Click for next page ( 26


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 25
IV. INTEROPERABILITY It was noted earlier that researchers in one discipline sometimes have been unable to use their data network to access the data base of another discipline. The Committee believes that OSSA recognizes this problem and wants to ensure that future systems are compatible, at J east to the extent necessary to achieve the required degree of interoperability. The present-day interoperability problems will pale in significance when compared to those of the future, especially with respect to the earth system sciences. According to OSSA's Earth Observing System (EOS) Data Panel, EOS will produce several orders of magnitude more data per day than any previous mission.* That group also reported the following: "On the scientific level, we anticipate system resource con- f~icts arising that will demand resolution within the confines of the data and information system. Multidisciplinary researchers and research teams wild have needs for particular observational sequences, while disciplinary researchers may weld have require- ments for entirely different measurements. Both groups wild be affected by spectacular events and the pressures (both scientific and political) to respond."** An Earth System Sciences Conceptual Model. One can see the complexity of the information flows within the Earth system sciences environment by reviewing the conceptual mode] prepared by NASA's Earth System Sciences Committee, shown in Figure 4.*** That committee notes that changes to our planet during the time span of human history have been modest when compared to those that have occurred over geological timescaJes. However * See page 8, citation from Report of the EOS Data Panel. ** Report of the EOS Data Panel, NASA Technical Memorandum 87777, Volume Ila, 1986, p. 26. *** Earth System Science Overview: A Program for Global Change, Earth System Sciences Committee (Francis Bretherton, et al.), NASA, May 1986, pp. 24-25. 25 r,

OCR for page 25
human activity over the next century or two may contribute to global changes comparable to those of geological history. The major components, shown as boxes in Figure 4, should be conceived of as groups of computer subroutines incorporating detailed knowledge of the relevant processes provided by the traditional Earth-science disciplines. The pathways (arrows) that connect these subsystems represent the information flow necessary to describe the interactions among them. The ovals and the attached arrows denote inputs from, or outputs to, an external environment. The use of models will help scientists to understand better the multiplicity of interactions among the Earth System components and to aid in predicting the future evolution of the system in response to selected changes in input variables. The mode] also illustrates the emphasis being pi aced on an integrated view of the interactions of the lithosphere; the physical climate system, including the atmosphere, oceans, and J and surfaces; and the biosphere, coupled to the other components through the biogeochemica] cycles. This integrated view is a fundamental aspect of Earth System Science and tends to indicate a continuing and increasing need for cross-discipline and multi-discipline data flows. In both solar terrestrial physics and planetary science, the science requirements for missions planned for the 199Os will be different from those in the past two decades. By 1989 spacecraft will have visited two comets and all of the planets of the solar system except Pluto. The first missions were primarily missions of discovery while the missions in the 199Os wild concentrate on increasing understanding. Missions are planned to Venus, Mars, Jupiter, a comet and an asteroid. The scientific problems in planetary science wild require the talents of experts in a variety of disciplines. For instance, one of the most intriguing features of the Jovian system is lo and its interactions with corotating Jovian plasma, which is of loan origin. An understanding of the phenomena associated with these interactions requires the expertise of planetary geologists, atmospheric scientists, and plasma physicists. In solar terrestrial physics, the emphasis will be on understanding the flow of energy and momentum through the coupled systems of the solar wind, the magnetosphere and the ionosphere. This will require coordinated multi-spacecraft, mu~ti-instrument and ground observations. In astronomy and astrophysics the 199Os will begin an era of observatories in space. Late in this decade the Rubble Space Telecope wild be launched. The Hubble Telescope plus the Space Telescope Science Institute wild constitute an observatory like those on earth, only with the telescope above the atmosphere. These changes and others indicate that OSSA faces a significant chal- Jenge in the development of interoperable information systems to support forthcoming interdisciplinary and multidisciplinary missions. There are no simple solutions; the issue is: Issue #2. How can interface requirements be established that would ensure interoperability with a minimum of standards? 26

OCR for page 25
i;; | Climate Change | I .o ~ ~ l .. ~ ~ co I On _ 1 UJ O 1 o ~ 1 c ~ l . ~ 8 ~ _ , l 1 l r l 1 1 Rae ~ _ Maximum | | I 141 1~(So ,NO )| 2 ~ L ~ ._ I _ ~ ~ ~ _ !, , ~ . ~ _ ~ _ e ~ . e | Dynamics | Chemistry l ~Ic~ z =] Stratosphere/Mesosphere | ~ ~ I . _ _+ 1 ~ 1 . ~ I_ ID ll l _ 1 1 1 3= l ! _ s ,o C) Q O C) _ = At) ~1 C as a, 3 LO | (H2O)~(S N. )| | UV. Particles | 27 ,~\ - Al _

OCR for page 25
Interconnection and Interoperability of Information Systems. Certain- ~y the easiest way to achieve interoperability is to have one system with identical or compatible hardware and software throughout. Practically, though, there are many reasons for keeping information systems separate, such as the need to keep old but relatively cheap systems in operation rather than lose the data or have to move it to another system. There- fore, there will always be a number of systems supporting OSSA's missions, and it appears there wild be an increasing requirement for interoperabili- these systems. OSSA already knows that ease in iJitv will depend directly on the degree of homogen- ty among a number of achieving interoperab ~ eity of its systems: the more alike they are the easier it is to intercon- nect them. Unfortunately, OSSA information systems differ substantially in their data base formats and languages, their operating systems, and the composition of their network protocols. OSSA, in concert with the Office of Space Tracking and Data Systems (OSTDS), GSFC, and JPL, has been addressing several aspects of the interoperability problem. Examples: On-Line CataJoq Activities. The trend toward decentralized and - distributed data bases is causing increasing difficulty for users who need information about the data bases or data from them. The NSSDC On-Line Data Catalog System (NODCS), consisting of the Central On-Line Data Directory (CODD) and the Distributed Data Catalog System (DDCS), is being developed to alleviate these prob- Jems. Once a user has decided to access a particular catalog in the DDCS, it will in some cases be possible to gain access auto- matical~y through CODD. In other cases, it will be necessary to follow procedures that CODD wild provide. However, there is no mechanism planned to allow a user in any catalog to gain direct access to data in any other catalog within the DOGS, since the interface languages and data query systems differ and the systems are structured to be top-down and menu driven. The Committee felt this was a useful first step, but that further efforts are needed, perhaps through the provision of gateways, to ensure interoperabil- ity across the ocean, land, climate, and planetary data systems and perhaps among other science data systems and odder meteorolog- ical and earth-observation systems. Data Interchange. Both OSSA and OSTDS have strong interests in (and have been heavily involved in) the development of Standard Formatted Data Units (SFDU) as a means of facilitating both data transfer and interoperabiJity among data systems. Most data formats now are mission-specific, but NASA wants a transition to generic structures and processes, since the goals are to maintain the data sets and relevant information in a consistent form and to enable users to obtain data with no manual intervention. OSIDS and GSFC have represented NASA on the Consultative Committee on Space Data Systems (CCSDS), a standards body that concentrates on standards for interface protocols including the SFDU. JPL also has done quite a Jot of work in this area, including the develop- ment of standards that are consistent with the CCSDS recommenda- tions regarding the SFDU. OPE's Planetary Data System will serve 28 1.

OCR for page 25
as the SFDU test bed for the planetary community. JPL also plans to require the use of the SFDU in other missions. Since the less two interconnected systems are alike the more they will need conversions across the various layers of the information systems architecture, OSSA must concentrate on eliminating differences among its systems. It will not be possible for OSSA to change its older systems, so the problem is not going to go away for many years. However, the planning must begin now. The Department of Defense (DoD) Experience with Transport Protocols. It is useful to consider the experience of the DoD, which encountered a similar problem in the 1970s, a period that saw the evolution of a number of new networks, using many different protocols. The DoD's Advanced Research Project Agency (ARPA) established a network--ARPANET--in the early 1970s to serve as a research vehicle and testbed for computer communications protocols. In 1978 the agency concluded four years of developmental testing on a Transmission Control Protocol (TCP) and an subsequently were mandated for u~ ac `, a Internet Protocol (IP), which . ~ standards throughout the DoD. It has been concluded that the momentum for the DoD program to establish interoperability among its networks resulted primarily from the directed use of TCP/IP. In the early 19SOs, the National Bureau of Standards (NBS), in coop- eration with the DoD, industry, and the International Standards Organiza- tion ("ISO"), developed a new Transport Protocol (TP-4) and a new Inter- network Protocol.* While DoD's TOP and IP have proven to be highly effective, and IP and the "ISO" Internetwork Protocol can easily be made compatible, TOP is not compatible with TP-4. TP-4 and the Internetwork Protocol were approved by the "ISO" as Draft International Standards in 1983 and 1984, respectively. Since commercial vendors normally consider Draft International Standards to be ready for implementation, there has been some expectation that commercial equipment manufacturers will employ the "ISO" standard protocols. If this were to prove true, organizations that employ other standards would find it difficult to find commercia1ly- avaiJable, off-the-shelf hardware for their networks. However, industry has not produced TP-4 products as rapidly as had been expected. As a result, DoD has indicated that it will adopt TCP/IP and TP-4/"ISO" IP as coequal standards after a satisfactory demonstration of the Jatter's suitability for use in military networks. A final commitment will be deferred until the demonstration has been evaluated and TP-4 products are commercially available.** In this report, "ISO" represents the International Standards Organiza- tion and ISO represents the OSSA's Information Systems Office. ** See Transport Protocols for Department of Defense Data Networks, a . report of the Committee on Computer-Computer Communications Protocols, Board on Telecommunications and Computer Applications, NRC, National fContinued on p. 30] 29

OCR for page 25
This approach will provide maximum flexibility for the DoD, by enab- ling it to determine whether to convert to TP-4 or to remain with TCP, depending on which standard proves to be the popular choice among equip- ment manufacturers. In either case, the DoD will obtain the benefits of standard commercial products at an early date. Once commercial products are available, development, procurement' and support costs should be low- er. The main points illustrated by the DoD experience are: (~) adherence to some standard is necessary to achieve (inter alia) interoperability, and (2) the decision to adopt or not adopt protocols that have been promoted by national and international voluntary standards organizations can be driven as much by factors such as product availability as by the goal of interoperability. Onen Systems Interconnection (OSI) Architecture. It should be noted ~ , that TP-4 is part of a broader scheme called the OST Architecture. The "ISO" began developing this architecture in the late 1970s, at the same time the International Telegraph and Telephone Consultative Committee (CCITT) began to develop its own OST reference model. By 1984 the CCITT had adopted the same language as the "ISO", and the reference model now is known as International Standard ISO 7498 and CCITT Recommendation X.200. As an indication of the seriousness with which the OSI architecture being regarded throughout government and industry, OSSA should consider the following recent events, which also tend to indicate that OSSA might wish to orient its thinking toward eventual migration to TP-4. Establishment of the Corporation for Open Systems (COS). In early January 1986, a group of computer and communications manufacturing companies incorporated COS in the Commonwealth of Virginia as a nonstock, not-for-profit membership corporation. The purpose of COS is: "to provide an international vehicle for accelerating the introduction of interoperable, multivendor products and services operating under agreed-to OSI, Itegrated Services Digital Network (ISDN) and related international standards to assure acceptance of an open network architecture in world markets." COS proposes to achieve its objectives through establishment of conformance and interoperability test programs to verify member-companies' product compliance with the "ISO" OSI standards and by identifying areas "Continued from 0.29] Academy Press, Washington, D.C., February 1985. The report notes that the services provided by TOP and TP-4 are functionally quite similary, but that some functions are provided in significantly different ways by the two protocols. This includes data transfer interface, flow control, connection establishment binding, and out-of-band signals. It was estimated that an experienced programmer would require about six months to design, implement, and test modifications of the three major, higher-level, DoD protocols (fide transfer, mail, and Telnet) to work with TP-4. 30

OCR for page 25
in which standards development needs to be accelerated. By early 1987, COS membership comprised 61 computer hardware, communica- tions, and software companies; computer and communications ser- vices providers, computer and communications users, and companies involved in the development of underlying technologies. The total includes three British companies, one Italian company, and agen- cies of the British and Canadian governments.* Establishment of the Government OS] Users Committee. In early . . September 1986, the government announced establishment of the OST Users Committee, whose goal is to determine an OS] standard for the government. The government also is considering a revision of its procurement policies to prohibit the purchase of commercial products that do not conform to the standard, which the committee hopes to develop during 1987. Fifteen agencies, including NASA, belong to the committee.** Considering these recent developments, it would appear that OSSA's future strategy has been set, but that it stir] needs to concern itself with the '.tactica1" problems of dealing with its older networks and, for those networks that require interoperabiJity, the establishment of a clear migration path to the OS] architecture. The following factors (and perhaps others) need to be examined, in order to develop criteria against which OSSA can identify and evaluate suitable options: functional and operational specifications (that is, will the pro- toco] designs meet OSSA's present and future operational needs?; interoperabiJity requirements (for example, must OSSA networks be interoperable at the applications level as well as at the network access level?; minimum procurement, development, and support costs; and ease of transition (migration) to new protocols. As with the issue of centralization, the Committee believes that OSSA is justified in taking a rather cautious approach to the interoperability Information supplied by COS, January 1987. ** The other 14 members of the Government OS] Users Committee are the Departments of Agriculture, Commerce, Defense, Energy, Health and Human Services, Housing and Urban Development, Interior, Justice, Labor, Transportation, and Treasury; the Environmental Protection Agency, the General Services Administration, and the Office of Management and Budget. 31

OCR for page 25
issue. While it is clear that cross-discipline data accessibility wild be increasingly important in earth-oriented discipline programs, it is not so clear that this will be the-case for programs in space physics, solar physics, astronomy and astrophysics, and planetary science. It is impor- tant to determine as soon as feasible which of the disciplines require or can benefit from coordinated or common data systems and which do not require coordination. To force commonality where it is not needed runs the serious risk of increasing costs and hampering the scientific work. Resolving these questions requires working with the discipline program managers and scientists. For those new networks that wild require interoperability, it would appear the proper approach would be to focus attention on non-proprietary standards that are in place and emerging for high-speed data networks. There is no evident need for OSSA to develop new standards or technolo- gies, but it needs to develop a strategy for the adoption of standards so that its future networks wild be compatible with those of the agencies with which it wild need to interoperate, such as the DoD, the National Oceanographic and Admospheric Administration (NOAA), the National Science Foundation (NSF), the Department of Energy (DOE), and other agencies with evolving networks. The Committee understands that OSSA is considering use of the NSFnet for data as weld as electronic mail distribution to researchers. Within the time-frame currently being assumed for the NASA environment, there is a reasonable expectation that the evolving NSFNet will be able to support the data rates implied by the l- and lO-Mbps rates discussed above. The current short-term plans call for networking technology based on T! carrier facilities (~.544 Mbps) and internet protocol (IP) routers--the switches of the NSFNet backbone--capabJe of 1,200 maximum-length packets per second. The limitations of current networking are not in the transmission facilities, which can go to one gigabit per second (Gbps) rates, but in the switches, gateways, and packet handlers. The development of these units to support fiber speeds is happening slowly. However, the foregoing assumes that the information that flows onto the NSFNet is created by networks that support the NSFNet protocol family -- namely DoD's TCP/IP, evolving to the "ISO" TP-4, which has also been recommended by the CCITT. Many of the constituent NASA networks, such as the Space Physics Applications Network (SPAN), do not conform to these standards. This creates a variety of problems. The most critical is the inability of researchers to reach across from, say, the ocean data net- works to the Earth science databases. also, the lack of clearly defined standards in electronic mail, file transfer, etc., which is common across the OSSA discipline networks, makes it difficult, if not impossible, for researchers to communicate with each other. In addition the app~ication- level gateways required would severely restrict the network throughput across such incompatible networks. Keeping in mind the approach being taken by the NSF, the DoD, and other agencies that have elected to stay with TCP/IP but eventually to migrate to TP-4, OSSA might want to consider a similar strategy in future information systems acquisitions. 32

OCR for page 25
Based on presentations on procurement made to the Committee, the information systems procurement process appears to be one that follows the sequence in which designs and design specifications are developed internally to a fairly detailed, well-coordinated level and then issued for competitive procurement. The process described by the presenters is quite lengthy, consuming from one to three years before a definitized, coordinated specification is generated. During the process, some dialogue takes place with potential contractors, but for competitive purposes many details are closely held internally within NASA. Contractors usually are given the opportunity to review and comment on the specification and statement of work. However, once the formal Request For Proposal (REP) has been issued, contractors must respond to a "design-to-specification" requirement in a relatively short time. This approach often results in systems with technologies that are not state-of-the-art, and architectures that are not cost-effective in terms of the system life. NASA presenters indicated that several significant NASA information system procurements have been issued or are planned to be issued using this approach. These include the Technical Management Information System, Program Support Communications Network, and Earth Observing Information System. A Procurement Strategy to Foster InteroperabiJity. The Committee suggests that an alternative information systems procurement strategy be considered--one that has been used successfully by NASA on many large space procurements--involving a competitive Concept Design Phase (COP) based on performance requirements that will lead to design specifications. Some of the characteristics and benefits of this approach are the following: Effort "Multiplier." It is expected that competing contractors wild commit substantial resources during the COP, which fact wild permit the project to move forward further and faster than would be possible with government funding alone. Enhanced Options. OSSA will be provided several design solutions, reflecting both the technological state of the art and the creativity of the competing contractors, enabling it to select system and subsystem design configurations that wild do the job at the best price. This would enable OSSA to request and evaluate proposals that include provisions for a migration path to TP-4. Industry Perspective. OSSA will J earn the industry's perspective on system costs and alternatives. Improved User Support. OSSA will obtain a better product by offering industry the opportunity to participate in its dialogue with the users on the mission to be satisfied. Included in this dialogue would be users who must access the system indirectly. 33

OCR for page 25
Faster System Development. OSSA will have the opportunity to prototype systems or elements of systems rapidly, as a step in shortening the development of design specifications (by intro- ducing user requirements to engineering design, for example, as a step in the specification generation process). The Committee believes that a procurement strategy and process foster- ing competition as early as possible in the acquisition cycle, particular- ~y on large information system procurements, would enhance the probability of satisfying user and mission requirements in a more timely manner' with- in life cycle cost-effectiveness goals. The adoption of common standards among the research networks of major agencies such as NASA allows users to have maximum flexibility with res- pect to access to their data, use of common campus networks and facili- ties, use of alternative access routes that provide flexibility, and use of nationally-supp~ied communications facilities. We believe OSSA is taking the proper approach to this issue. Time is the enemy now, however, and OSSA is faced with critical choices that must be made before the massive flood of data from the new NASA initiatives and the desire of researchers to access such data both increase the frustration of the users and make it difficult to convert from the status quo to state-of-the-art information systems that satisfy NASA's and the users' needs. 34