169
Conference Reports
The
IEEE
Computer Standards Conference
March 21-23, 1988 The recent IEEE conference frequently produced excitement that contradicted the " d r y " reputation often attributed to the subject of standardization. From the opening remarks of the keynote speaker to the closing challenges by one of the leaders in information technology, the audience's traditional biases and adherence to past standardization processes were examined and often heatedly debated by the audience. This ought to have been expected. In a few short years the IEEE has produced enough computer standards to match the sum output of all other similar organizations - many of whom have been at work for decades. And this energy has produced m o m e n t u m which continues to produce standards I at a firehose rate and places the IEEE Computer Society at the forefront of professional organizations actually having an immediate impact on our industry. Beyond any doubt, this phenomenal success is due to the vision and management skills of its leaders. Howard Yudkin, president and chief executive officer of the Software Productivity Consortium, keynoted the conference with two important points about present trends deserving international attention. First, the growing importance of software engineering will increasingly require good standards. He outlined his concept of software modules which have standard interfaces that once developed could be reused again and again. He compared them to the present catalogs of electronic parts available to engineers who, as a result, are not being required to reinvent every part for
1 See "A Status Report of the IEEE Computer Society Standards," edited by H. Wood, Computer Standards & Interfaces, Vol. 7, p. 281. North-Holland Computer Standards & Interfaces 8 (1988/89) 169-172
every large, complex system. This reuse of standard parts underscores the impossibility of building 747s if their construction required all new components rather than assembling the plane from many existing catalogs. Catalogs of standard software modules must proceed the necessary progress in the automation of software engineering and the development of software tools. Engineers can then concentrate on the new ideas rather than expending limited resources to construct old ideas in a new form simply to permit their attachment to new systems. Second, Yudkin noted with concern the apparent reluctance of government to support standards even in the face of government's dependence on standards for procurement of technology critical to its operations. He urged a reversal in this trend and cited as a specific example in the US of cut-backs in the National Bureau of Standards budget for computer standards. Responding to questions from the audience, Yudkin commented that consortia, reflecting a new concept and force in US economics, have not yet demonstrated whether they will succeed or not. We still wait for the answer to whether they can solve the problems for which they were created. He noted that the protection afforded from accusations of anti-trust violations or collusion is frequently overstated and that his organization must still tread its way lightly through a mine-field of potential legal issues. Further, of the 60 + consortia identify by the US Department of Commerce, less than 12 are true technology developers. Additionally, while consortia may have introduced more industry input to ISO, the politics in lso has remained the same. In a response that may have startled many of the audience who labor over standard specifications in formal standards committees, Yudkin noted that certification of the standard software module interface was critical to the idea's success but he saw no advantage or need for the specifica-
0920-5489/88/$3.50 © 1988, Elsevier Science Publishers B.V. (North-Holland)
Conference Reports
170
tion to be produced by a formal committee. Indeed, he noted that they frequently produce political specifications rather than useful ones. He left open the possibilities of alternative specification development processes. A panel on Interface Standards quickly initiated the active give and take character of the conference. After establishing that the audience was 30% vendors, 30% users, and 40% who belied the apparent dichotomy of the question, the speakers initiated a heated debate among themselves and the audience on whether: - s o f t w a r e tools have been developed ad nauseam, the Standards Graphics Mark-up Languages defies understanding and clearly should not continue to exist, requires firm management and is not getting it, - the need for standards exists because in today's complex, global world no one manufacturer can provide all needed tools, - standards may be developed to protect existing positions rather than to help users, standards specifications, in an attempt to have only one but one which includes everything, tend to overload standards until the standard collapses under its own weight. In a panel on Models, Management, a speaker from IBM provided a dramatic name for the impact standards frequently have on vested or inplace interests. She referred to the need for internal negotiations on standards issues to take into account the personal needs of the individuals whose projects or products may be affected or even killed by the emerging standards. These needs were called, "ricebowls" inside IBM. This is a term I recognized from the movie and book, " T h e Sand Pebbles" where it named a informal but strangely codified concept. The concept reflected a group recognition and mutual support of the notion that the first concern of the group was not to jeopardize the livelihood, or ricebowl, of any of its members. In the book this leaded to conflict between the group and an individual moved to seek excellence. Are there similar conflicts inside all vendor organizations? Panelists considering the federal governments' role in standardization noted the leadership role in standards thrust on government by its obligation -
-
-
s
t
a
n
d
a
r
d
i
z
a
t
i
o
n
to taxpayer and public for efficiency and safety. PRC, an independent software company, attested to the importance of standards in dollars and cents. Standards which permitted them to tackle and stay ahead of the momentous task of automating the US Patent Office's data collection. Concrete savings were described which resulted from a system design based on publicly available standards and which permitted sub-contracting and overall management from a central point of accountability. The panelists agreed that the federal government's insistence on standards enforced through procurement practices has opened government contracts to greater competition with resultant improvements in cost and quahty but frequently this requirement forced dual product lines on vendors. Dave Burk, the first of two DEC speakers to note emphatically the great importance DEC places on open standards, described a future market for distributed computing amounting to 150 billion dollars that required peer to peer networking. Users everywhere recognize the role of standards to: Stimulate product demand and competition, - Promote open markets, - Increase interoperability, - Reduce risk of vendor and user investments, Ensure high portability, Foster economies of construction, installation, and use. The panelists noted a clear trend towards an increased demand for standards based on business requirements. Standards provide a base line for comparisons in the many requirements for measuring quality, performance, and functionality. There is also growing effort being given towards finding " t h e strategic interface level" where considerable attention and importance is given to standards. In such strategic areas, the federal government is prepared to make its own standards rather than wait for others. Examples cited were posix, a portable operating systems based on unix, and GOSIP, the government's osx standard. Certification of software has growing importance but speakers noted the difficulties in specification writing and interpreting to avoid errors in the specification itself. The session on Managing Standards reminded the audience of the magnitude and scope of typical standards specifications and their documenta-
-
Conference Reports
tion. The speakers 2 addressed specific issues as possible solutions. One suggested using automation to handle the documentation by providing many tools for dealing with the size of current documents. The importance of a conceptual overview to provide general guidance a n d direction to the distributed efforts normally associated with standards development suggested the value of reference models to John Holmes and his associates in the UK's IEE. He reported on suggestions for understanding and augmenting a taxonomy of reference models which were to assist standards development and which were specifically aimed at solving the problem of interoperability among heterogeneous systems through standards. More immediately, this work would provide coordination in today's standardization efforts. Readers should be aware of the current strategic planning efforts now being initiated in ISO which this work could assist. Carl Cargill, of DEC, identified the increased importance of standards to his company and the market in general. Implicit in his ideas is the notion that engineers and technicians make up standards committees and tend to seek a technical solution when a marketing solution is what's really wanted. He challenged the current process that produces standards and suggests an alternative that incorporates marketing and business considerations. One idea this writer finds satisfying is a formal recognition during the standardization process that it may be better in the long run N O T to create a standard but to permit a proprietary solution that produces profits for future research. Unfortunately, the implementation of such a consideration may be more difficult and time consuming than standardization. Should we endorse SNA? Nevertheless, the central idea of seeking market utility from a standard is strikingly correct and novel. To do so brings issues about how a standard affects existing competitive positions out of their current secrecy and emphasizes the user's control through the procurement process. Certainly all of standardization will benefit from the earliest statement of what the buyers will buy coming from those who will be doing the buying. 2 Papers on this session are in the proceedings which can be obtained from The IEEE Computer Society, order number 791, 445 Hoes Lane, PO Box 1331, Piscataway, NJ 088551331.
171
The US federal government's various standardization efforts reflect a statement of buying power. A panel addressing Integrated Data Systems attempted to identify the integrating influence of the standardization process, but, for this listener, succeeded only in revealing the complexity of standardization and divergence of current interests. Richard Foote, of GTE, noted that standards committees frequently permit one or two editors to update the standard so as to reflect the consensus of the last meeting. Lately this process has come under scrutiny because of challenges by meeting participants about the accuracy and fidelity of the editing to the intent. Consequently, additional rules and safeguards are felt to be needed. The changes will define the scope and authority of the editor, provide limits to permissible change, require recordation of the editing meetings decisions, and suggest national body support for the editor. The process suggests guards guarding the guards. One must ask if the intent is to make the standardization committee more, or less, responsible for its work. Pat Harris, from the National Information Standards Organization, identified duplication and potential duplication between the computer standards normally discussed by the 1EEE and the bibliographic or library standards which NISO addresses. In her comments during the panel, Impact on Market Place, Helen Wood, NBS, quantified the information technology market which forms the core of the conference's several standardization objectives. The numbers are indeed impressive. She estimated $354,000,000,000 of product in use. Annual expenditures of the US federal government will be $16,000,000,000 and private industry's will be $150,000,000,000. This compares well with other estimates 3 and underscores the importance of standards IF standards have impact on spending. Mike Kaminski, GM, who has been shepherding the MAP effort, pointed to clear savings already produced by MAP and noted that MAP contributes benefits to all its participants - - not just GM. Don Abelson, a US trade representative from the President's Office, reported that the globalization of information technology has led to the need for
3 H. Poppel and B. Goldstein, Information Technology: The Trillion-Dollar Opportunity, McGraw-Hill, 1987.
172
Conference Reports
the General Agreement on Trade and Tariffs (GATT) to address specifically the possibility of standards used as trade barriers. The closing speaker, Kenneth Olisa of WANG, noted a cultural reversal in the way vendors treat standards. Once a vendor accepted multi-vendor sites as a given, the necessity for standards becomes obvious. Quoting the slogan, '"Reads', 'Feeds', and 'Speeds' are dead; customer needs are king," Mr. Olisa commented that the old emphasis on technical metrics has been replaced by the need to provide customers with solutions to their needs. Noting that 22,000,000 raM-type Pc's will be in place in 1988, he nevertheless disagreed with those who predict the disappearance of mainframes. He listed some specific needs of current businesses: - Systems for today and tomorrow, - Complete solutions, Multi-vendor environments, Realization of past investments, Sensitivity to the user, Long term relationship. The vendor challenge, he felt, was to break the current proprietary molds, integrate the key technologies, listen carefully to needs, be timely, and make a profit. Similarly, he listed the customer challenge as articulating their needs well, ensure realizing past investments, moving their discipline
in the direction of the Chief Information Officer, and meeting their organization's goals. In summary, the three day conference addressed, reviewed, and updated many of the past standardization concerns and offered a few new ideas. While we could cover and report on only a few, the proceedings are available for our readers who seek more information. The attending audience's interests were quite varied leaving the impression that while Iso's osI standards dominate current standards work, many more prosaic efforts are both required and underway. A major new trend is the increasing vendor support of standards with the intention to open customer bases to multivendor systems. However, the reader should certainly take to heart a recurring warning: the success of osI and the various consortia promoting osI is far from assured. The work may only now be reaching a mature stage where work and experience accumulated in earlier standardization work can provide guidance. Success will require continued support by both vendors and users. And, more important, we may be beginning to recognize that changes are needed in the standardization process to take into account the brisk pace of computer and communication technology change. John
L.
Berg
Editor-in-Chief