The Network Maturity Model for Internet Development by pgu13428


									Internet Watch

                                                                                                                 Currently, the individual networks that

                                The Network                                                                   make up the Internet use an ad hoc devel-
                                                                                                              opment approach that varies depending
                                                                                                              on the number of sub-Internet entities.
                                                                                                              Although some individual networks were

                               Maturity Model                                                                 built with a certain level of quality, the
                                                                                                              network of networks that comprise the
                                                                                                              Internet has evolved without an overall
                                                                                                              QoN concept. Even though an individ-

                                  for Internet                                                                ual network has a high level of operation-
                                                                                                              al quality, if it was built in an unstruc-
                                                                                                              tured fashion, it can have difficulty
                                                                                                              cooperating with other networks. For

                                 Development                                                                  example, this incompatibility in system
                                                                                                              quality poses problems in trying to main-
                                                                                                              tain even minimum end-to-end security
                                                                                                              requirements effectively and efficiently
                                              Network Maturity Model Development Team                         across a local or wide area network.
                                                                                                                 The Network Maturity Model reverses
                                                                                                              the ad hoc network development trend.
                         he Internet has attracted more                                                       NMM is a process for defining and imple-

                 T       than 50 million users during the
                         past four years, and there is
                         every indication that its growth
                         will continue at a similar rapid
                 pace. As use of both enterprise networks
                 and the Internet continues to grow,
                 addressing the need for quality is an
                                                                                                              menting business decisions about allo-
                                                                                                              cating development resources, engi-
                                                                                                              neering decisions about network layout
                                                                                                              and performance, and management deci-
                                                                                                              sions about network operations. NMM
                                                                                                              introduces sound business and engineer-
                                                                                                              ing practices from the disciplines of sys-
                 increasingly important issue. The 13                                                         tems engineering and project and quality
                 April 1998 AT&T frame relay network                                                          management to the design and operation
                 crash is a reminder of the dire conse-                                                       of complex networks.
                 quences of a network failure. Business
                                                                   Networks can use NMM
                 decisions that impose financial penalties              to target areas                        NETWORK ENGINEERING RESEARCH
                 for network failures further emphasize               for improvement.                           Errors in development procedures are
                 the need for quality in network engi-                                                        causing network outages more frequently
                 neering and operations.                        dictable, repeatable, and timely fashion.     now than in the past, and these outages
                                                                   At the heart of any IT system—and          have a greater impact because more peo-
                 QUALITY OF NETWORK                             thus at the heart of the issue of quality—    ple are using the Internet. These errors
                     Efforts are under way in Internet 2 to     is the network infrastructure that com-       occur because we don’t have a well-
                 offer runtime quality of service (QoS) for     prises the IT system. Whether it is several   defined network engineering process. In
                 user-level performance features such as        computers in a room hooked up to a            their 1998 annual report, the Network
                 bandwidth, delay, and jitter, but little       common server, a company-wide intra-          Reliability and Steering Committee indi-
                 attention is being paid to system-level net-   net, or the Internet itself, QoN depends      cated that 32 percent of network failures
                 work quality. Quality of network (QoN)         on the hardware, software, procedures,        and outages were due to procedural
                 is the ability of an information technol-      and human components comprising the           errors, and that trend is increasing at a
                 ogy (IT) system to meet user expectations      entire network system.                        rate of 3 percent a year.
                 by providing the intended functionality           Inattention to QoN could cause net-           The immense practical utility and
                 or desired information in a reliable, pre-     work failures to increase at a rate com-      impact of research and development on
                                                                parable to network growth. To ensure          software and systems engineering, pro-
                   Editor: Ron Vetter, University of North      QoN, we need to develop standards that        ject management, and total quality man-
                   Carolina at Wilmington, Department of        use network characteristics such as scal-     agement are the motivation for research
                   Computer Science, 601 South College Rd.,     ability, interoperability, security, and      in network engineering processes. The
                   Wilmington, NC 28403; voice (910) 962-       portability as design parameters. A           need to create a better design, maintain a
                   3667, fax (910) 962-7107; vetterr@           quality design process provides network       high level of quality, and provide a path
                                            stability, modularity, connectivity, avail-   for growth is at least as essential for the
                                                                ability, and ease of maintenance.             Internet as it is in the software arena.

   2                          Computer
  Both software and networks are engi-       mine the state-of-practice in tasks such           aving higher quality networks will
neering products that

  • evolve from input design require-
                                             as configuration management, client
                                             management, and fault analysis and cor-
                                             rection. Assessors would evaluate against
                                                                                          H     have a positive impact on user sat-
                                                                                                isfaction by improving overall Inter-
                                                                                          net quality and reliability. The Network
    ments, which frequently change           “base practices” that describe what          Maturity Model will ensure that proces-
    based on stakeholders’ needs;            needs to be done to implement a specific      ses supporting the development of qual-
  • require careful, cost-effective main-    quality process. For example, organiza-      ity networks are defined, established, and
    tenance;                                 tions need to define technology road         continuously improved. Our Web site
  • compound in size and dimensions,         maps that describe how their network         ( provides further
    with severe cost implications; and       will evolve over time and how it will        information about NMM research. ❖
  • have a high impact because they are      incorporate technological advance-
    in widespread use.                       ments.                                       The Network Maturity Model Develop-
                                                Written and oral reports would pro-       ment Team includes S. Bhattacharya,
Current status                               vide feedback for the assessed organiza-     J.M. Capone, K. Dooley, S. Palangala,
   The state-of-the-art in network engi-     tion to use in generating specific im-       and H-S. Yang, Arizona State University,
neering is similar to the hackers’ para-     provement projects and to serve as in-       and W. Baumann and J. Fritsch, Motor-
digm of software development. Com-           put into the organization’s annual strate-   ola Inc. Contact the authors at {sourav,
puter networks were first developed and       gic plan. The overall system would be        jcapone, kevin.dooley, srihari}@,
maintained by local gurus who under-         characterized according to one of five or {William_Bau-
stood the magic of token rings. Network      possible (sequential) NMM levels: basic      mann- P12002, Jim_Fritsch-AJF002}@
operations success came through the          practices, managed processes, defined
Herculean efforts of individuals. Despite    processes, measured processes, and con-
an exponential increase in the complex-      tinuously improving processes.
ity of network requirements and demands
(document sharing, enterprise resource       WHERE ARE WE HEADED?
planning, electronic commerce), this hero       We have structured the NMM into            National Science Foundation
mentality still reigns. Designs are often    three high-level components: enterprise       Exploratory Workshop
determined on the back of an envelope,       management, network engineering, and             The National Science Foundation is
and the main focus of network operations     network operations. Enterprise manage-        sponsoring an exploratory workshop
is disaster management and recovery.         ment encompasses managerial leadership,       to investigate quality and process re-
Users’ concerns about quality are still      stakeholder focus, human resources devel-     search issues in the engineering and
more centered around basic issues of net-    opment, technological leadership, and         operations of complex information
work availability and speed rather than      supplier management. Network engi-            network systems. The workshop will
the actual content they are seeking.         neering includes requirements analysis,       identify possible research topics such
   Because of the strong similarities        design, implementation, verification, and      as underlying network technology, net-
between software engineering and net-        validation. Network operations issues         work system engineering, and quality
work engineering, we can apply the           include maintenance, client management,       in network operations.
lessons learned from the software process    performance management, configuration             Open participation in the workshop
model to develop a similar design process    control, scalability, and security. These     is encouraged. Participants represent-
for quality-oriented research in the net-    areas are further defined by features and      ing the following interests are invited:
work arena. NMM is a feedback-based,         practices.                                    clients of enterprise networks and net-
iterative process designed to assess            We are currently validating the com-       work services, producers of networks
the enterprise management system, the        pleteness and accuracy of our identified       or network elements, network service
network development process, and the         practices, using both benchmarking and        providers, network researchers and
network operations system against well-      expert panels to derive a final, usable set    potential government funding agencies.
accepted and proven principles from sys-     of criteria. The benchmarks target a             Tentative dates for the workshop are
tems engineering, project management,        broad spectrum of businesses that             25-26 February 2000. The workshop
and quality control. In doing so, an orga-                                                 is hosted and cosponsored by the Tele-
nization can evaluate its network engi-        • perform network engineering devel-        communications Research Center of
neering practices and determine its              opment,                                   Arizona State University, Tempe, Ariz.
process strengths and weaknesses as well       • operate data information networks,        Limited workshop support is available.
as target areas for improvement.               • provide network services and busi-        Contact Drs. Sourav Bhattacharya,
                                                 ness modeling, or                         Jeffrey Capone, or Kevin Dooley at
Assessment process                             • are heavily dependent on or signifi-       {sourav, jcapone, kevin.dooley}@asu.
 In a typical assessment, NMM experts            cantly impacted by information net-       edu for further information.
would perform onsite interviews to deter-        works.

                                                                                                             October 1999               3
    Internet Watch

4             Computer

To top