|
|
| GSBCA No. 7927-P
1985 WL 16860 (G.S.B.C.A.), 85-3 BCA 18,289, GSBCA No. 7927-P GSBCA Protest of Appearance for Protester
Appearance for Defense Logistics
Agency Opinion by Administrative Judge Lieblich The issue raised by this protest is whether certain provisions of a solicitation for computer disk drives exceed the Government's minimum needs and unduly restrict competition. The protester challenges three aspects of the solicitation: (1) a "performance requirement" that the disk drives average no more than one hard failure per 35,000,000 input/output executions; (2) a provision calling for sixteen or thirty-two megabytes of cache storage and adding a substantial dollar penalty in the evaluation of those proposals that do not offer cache; and (3) a provision requiring the contractor to give the Government a credit equal to the entire monthly maintenance charges for a given data processing site if the performance of the disk units at that site drops below the level prescribed in the solicitation. We conclude that the Government
has overstated its minimum needs by imposing an absolute performance
requirement that is justified by neither its actual requirements nor
the reliability measurement it uses. We further conclude that it has
unduly restricted competition with the specific provision for a maintenance
charge credit that it has inserted into the solicitation. To that extent
we grant the protest. We do not find the cache storage provision objectionable. Findings of Fact 1. The solicitation in dispute was issued by the Defense Logistics Agency (DLA) on March 29, 1985, with a closing date of April 24, 1985, subsequently extended by amendment to May 7, 1985. Protest File, Exhibit 3. The general description of DLA's requirements, found in paragraph B.1 ("Scope") of the solicitation, is as follows: [FN1] DLA has a requirement for state-of-the-art, high-capacity, high-performance direct access storage devices providing approximately 450 billion bytes of online storage capacity to be installed at multiple DLA sites. The major program addressed by this procurement is the Agency's modernization of APCAPS, SAMMS and MOCAS Automated Information Systems (AISs). This Agency modernization effort includes supporting a state-of-the-art data base, its associated files and providing additional online access to the data base information by DLA and other users of these AISs. This solicitation is for the acquisition of an estimated 186 IBM 3380 or equivalent Direct Access Storage Devices and approximately 88 IBM 3880 or equivalent Associated Control Units for the Defense Logistics Agency. The offeror shall be prepared to furnish all hardware, software, installation, documentation, and maintenance necessary to this acquisition. Descriptions/Specifications can be found in Section C. A direct access storage device (abbreviated "DASD" and pronounced "dasdee)" is also commonly referred to as a disk drive. . . . . IV. Disputed Provisions of the Solicitation. 28. The solicitation was issued March 29, 1985, with a closing date of April 30, 1985, extended to May 7, 1985, by amendment. Protest File, Exhibit 3. The provisions of the solicitation challenged by the protester are in several parts of the solicitation. From Part C, "Description/Specs/Work Statement," as amended by Amendments 2 and 3 to the solicitation, come the following: C.2.1.1 Initial Performance Requirement For the proposal to be considered acceptable all DASD equipment proposed to meet the DLA requirements must have achieved the following performance levels. a. 3380 compatible equipment must have achieved an average of 35,000,000 I/Os [inputs/outputs] per hard failure for three months out of the four month period of Feb 1985--May 1985. b. These performance levels are calculated by the Reliability Plus proprietary software product and are extracted from the monthly DLA Reliability Plus database report. C.2.4.1 Cache Storage The proposed storage controllers shall provide the electronic cache storage capability of 16 MB [megabytes] and 32 MB. This optional capability will be required on approximately 25% of the storage controllers. Cache storage allows a copy of the most recently used data to be retained in storage where it is available when next requested or the space is needed for additional data. Proposals may offer on cache memory, one or the other size of cache memory, or both sizes of cache memory. Proposals that offer no cache memory or only one size of cache memory will be assessed a cost(s) as described in Section M.5.9.6. 29. From Part E, "Inspection and Acceptance," comes: E.5.1 Performance Standard: Each disk system shall operate for a period of 30 consecutive days and achieve an average pool of equipment performance level of 35,000,000 I/Os per hard fail for 3380s as calculated by the Reliability Plus proprietary software package. Acceptable performance will be achieved if the disk drives have less than 35,000,000 I/Os for 3380s and experience no hard failures. 30. From Part H, "Special Contract Requirements," come: H.3.2 Performance Period The first day of the initial performance period shall be the first day of the month following the month in which the acceptance test was successfully completed for the equipment supplied. A performance period is defined as the lesser of: a. Four months; or, b. The number of months required for a DLA Data Processing Installations (DPA) pool of installed DASD to achieve an average of 40 million I/Os. A new performance period starts on day one of the fifth month or day one of the month following the period in which the pool of DASD achieved an average of 40 million I/Os. H.3.3 Performance Measurement Methodology DLA will use the following methodology to measure performance of disk units supplied by the offeror and in the use at each DLA DPI. a. Data from each DPI will be used in measuring performance in each performance period. Data source for computing performance will be the SYS1.LOGREC data set at each DLA DPI. b. Disk equipment performance at each DLA DPI will be computed by using the Reliability Plus proprietary software package from UCCEL, formerly the University Computing Company of Dallas, Texas. H.3.4 Performance Computation Performance will be computed each month using Reliability Plus for the pool of installed disk equipment (Control units and Head Disk Assemblies (HDAs)) at each DLA DPI. The performance computation will include data for all 3380 disk equipment installed at the DLA DPI. H.3.5 Performance Evaluation Each DLA DPI's disk equipment performance will be evaluated based on the use- to-hard-failure-rate figures extracted from the monthly Reliability Plus database reports. Each using DLA DPI shall be responsible for maintaining its own equipment performance reports from Reliability Plus, and performance reports shall be made available to the contractor. V. Reliability Plus and Why DLA Chose to Use It. A. "R+" and What it Does. 31. Reliability Plus is a proprietary computer program developed and owned by Reliability Research, Inc. (RRI). The name of the program is almost universally abbreviated "R+" and spoken "R Plus." Mr. Larry Martin, one of the two developers of the program, testified at the hearing. He said the primary reason for developing the program was "to give users the capability of selecting hardware on the basis of how reliable it is." Transcript at 253-54. . . . .
Discussion Two of the three issues in this protest (R+ and cache) have been couched in terms of the Government's "requirements" or "minimum needs." In our view, however, the R+ issue is only masquerading as a requirements issue, which is why we grant the protest as to that issue. The cache issue, on the other hand, is a legitimate requirements issue, and we find no basis for overturning DLA's cache requirements. The third issue (maintenance fee credit) relates to whether the provision is a penalty in the guise of a liquidated damages provision. We hold that it is and sustain the protest on this ground as well. I. Standard of Review. Our jurisdiction over this protest derives from Section 2713 of the Competition in Contracting Act (CICA), Pub.L. No. 98-369, 98 Stat. 1182-84, 40 U.S.C. s 759(h), which amends the Brooks Act, 40 U.S.C. s 759 (1982). Subsection (h)(5) of the Brooks Act, added by the CICA, provides in part: (A) In making a decision on the merits of protests brought under this section, the board shall accord due weight to the policies of this section and the goals of economic and efficient procurement set forth in this section. (B) If the board determines that a challenged agency action violates a statute or regulation or the conditions of any delegation of procurement authority issued pursuant to this section, the board may suspend, revoke, or revise the procurement authority of the Administrator or the Administrator's delegation of procurement authority applicable to the challenged procurement. . . . . So long, therefore, as we are in an area where an agency has legitimately determined its requirements, we cannot interfere. But we have underscored "legitimately" to make clear that an impermissible limitation on competition in the guise of a requirements determination is as vulnerable as any other method of unduly restricting competition. The GAO decisions bear this out. Compare, e.g., Storage Technology Corp., B-215336.2 (August 17, 1984), 84-2 CPD p 190 (procuring agency's technical conclusions as to its minimum needs are entitled to great weight and will be accepted unless there is a clear showing that they are arbitrary) with Memorex Corp., B-213430 (July 9, 1984), 84-2 CPD p 22 (preclusion of 3350 class double-density drives lacked technical justification and therefore unduly restricted competition). If one reads a number of GAO decisions in this area (and we have), one develops a sense of two lines of cases that have at least a hint of inconsistency. When GAO agrees with an agency's determination of its requirements, it says, as it did in Storage Technology, that it will upset such a determination only upon a strong showing. But when it disagrees, as it did in Memorex, it says the agency may only solicit its minimum needs and gives no more than lip service to how strong a showing must be made to upset an agency's requirements determination. We think the inconsistency, if there is one, lies more in the language of the decisions than in their results. . . . . Our substantive standards of review should derive from our procedures and from the burdens assigned in the foregoing paragraph. Reverting to the standards followed by GAO in "requirements" and "minimum needs" cases, we hold that the substantive standards we will apply in considering the R+ and cache issues in this protest are (1) we will not take issue with an agency's narrowing of competition in pursuit of legitimate agency requirements but (2) we can and will overturn those requirements that improperly limit competition. We will accord deference to the agency's technical judgment, but we will not slavishly follow it where the result is lacking in justification. The burden is on the protester to establish by a preponderance of the evidence that the agency has impermissibly narrowed competition. There are limits on competition inherent in all procurements. A decision to purchase an electronic pocket calculator excludes the mechanical analog calculator and the abacus. A decision to buy an automobile excludes the bicycle and the airplane. Everyone takes these sorts of limitations on competition for granted, and common sense thus suffices to dispose of virtually all complaints against limits on competition. Even some calls that might seem close evoke no objections; here, for example, the Burroughs Corporation, parent to the protester Memorex, has brought no protest against the requirement in the solicitation that the disk drives supplied be IBM or compatible. The issue of limits on competition should arise only where the protester can make at least a colorable argument that the exclusion of its product or services narrows competition more than is necessary. Given that most exclusions of competition are inherently reasonable and incontestible (it is hard to imagine an abacus maker's protest against the procurement of electronic calculators surviving a motion to dismiss as frivolous), we deal in protests only with those that are at least potentially unreasonable. These considerations, we suspect, have led to at least some of the strong language used by GAO when denying protests under its standard of review. But there is really only one issue: are the agency's needs such that it acted properly--reasonably, legitimately, permissibly--in narrowing competition through the provisions of its solicitation? Add negatives where appropriate and you have what the protester must establish by the preponderance of the evidence. That is the standard we will follow in this and other cases where that is the issue. II. The Specific Issues A. Reliability Plus. In an attempt to make this look like a requirement, DLA has even given it the label "Initial Performance Requirement." This barely qualifies as a nice try. What DLA is asking for is not a given level of performance but a track record, and a track record measured not by DLA or any other Government agency but by a third party. As icing on the cake, DLA has made the test all-or-nothing and has disqualified all who fail to meet it. DLA has cited us to Storage Technology Corp. for the proposition that "equipment reliability is a proper factor to consider in selecting ADPE." Respondent's Brief at 49. We have no quarrel with that proposition. The problem is not DLA's goal but its implementation. The facts of Storage Technology help clarify the point. In Storage Technology, the procuring agency was the Veterans Administration. In its solicitation it listed seven technical factors that it would consider, in addition to price, in making the award, first of which (and most heavily weighted) was "peripheral equipment reliability." To determine reliability it planned to survey some of the data processing installations of each offeror, selected so as to reduce the chances of bias, to obtain LOGREC data on failures, and to apply various factors to normalize the results. GAO ruled that an agency's determination of its minimum needs must be accepted unless there is a clear showing that the conclusions are arbitrary, that reliability was a proper factor for consideration, and that the proposed survey did not appear unreasonable. It therefore denied the protest. DLA has gone well beyond what the VA did in Storage Technology, and in so doing it has gone beyond what is permissible. It has both made reliability an all-or-nothing proposition, rather than one factor for evaluation, and used a third party's figures rather than conduct its own review. By this combination of actions it has abdicated all discretion and delegated the eligibility decision to an outsider. We are not saying that this sort of conduct is universally impermissible, but it certainly is in this case. . . . . What DLA has done is to hang the "minimum needs" label on the R+ ratio of at least 35,000,000 I/Os (actually "accesses") between hard fails. DLA has given us no reason to believe that it needs disk drives that meet this level of performance. (See Finding 50.) We do not question the importance of DLA's mission or its need for reliable disk drives. We see nothing wrong with a procedure such as was followed in Storage Technology, where the agency made its own reliability determinations and used them as one factor in evaluation rather than a make-or-break criterion. We probably would not even take exception to the use of Reliability Plus figures, properly normalized and evaluated, as one measure of performance in an evaluation factor for reliability similar to the one used by the VA in that case. The fault lies neither in wanting reliable hardware nor in making reliability a factor in the procurement. Rather, where DLA erred was in taking a third party's standard and making it an absolute criterion of eligibility for award. This goes beyond a determination of minimum needs into the proscribed area of unduly restricting competition. Nothing in the procurement statutes and regulations is more fundamental than the requirement to maximize competition. An undue restriction on competition is a violation of that requirement. It lies within the remedial jurisdiction of this Board under the CICA. The specific remedy is considered in the next principal section of this opinion. B. Cache Storage. This is a very different issue from R+. This fits the statutory definition of "selection of the types and configurations of equipment needed," one of the areas in which the Administrator is not supposed to "impair or interfere with the determination by agencies." 40 U.S.C. s 759(g). We differentiate it from R+ not because of the way the requirement is phrased in the solicitation but because of the nature of what is being procured. A cache is an item of hardware, part of the equipment configuration. Its purpose is to provide a point of storage intermediate between DASD and processor memory for items that are frequently "hit" and that have a high read- to-write ratio. While cache may not be the only way of solving a particular problem in the design or operation of a large data processing system, and while it may actually impair the operation of a system if used in situations where no data presents the right combination of frequent hits and high read-to-write ratio, the protester has not established by the preponderance of the evidence that DLA has unreasonably limited competition by requiring cache memory in some of its disk drive controllers. The GAO Memorex decision held that there was no basis for soliciting only single-density drives and not double-density as well. The reason for the decision was a technical one: the agency contended that the double-density drives were more likely to encounter interference in the competition for the data on them, but GAO found that the likelihood was so slight with either sort of drive, and the consequences so minimal, that there was no legitimate basis for the requirement. Had GAO found otherwise, it would probably have said an agency must be free to determine its own requirements and denied the protest. Similarly, if we were persuaded that there is no legitimate basis for the cache requirement, we would cease our deference to the agency's determination of its requirements and rule as we did on the R+ issue. What we are saying, therefore, is that we do not think DLA unjustified in asking for cache, and we cannot tell it to strike the provision from the solicitation. The protester suggests that DLA's one apparent success with cache was tainted because the processor was changed at the same time. It further contends that the particular application of cache was for a read-only file that can as easily be handled by a solid-state disk. As narrow points, these sound persuasive, but in the big picture they are far from compelling. This is a procurement for a requirements contract covering an estimated 186 disk drives and eighty-eight controllers. It estimates a need for cache on twenty-five percent of its controllers, or approximately twenty-two. DLA is a large agency with many data processing installations and a variety of large Standard Automated Information Systems. What experience it has had with cache has been, at the very least, not unfavorable. Within its systems are many different kinds of data, and we find it hard to believe that none could benefit from cache storage. Granted, cache is useful only in specific situations, and careful study is required to determine them. Under its requirements contract, DLA will have the opportunity to give the problem careful study and install cache only where it is thought advisable. That is all we think can properly be required. Our decision on this issue is bolstered by DLA's effort to allow companies that do not offer cache to compete. Instead of completely ruling out an offeror that cannot furnish cache, DLA will increase its price by the cost of obtaining cache elsewhere. (Quaere whether that makes sense technically if cache is so integral to the disk controller that there is no compatibility between one vendor's controller and another's cache.) If this does not completely eliminate the disadvantage to vendors that do not have cache, it is still better than totally disqualifying them. We note that the protester offered no evidence of the effects on competition of the particular pricing adjustment in this solicitation. We do not grant the protest on this ground. C. The Maintenance Fee Credit. This argument goes to the bare language of the solicitation. On its face the language is defective. A liquidated damages provision can be the subject of a bid protest, and the GAO will sustain the protest if the damages bear no relation to the injury incurred. D.J. Findley, B-215230 (Feb. 14, 1985), 85-1 CPD p 197; Environmental Aseptic Services Administration, 62 Comp.Gen. 219 (1983). In this case, whether the contractor gets all or none of its monthly maintenance fee depends entirely on whether it meets a certain R+ score or not. This sort of all-or-nothing liquidated damages provision meets with almost universal disfavor. In particular, the Armed Services Board of Contract Appeals, which would hear any appeal from a dispute relating to the contract awarded under this solicitation, declines to enforce such clauses if the amount of damages bears no rational relation to the injury. Tor Industries, ASBCA No. 21333, 78-1 BCA p 12,993; Atlantic Services, Inc., ASBCA No. 19876, 75-1 BCA p 11,190. It makes no sense to permit a provision in a solicitation to stand if the board charged with adjudicating cases under the contract is likely to strike it down. DLA argues, citing portions of the record submitted in camera or under protective order, that the effect of these liquidated damages provisions on the protester will be de minimis. That may or may not be the case. What is indisputably the case is that the provisions in this solicitation are patently defective. No matter what company ultimately obtains the contract, it ought not to be saddled with them. . . . . Decision The protest is granted to the extent indicated. The Administrator's delegation of procurement authority to DLA is hereby revised to include the following provisions: 1. The solicitation may not include any provision requiring an offeror's equipment to have attained any particular level of performance as measured by the Reliability Plus program. 2. The solicitation may not provide for liquidated damages except on a sliding scale in proportion to the loss of use of the services of a given data processing system or center. The conformity of the solicitation to the delegation of procurement authority is the responsibility of the procuring agency, DLA. In aid of this decision, the delegation of procurement authority is suspended until the present solicitation has been either (1) revoked and replaced by one that conforms to the delegation as revised or (2) amended to conform to the delegation as revised. This decision is addressed only to the merits of the protest and does not purport to determine the protester's entitlement to either costs of the protest or proposal preparation costs. ROBERT E. LIEBLICH We concur: CYRUS E. PHILLIPS, IV
LEONARD J. SUCHANEK |
|
|