spent on interim computer upgrades. In constructing its approach to this major agency automation project, ASCS will need to resolve the weaknessesfound in our review of important information systems development projects. Page 11 GAO/IMTEGBO-SS Agriculture Automated Systems c r E240109 Becauseof the managementand oversight problems evident in these Recommendations ASCS automation projects, we recommendthat the Secretary of Agriculture l direct the Administrator, ASCS, to improve project managementand oversight for new information systems development or enhancement projects by ensuring that (1) information needsare sufficiently identi- fied to describethe scopeand magnitude of software systemsprojects before advancing to the development phase, and (2) project size and scopeare reasonably controlled; l direct the Administrators of ASCS, AMS,and FNS to complete an agreement governing software maintenance for PCIMS and establish an effective res- olution processto arbitrate interagency disagreementsto ensure orderly and timely system implementation and maintenance; . direct that ASCS’fiscal years 1990-92funding request for county office minicomputer upgrades, maintenance, software, and other equipment be reestimated and that any additional funding neededbeyond the approxi- mately $26 million remaining in existing procurement authority be con- tingent upon correcting the flaws in ASCS’forecasting to more accurately estimate equipment upgrade needs;and . direct that approval of AHX’ requests for county office minicomputer equipment procurement authority for fiscal year 1993 and beyond be contingent upon the agency’simplementing a more effective disk capacity managementprogram, as verified by the Department’s Office of Information ResourcesManagement. We conducted our work in accordancewith generally acceptedgovern- ment auditing standards. The views of ASCS officials were sought during the course of our work and their informal commentswere obtained on this report. We have incorporated these commentswhere appropriate. At your office’s request, we did not obtain written ASCS commentson a draft of this report. As arranged with your office, unless you publicly announcethe contents of this report earlier, we plan no further distribution of the report until 30 days from the date of this letter. At that time, we will send copiesto the Secretary of Agriculture, the Administrators of the Agricultural Sta- bilization and Conservation Service,Agricultural Marketing Service, and Food and Nutrition Service, and other interested parties. We will also make copies available to others upon request. Page 12 GAO/IMTEG90-85 Agriculture Automated Systems \ *iI’ \ E240109 This work was performed under the direction of JayEtta Z. Hecker, Director, Resources,Community, and Economic Development Informa- tion Systems,who can be reached at (202) 275-9676.Other major con- tributors are listed in appendix IV. Sincerely yours, Ralph V. Carlone Assistant Comptroller General Page 13 GAO/lMTEG90436 Agriculture Automated Systems I Contents Letter Appendix I 16 Objectives, Scope,and Methodology Appendix II Dates of Key Events and Cost Estimates for GIMS and PCIMS Appendix III GAO Methodology for Estimating ASCS County Office Minicomputer Upgrade Needs Appendix IV Major Contributors to This Report Table Table 1: Distribution of County Office Disk Space Occupiedby Files for Deletion or Alternative Storage Page 14 GAO/lTMTEC9O-S5 Agriculture Automated Systems I Content43 Abbreviations ADP automated data processing AMS Agricultural Marketing Service ASCS Agricultural Stabilization and Conservation Service FNS Food and Nutrition Service GAO General Accounting Office GIMS Grain Inventory ManagementSystem IBM International BusinessMachines IMTEJC Information Managementand Technology Division IRM information resourcesmanagement KCMO KansasCity ManagementOffice PCIMS ProcessedCommodity Inventory ManagementSystem Page 16 GAO/IMTECBO-S5 Agriculture Automated Systems ; ‘h / Appendix I Objectives, Scope,and Methodology The objectives of our review were to provide information on AX% (1) contract to develop two large inventory managementsystems,GIMS and PCIMS,and (2) planned interim county office computer equipment upgrade. We focused on the inventory systems’ original and current esti- mated completion dates, reasonsfor cost increasesand scheduledelays, and whether the two systems meet users’ needs.We were also to provide information on the methods ASCS uses,or could use, to determine when county office computer upgrades are needed,and the actions taken or available to improve the use and extend the life of county office com- puter equipment. To accomplish these objectives, we interviewed and obtained documents from officials and personnel at the Agricultural Stabilization and Con- servation Service in Washington, D.C., and at ASCS’Kansas City Manage- ment Office and Kansas City Commodity Office. In addition, we interviewed Department of Agriculture Office of Information Resources Managementofficials, Agricultural Marketing Service officials and per- sonnel, and Food and Nutrition Service officials and personnel in Wash- ington, DC. To respond to the questions on the development of these two inventory systems, we reviewed pertinent contract, agency procurement, and sys- tems design documents. We also reviewed documents relating to AS& development and installation of GIMS and FCIMS. We interviewed a con- tractor representative involved in these system development efforts. We also interviewed users of the systems at ASCS,AMS, and FNSto determine how well their needsare being met. To respond to questions about the planned interim computer upgrade, we reviewed federal regulations identifying actions agenciesshould take in planning, acquiring, and using computer resourcesas part of an agency’scapacity managementactivities. We also determined ASCS’cur- rent and planned disk capacity managementactivities for county office computers and the processused to determine the need for the interim computer upgrades at these sites. We analyzed the impact of improving disk managementat the county offices and reviewed ASCS' methodology for projecting equipment needs.Finally, we independently estimated ASCS’future equipment funding needs. Our audit work was conducted between September 1989 and August 1990. We performed the review in accordancewith generally accepted government auditing standards. Page 16 GAO/IMTE.G9O-86 Agriculture Automated Systems _..__... -..._ -.__ Appendix I Objectives, &ope, and Methodology We did not obtain official agency commentson a draft of the report. However, we discussedits contents with responsible ASCS officials, and have incorporated their comments where appropriate. Page 17 GAO/lMTEC!#O-S5 Agriculture Automated Systems + ;*h Appendix II Dates of Key Events and Cost Estimaks for ’ GIMS and PCIMS Dollars in millions Percentage Estimated CosV Increase Over Cumulative Initial Estimated Date . ._- __.-... - Event GIMS PCIMS Total cost Dec. 1983 Contract awarded $4.4 $2.6 $7.0 . Jan. 1986 Additional GIMS costs after user requirements definition and design prepared 5.8 l 12.8 80 tvlay 1986 PCIMS development terminated after user requirements definition and design prepared . . 12.8 80 Aug. 1986 Additional GIMS user requirements identified, plus new requirements as a result of legislation 7.7 l 20.4 190 Mar. 1987 GIMS phased installation completed . . 20.4 190 Sep. 1987 Contract modified to restart PCIMS and to reflect design requirements of ASCS, AMS, and FNS . 18.9 39.3 460 Feb. 1989 PCIMS requirements identified after reviewing design and adding functions originally planned to be developed with in-house resources . 7.1 46.4 560 Apr. 1989. Additional PCIMS requirements identified at acceptance testing July 1990 . 6.2 52.6 650 Aug. 1990. Planned contract increase for PCIMS implementation support July 1991 . 9.3 61.9 780 July 1991 PCIMS expected to be installed . . 61.9 780 Totalab 917.8 $44.1 $81.9 780 %epresents projected cost of completion at specified time. Additionally, for ease of presentation, some events are combined and only major cost factors shown. bTotals may not add due to rounding Page 18 GAO/IMTEC9O-86 Agriculture Automated Systems ApprmthixIII GAO Methodology for Estimating ASCSCounty Office Minicomputer Upgrade Needs We independently estimated ASCS’ minicomputer disk and associated funding needsfor the remaining portion of fiscal year 1990 through fiscal year 1992. In basic terms, our methodology involved the following steps. First, we calculated the disk spacecurrently being used by each county office, assumingthe removal of files and software ASCS identified for deletion or migration to other media, such as tape drives or dis- kettes. Next, we calculated an estimate of disk spacegrowth for all county offices basedon historical data from the past 2 years, a period for which ASCS could readily provide the requisite data. We computed the averagemonthly growth in megabytesover the 2-year period and, using linear projections, estimated future county office disk storage needsthrough September 1992. Using these results, we estimated the cost of purchasing and maintaining additional computer equipment as well as the currently installed equipment and other related costs through September 1992. Each step is further explained below. To determine the data and software files that could be deleted or tempo- rarily migrated to other devices,we relied upon ASCS criteria. These files included (1) those identified by ASCS for deletion, (2) those identified by ASCS guidance as temporary files not requiring storage on disk, and (3) county office training files that ASCS instructions state can be migrated from disk when spacebecomeslimited. To determine the growth estimate for future disk spacerequirements, we obtained ASCS monthly data on averagedisk spaceuse for all coun- ties between May 1988 and May 1990. Using this information, we esti- mated the average amount of disk storage growth for all files at all county offices using linear regressionanalysis.’ To account for differ- encesin monthly disk spacegrowth for individual counties, we multi- plied the averagegrowth rate by the ratio of each county’s existing disk spaceusageto the overall averageexisting disk spaceusagerequire- ment.2For example, if the averagegrowth rate for all counties was 5 megabytes per month and the overall averageexisting disk usagewas 600 megabytes,the estimated growth rate for a county whose existing ’ Linear regression analysis is a statistical technique for evaluating a linear relationship of one or more independent variables to a single continuous dependent variable. It can be used to determine the best interpretive mathematical model for describing this relationship, and subsequently estimating future values of the dependent variable. We used time series regression analysis because of the tem- poral ordering of the independent variables (periods of tie), which allows one to see how the depen- dent variable (megabytes of disk space) has varied in the past and how it is likely to vary in the future. “An ASCS Kansas City Management Office assistant deputy director agreed that this approach for estimating disk space growth is appropriate and should provide a reasonable estimate of ASCS needs. Page 19 GAO/IMTEC-SO-95 Agriculture Automated Systems Methoflology for Edmating ASa County Offlee Minicomputer Upgrade Needa disk spaceusagewas 400 megabyteswould be 4 megabytesa month (400/600 times 6 megabytes). We calculated each county office’s disk needsfor each month between June 1990 and September 1992, by (1) calculating disk spacethat would remain as of June 1990, after removing from each county office’s disk those files ASCShas identified for deletion, and (2) increasing each office’s disk spaceusageby the county’s estimated monthly growth rate, After computing each county office’s disk spaceneeds,we estimated the funding requirements for this period using AS& equipment cost infor- mation The projected funding estimates include the cost for computer and equipment upgrades plus other costs,including the maintenance of existing and new equipment, system software, and other equipment such as printers and terminals. Using this methodology, our estimates indicate that ASCS funding requirements for these items for the remainder of fiscal year 1990 through fiscal year 1992 would be approximately $20 million. ASCS’May 1990 draft procurement request indicates that it needsabout $83 million for these items for this period. The agency has approximately $26 mil- lion in procurement authority remaining until the end of fiscal year 1992. Thus, the agency’sestimates indicate that it will neeti an addi- tional $67 million through the end of fiscal year 1992. Our analysis questions the validity of ASCS’ estimates of additional minicomputer disk storage needsthrough fiscal year 1992 and the associatedrequest for an increasein procurement authority. Our methodology sumesthat the disk storage growth rate computed from the past 2-y ar period would continue through 1992. If we increaseour estimate bf 25 percent to account for new requirements imposed by imponderable events, such as passageof the 1990 farm bill (the practice employed by ASCS in its own funding estimates), our estimate would rise to about $26 million. This would be far less than the $67 million in additional procurement authority ASCS is requesting to cover July 1990 through September 1992 and would amount to about $1 million less than ASCS’ remaining procure- ment authority. Page 20 GAO/IMTEC90-95 A@icultwe Automated Systems . *p Appendix IV Major Contributors to This Report Theodore C. Gearhart, Assistant Director Information David L. McClure, Assignment Manager Management and Dennis L. O’Connor, Reports Analyst Technology Division, Washington, DC. -Kansas City Regional CeceliaM. Ball, Staff Evaluator Office Thomas M. Codk, Staff Evaluator Doris J. Hynes, Staff Evaluator Jerry D. Hall, Computer Analyst (510424) Page 21 GAO/IMTECBO-EK Agrkmlture Automated Syeteme ,““1”1~- - -- __-_... - _....... I-“.III_III.I. -II. - -~..---I.-^.1-__1 ---_---I.-I- ~ ..I .~.e-...-- ---.-~_.- _-__- -- Orcltviug Inforruatiou I J.S. General Accounting Office I’.<). Hex 6015 Gaithersburg, MD 201377 Orders may also be placed by calling (202) 2756241.
Information Resources: Management Improvements Essential for Key Agriculture Automated Systems
Published by the Government Accountability Office on 1990-09-12.
Below is a raw (and likely hideous) rendition of the original report. (PDF)