ftp.umcs.maine.edu/2015-02-07.ftp.umcs.maine.edu.tar/ftp.umcs.maine.edu/pub/WISR/wisr4/proceedings/detex/moore.detex

Text File | 1992-04-05 | 6KB | 149 lines

 [12pt] article









  A National Infrastructure for Defense Reuse

   James W. Moore
 IBM Federal Sector Division
 Gaithersburg, MD 20879
 [phone]
 [e-mail]





  Position

 Interest in software reuse is not confined to the Department of
 Defense; commercial industry is vigorously pursuing such
 techniques.  There are, however, some problems specific to
 defense reuse.  For example, there is a difference in scale; few
 commercial programs require the combination of size, reliability
 and responsiveness that characterizes many defense programs.
 Perhaps most importantly, though, there are comparatively few
 opportunities for defense reuse.  Few similar defense systems are
 started within any short period of time and the development of
 those systems may be awarded to different contractors.  This
 means that effective defense reuse must be done across company
 boundaries rather than within a particular company.  Coping with
 this characteristic requires the development of a national,
 inter-company infrastructure for the preservation and
 presentation of reusable software assets.

 The Army's RAPID (Reusable Ada Products for Information Systems
 Development) reuse library  ,
and similar programs have
 performed pioneering work in setting up operational libraries to
 provide reusable assets to local communities of users who share
 requirements and values.  A ``qualification'' or ``certificatio''n
 process
applies these values to the screening of
 candidate assets.  The effectiveness of value-based qualification
 is diminished, though, when applied outside of the community of
 users who share those values.  It seems unlikely that such an
 approach can be scaled up to the national level.

 The USAF CARDS (Central Archive for Reusable Defense Software)
 program, chartered by congress to lay out a blueprint for DoD
 software reuse, will take a different approach.  This program is
 initially focusing on domain specific reuse processes applicable
 to command and control software.  Knowledge from a fairly
 rigorous domain analysis of command and control centers will be
 tangibly expressed in a knowledge-base  , resulting in a
 generic architecture.  Reusable components will be qualified on
 the basis of their ``form, fit and function'' within the generic
 architecture.  This, in effect, permits the government to
 ``standardize'' upon the domain analysis and to create multiple
 instances of command centers derived from the generic
 architecture, thus achieving high levels of reuse.

 Even rigorously structured domain-specific libraries like CARDS
 will not lead to a single national reuse resource.  With today's
 reuse technology, there are simply too many unsolved problems to
 select a single design point for solution.  Even after the
 problems are solved, there will still be a need for diversity
 because different communities will specialize in different
 application domains, will want different fee structures, and will
 prefer different user interfaces and communications mechanisms.
 Most importantly, though, institutionalization of software reuse
 requires a cultural change where thousands of individuals will
 have to make individual decisions to prefer reuse to new
 development.  The probability of a single design point appealing
 to thousands of users is vanishingly small.  Moreover, a
 diversity of choices will provide an element of intellectual (and
 perhaps commercial) competition which will ultimately have the
 effect of reducing costs, improving services, and accelerating
 technology infusion.

 In order to present the resources of many different reuse
 libraries to many different users, a national marketplace is
 required.  The DARPA ASSET (Asset Source for Software Engineering
 Technology) program has the mission of providing this
 marketplace.  ASSET will perform the functions necessary to
 create an interoperating network of reuse libraries so that any
 user of any library will be able to access the services of all of
 the libraries.  ASSET would also serve any central coordination
 functions such as the administration of schemes for nomenclature
 and identification.  ASSET is following a strategy of encouraging
 the participation of private entrepreneurs to provide value-added
 services in all aspects of its operation.  By attracting a
 variety of participants, it is hoped that the electronic
 marketplace will become populated by a diverse set of producers
 providing goods and services for a diverse set of consumers.

 Of course, interoperability of reuse libraries requires some
 level of standardization.  A cooperative government/industry
 consensus group, the Reuse Library Interoperability Group (RIG),
 has been formed to draft and propose such standards.   The group
 hopes to identify and standardize well-understood technologies to
 provide a stable base for value-added activities by the various
 reuse library efforts.  In its six months of existence, the RIG
 has attracted twenty-three organizations as members, including
 major corporations, government agencies, and reuse library
 programs.

 In looking to the future, we can see a ``reuse industry'' serving
 local community (or intra-company) reuse libraries, domain-
 specific libraries which cross corporate boundaries, and a
 national infrastructure for sharing among libraries.  The
 national marketplace will be populated by providers of
 components, value-added tools, consulting, and system
 integration.  All of the services will be delivered to end-users
 via their local library.


   nied91

Major Alvin Nieder, ``RAPID: Implementing a
           Comprehensive Reuse Program,''    Joint DoD Software
           Technology Symposium , Salt Lake City, UT, Apr 91.

Joanne C. Piper and Wanda L. Barner, ``The RAPID Center
           Reusable Software Components (RSCs) Certification
           Process,'' U.S. Army Information Systems Software
           Development Center - Washingtoni, 1991.

James J. Solderitsch, Kurt C. Wallnau and John
           Thalhamer, ``Constructing Domain-Specific Ada Reuse
           Libraries,''    7th Annual National Conference on Ada
           Technology , 1989.