Click for next page ( 108


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 107
5 Computer Networks INTRODUCTION This chapter describes briefly the technology by which inclepen- dent computers cooperate with each other to exchange data and software, and to share specialized computing and storage equipment using modern data communications. From this technical discussion, some important public policy issues are drawn concerning export control of computer networks. A cooperating network of computers depends on many technolo- gies that cannot be covered in so brief a span as this chapter, but which are necessary to understand the context of computer networks. Fundamental to aB communications is the basic telecommunications network of cables, wires, fibers, internal switches, and other technical devices, but there is little coverage of telecommunications here. Data communications, which is another technology that had to be slighted, focuses on the devices that interface directly with the telecommu- nications network. Typically, this includes the telephone handset, the computer modem, and the Ethernet cable attachment inter- face. This chapter refers to the telecommunications network as the "media," and the data communications hardware and software that interact with it as the "media access" layer of control. Computers interface with the data communications technology, which allows a 107

OCR for page 107
108 GLOBAL TRENDS IN COMPUTER TECHNOLOGY data stream of bits to flow between the computers. It is the technol- ogy of control and interpretation of those data streams by modern computer systems that is the focus of this computer networking technology discussion. In particular, the chapter focuses on the "loose coupling" of inde- pendent computers through telecommunications, "loose" in the sense that the computers in the computer network are general-purpose ma- chines that leave or join the network at arbitrary times. "Tightly coupled" computer networks employ special processors and high- speed bus hardware as the communication medium, with the ensem- ble collectively called a parallel computer. Such nets are addressed in Chapter 2 in the context of high-performance parallel and multipro- cessor computers. However, the distinction is becoming moot with very high speed (above 100 Mbits/s) fiber-optic local area networks (LANs). The key differentiator, then, is the use of general-purpose (Ioose coupling) versus special-purpose (tight coupling) "protocols" in the communication "handshakes" between processors. Discussed here are the basic trends toward packet-switched com- puter networks based on International Organization for Standardiza- tion (ISO) data exchange protocols. Although an uncontested leader in the development of protocols, the U.S. government is caught in the dilemma of embracing ];SO protocols for the future and promulgat- ing standard protocols for the Department of Defense at the present time. Breakthrough possibilities are examined in a field paced by slow standards adoption. Computer networking R&D is paced by the universities and DOD, but the business is paced by national telecommunications agencies and companies. The USSR is far be- hind because of its lack of telecommunications infrastructure (see Chapter 6~. Protectability of U.S. leads in technology is difficult if that tech- nology is standardized, but production technology may provide a competitive edge. Security is a unique problem for computer net- works and international standards. It is a requirement for serious commercial growth on the one hand, but on the other hand it must use export-restricted encryption and security technologies. This sit- uation leads to an export policy deadlock.

OCR for page 107
COMPUTER NETWORKS 109 MAJOR TECHNOLOGY TRENDS A basic trend in computer communications has been the explo- sive growth of packet-switched computer networks over more tradi- tional line- and message-switched communications technologies dur- ing the past decade. This trend appears to be continuing and accel- erating as it feeds on the commoditization of the personal computer and engineering workstation coupled to commercial telecommunica- tion nets through Tow-cost modems. The two most popular types of computer networks are the Tong-haul wide area network (WAN), which services the nation or a geographical region, and the local area network, which services a building or a campus. Like most computer technologies, basic networking depends on many other technical trends discussed elsewhere in this report. Among the more significant are Tow-cost, high-performance com- puter hardware to manage the digital signal processing of modern telephony, and computer software to manage the hardware and ap- plication protocols. Alternatively, other technologies depend on com- puter networks for productivity, providing data access and distribu- tion, andfor sharing hardware and software resources. For example, modern supercomputers need high-speed LANs and WANs for their user access and for the large volumes of data they produce. Fac- tory automation is moving data among administrative, logistic, and production facilities automatically through networks. Office automa- tion employs computer networks to share equipment and data and to move information to people rather than people to information. Computer networks strengthen interdependencies among gov- ernment, business, technologies, products, services, and private and public communities of interest. The "glue" that makes this happen is the standardization of communication protocols. Standard Peer Protocols The physical media of computer networks are quite varied, in- cluding twisted pairs of telephone wire, coaxial cable, fiber-optic cable, and radio. The common ingredient is the use of serial bit streams of standard formatted data groups frames, packets, clata- grams, and messages- parsed by modern computer software logic into control and data portions, that is, into protocols. Control infor- mation is used to identify the sender, the receiver, error correction, and so on, in the routing and delivery of the data by the computer

OCR for page 107
110 GL OBA L TRENDS IN COMP UTER TE CHNOL O G Y network components. The data constitute the sender's private infor- mation. The objective of computer networking is to permit an arbitrary process in one computer to exchange data with another process in an- other computer with a "seamless" interface. The computer network services are organized in a hierarchy of delivery protocols. When the service is provided between two equivalent or peer processes, it is called a "peer protocol," employing common syntax and semantics in the control and data formats. The implementation of a peer protocol is in support services from its computer environment obtained from an "adjacent" Tower-level (layer) protocol service in the protocol hi- erarchy. As a message moves through the network, beginning at the highest levels of application, the adjacent computer network layer encloses the message- both data and control in another framing structure, like putting a small envelope in a larger envelope. Each layer provides different services that require different control infor- mation. The success of the scheme is in its conceptual simplicity and in the flexibility of an all-digital, software-mechanized implementa- tion. The difficulty has been in finding the correct layering of the network services. A Simple Protocol Analog The (liplomatic use of the term "protocol," as a code of etiquette and precedence derived from the Latin roots "to glue together," comes close to its meaning in computer networks. The messages that flow between computers follow an established set of rules (etiquette) in proper sequence to glue the network into a cooperating community. The following human analog tries to capture the concept of a protocol hierarchy or stack. An American chief executive officer (CEO) wishes to complete a business transaction with his or her counterpart (peer) in Japan. The American CEO, representing the "application layer," composes thoughts in a manner that a Japanese peer will understand (peer protocol), and dictates a letter to a secretary (presentation layer). The secretary converts the communication from one format (voice) to another format (written), representing the service the secretary flower layer) provides to the boss (higher layer). The secretary then puts the letter in an envelope and puts the Japanese CEO's address on the envelope, thus making a session on behalf of a higher layer entity. The letter is mailed (passed down to a Tower layer

OCR for page 107
COMPUTER NETWORKS 111 entity) to the U.S. Postal Service, a reliable "datagram" transport mechanism. The post office passes the letter to regional collection centers (switching centers of the network layer) and then on to the destination post office via the routing information in the letter's address, usually the ZIP code (control information). The passing is handled by bundling many different letters with similar ZIP codes in bags carried by truck, plane, or ship to the destination (the physical media, lowest layer). The process now repeats in reverse post office to secretary to Japanese CEOphysical to network to transport to presentation to application layer. The Japanese secretary sees to it that the English letter is translated into Japanese, the destination's presentation layer. At each layer of the protocol hierarchy, there is a peer protocol that understands the rules (etiquette) of its peers; the letters are formatted the same, the envelopes are addressed the same, the mail- bags are labeled in an agreed-upon format, and so on throughout the process. Furthermore, each layer provides a service to its higher layer, and interface protocols express the service requests. Figure 5.1 shows this hierarchy of layers: the popular Open System Interconnect cost' model of a seven-layer hierarchy. . DOD Versus ISO Proto co] Standards The ARPANET implemented the first successful protocol suite (known as TCP/IP) in the mid-1970s. It has evolved and been adopted by the DOD as its model. The ISO proposed the OSI model of seven layers of service that has been adopted by most countries. A comparison of the two models is made in Figure 5.1. Research over the past decade by the commercial common carri- ers and the DOD has led to the competing protocol suites of Figure 5.1. Efforts are under way to build a common model of seven service layers. At the high layers (5 through 7) the emphasis is on application services for electronic mail, file transfer, name services, and remote terminal services. The middle layers (3 and 4) deal with transporting messages, packets, and frames across the networks, reliably, compat- ibly, and flexibly. The lowest layers (1 and 2) manage the various communications media. As standards have matured, industry has implemented the pro- tocols, particularly the lower-layer protocols, which have the greatest maturity and stability, in hardware chips and PC boards. The United States leads in protocol development, particularly protocol testing

OCR for page 107
112 GL OBAL TRENDS IN COMP UTER TECHNOL O G Y OSI X.400, FTAM, VTP Model (7) Application (6) Presentation Various Subsets and Options (5) Session DOD SMTP, RFC-822 (5/6) Utility FTP, TELNET . TP0-TP4 ~q Owl, ~~ TOP, EGP, GGP X.75, CLNP | (3b) Internet IP, ICMP X.25-3 (3) Network (3a) Network 1822, X.25, ARP X.25-2 (2) Data Link 802.2, LAP X.25-1, X.21 (1) Physical 802.3,4,5, FDDI KEY: X.400 = CCITT-STD, Electronic Mail Interconnection FTAM = File Transfer and Manipulation VTP = Virtual Terminal Protocol SMTP = Simple Mail Transfer Protocol, MIL-STD-1781 RFC-822 = Internet Test Message Format FTP = File Transfer Protocol, MIL-STD-1780 TELNET = DOD VTP, MIL-STD-1782 TP0-4 = Transport Protocols with Options: TP4 max services TOP = Transport Control Protocol, like TP4, MIL-STD-1778 EGP = External Gateway Protocol GGP = Gateway to Gateway Protocol X.75 = X.25-3 to X.25-3 Gateway Management Protocol CLNP = Connectionless Network Protocol, like DOD IP IP = Internet "datagram" Protocol, MIL-STD-1777 ICMP = Internet Control Message Protocol, RFC-792 X.25-3 = CCITT-STD Level 3, Virtual Circuit (Packet) Protocol 1822 = ARPANET Host-lMP Protocol, being replaced byX.25-1,2,3 X.25 = CCITT Standard Protocol if a number of sub Protocols ARP = Ethernet Address Resolution Protocol, RFC-826 X.25-2 = CCITT Level 2 Standard, HDLC Framing Protocol 802.2 = IEEE-STD Logical Link Control Protocol, like HDLC LAP/lAPB = DOD-STD Link Protocol, like HDLC X.25-1 X.21 802.3 802.4 802.5 FDDI CCITT Level 1 Standard, Physical Interface Protocol Physical Level Protocol, like X.2~1 IEEE Ethernet Carrier Sense Multiple Access/Collision Dstection, CSMA/CD Protocol IEEE Broadband Token Bus Protocol IEEE Token Ring Protocol Fiber Distributed Data Interface FIGURE 5.1 Open System Interconnect (OSI) and DOD models with representative protocols.

OCR for page 107
COMPUTER NETWORKS 113 and chip production. This lead should be supported to strengthen U.S. competitiveness as this technology matures. Computer networking is a dual-use technology. It serves as the backbone of modern military command and control systems, of mili- tary base-level communications systems, and of intelligence systems interconnection, as well as the DOD backbone Defense Data Net- work (DDN) replacement of AUTODIN. Computer networks are the leading technology for office automation, second only to the PC. Classified military applications use many DOD and OST general-use protocols TCP/IP, TP4/CENP, file and mail protocols that are not controllable, because they are now standards with commodity status, available from many vendors, including university and foreign sources. Classified systems also use specialized protocols, many of which should be controlled for national security reasons. The easiest method for their control is to subsume any separate export controls for the computer network protocols in the export license currently required for export of a classified system of which they are an integral part. OS! Protocol Profiles Protocol developments in Europe continue to refine the OS] suite of protocols for application communities. This work entails packaging specific protocols in hierarchies based on various options required by the community. The packages are caned "profiles." Europe is ahead of the United States in implementing OST, and its software exports to the United States are increasing. The successful transfer of OST to the United States may weaken U.S. competitiveness by giving OST global validity even though it originated as only a European local standard to address needs and conditions on the Continent. The irony is that the technology was invented by the DOD and has generated more sophisticated implementations. The importance of protocols can be best appreciated by ex- LANs ~ r ~ 1- ~ ~ amining the growth of computer networks in the factory. are growing in the administrative, financial, inventory, and factory contexts and are being integrated into computer-integratec! manufac- turing. The computers, or hosts, on the different computer networks cover a wide range, from business machines in the a(lministrative and financial nets, to automated cutting, weakling, and manipulation

OCR for page 107
114 GL OBA L TRENDS IN COMP UTER TEClINOL O G Y processors and robots. With each need, new peer protocols have develop ed.i American manufacturing competitiveness is at stake. Advanced Research in Protocols Protocol R&D in the United States continues to explore two dif- ferent dimensions: media an ~ performance exploitation, and different functionality. Unlike other countries, the U.S. research community has ready access to computer networks, tools, and laboratories to accelerate advances in protocols. By running faster, the United States maintains leadership in future products and services. Pro- tocol production technology is key to a competitive leacI.2 DARPA is exploring a variety of new protocols that trade functionality for speed or simplicity.3 Computer Network Management Packet nets are complex and difficult to control. Their com- ponents are numerous and some are always broken. Self-balancing regulatory mechanisms are still evolving. The need for dynamic con- figuration control and management of resources is called "network management" and the biggest area for future growth. The problem is manifold. There are conflicting commercial interests; standards have helped mature the technology, but there are few standards in the 1 U.S. examples of such protocol profiles are: GOSIP (Government OSI Protocol), an NIST-specified OSI protocol suite for U.S. government users that uses various level 1 and 2 protocols; CLNP, TP4, OSI "stacks" at levels 3 through 6, X.400, FTAM VTP; MAP (Manufacturing Applications Protocol), a General Motors-specified OSI suite employing 802.4, 802.2, various OSI stacks at levels 3 through 6, and a research Manufacturing Message Protocol (MMP) at level 7; TOP, a Boeing profile equivalent to MAP for office automation employing 802.3 Ethernet as the LAN. 2Many universities and industrial laboratories are exploring the performance limits of TCP/IP overhead. Given ideal conditions, bandwidths of several megabits per second are being reported over Gambit Ethernets. This work has led to exploring protocols for high speed. ANSI X3T9.3 committee is exploring standards for 2-Gbits/s FDDI CPU-CPU protocols, and X3T9.5 is looking at 50 Mbits/s. 3These include: NETBLT, a high-speed Block Transfer protocol alternative to TCP/IP; a class of "lightweight" protocols, such as Lawrence Livermore Laboratory's "Delta-T," which eliminates error checks assuming highly reliable networks and/or re- dundant data (e.g., voice, graphics); and VMTP, Versatile Message Transfer Proto- col, which permits variable rate control and other negotiated options. Another area of research protocols is for distributed operating systems, Remote Procedure Calls (RPC) for Inter Processor Communication (IPC). Examples include MACH (Carnegie- Mellon), VKernel (Stanford), LOCUS (UCLA), GEMSOS (Gemini Computer Co.), and PLURIBUS (BBN).

OCR for page 107
COMPUTER NETWORKS 115 network control and management area. Such standards deal more with the structure of the components, traditionally a proprietary area, than with the data exchanges. There are fuzzy boundaries between networks, which create control and management problems. When a LAN injects erroneous traffic into a WAN gateway, there is no management responsibility to detect and react to repair. Public Computer Network Services Beginning with ARPANET and followed by TYMNET and TEI'ENET, there has been a steady growth of packet-switched com- mercial computer network utilities such as CompuServe, Mark TIT, BITNET, and ACCUNET. Initially, these utilities satisfied private commercial and government needs, but lately they have expandecl to service the growth of personal computers and home workstations. These utilities provide access to a variety of information services directly by modem or by "server hosts" attached to the networks via specialized databases and "bulletin board systems" (LEXIS, DIALOG, The Source, STARDYNE). The growth of entrepreneurial "on-line retailers" is a phenomenon unique to the United States. One monthly magazine lists nearly 1,000 "systems operators" around the United States that support special interest, dial-up bulletin boards. The most similar retailing case overseas is the videotext system in France, the network part of which is government sponsored and op- eratecI, as are many of the services. With globalization of business, increasing numbers of U.S. products and services are managed by computer networks, which may often provide effective (and possibly uncontrolled) export by multinational corporations, governments, and educational and scientific laboratories. Even individual states are getting into the network services business as a means of aiding focal industry (Brown, 1988~. The key elements in this growth have been the availability of personal computers equipped with Tow-cost, high-quaTity communi- cation modems, which modulate and demodulate data signals carried on public-switched networks. Even as modem speeds improved by a factor of 10 in a decade, costs declined by 50 to 75 percent (Fig- ure 5.2~. The combined effect was an impressive 43 percent com- pound annual growth rate (CAGR) of unit sales to the multimillion commodity volumes, as shown in Figure 5.3. Volume growth con- tinues to be fueled by the expanding PC base and the constant equipment upgrades to higher speeds by users.

OCR for page 107
116 4000 3000 CO ~ 2000- _ _ 1 000- 500 GL OBA L TR ENDS IN COMP UTER TE CHNOL O G Y 9600 bits/e 2400 bits/e - 1984 1985 1986 1987 1988 1989 1990 FIGURE 5.2 Modem price estimates by ye=. SOURCE: Courtesy of Gartner Group. 5000 4500 In o s In ._ o CO a) Q E of 4000 3500 3000 2500 2000 1500 1 000 500 o 1 991 1 992 1981 - 1 986 - 1594 1 654 930 - 380 120 F';;;;1 ~ . . . F:::::::::1 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 189 1991 CAGR = 43% 1988 CAGR = 23% 4189 FIGURE 5.3 U.S. personal computer modem shipments by year. SOURCE: Courtesy of Future Computing, Inc.

OCR for page 107
COMPUTER NETWORKS 117 BREAKTHROUGH POSSIBILITIES Product Innovation Computer networks is a field in which product innovation has slowed to the rate of standards agreements, which are notoriously slowed by technological complexity and competing commercial in- terests. Protocol software is among the most complex product yet invented. There are minions of asynchronous machine states, poorly specified protocols, and protocol implementations manufactured by competing vendors that lack seasoned, thorough, standardized life- cycle test suites. Furthermore, there are competing standards by DODandISO. Breakthroughs are most likely to occur in new protocols in the upper layers of the OSI/DOD models, driven by higher speed me(lia, large volumes of interconnected workstations, and new services of the variety noted earlier. Early computer nets were constructed by host computers simulating such T/O devices as a disk or a Teletype terminal. Later, these simulated devices were made more general to create a "typical" computer host with standard functional character- istics, so-called "virtuaTization." A real host would then use software to map its real characteristics to those specified in the virtual de- vice, for example, a virtual computer with virtual terminals, virtual memory, and virtual ports. By virtuaTizing the hosts, any user on the computer network could use the resources of a distant host in a manner similar to his local host, regardless of the computer vendor, system software, or peripheral device configuration. Today's technology has moved to virtuaTize the software services of the host computers; there are virtual operating system, files, mail, process calls, and command languages. Future technology wiB climb the software ladcler to higher levels of software applications. The time is now for virtual spreadsheets, virtual word processing, and virtual database management systems (DBMS), called in the popular literature "network versions." With agreed and de facto standards has come growing portability of data objects (disk files) between products by different manufacturers of these application- level tools. Most of the high-end word processing, spreadsheet, and DBMS software has network versions that permit shared use by multiple users on the net and distributed applications protocols built from these standards. IBM's Structured Query Language (SQL) is rapidly becoming such a protocol for distributed database systems.

OCR for page 107
118 GLOBAL TRENDS IN COMPUTER TECHNOLOGY Production Technology Breakthroughs are likely in production innovation. Produc- tion technology for computer nets would make the product faster, cheaper, smaller, less in need of power, and so on. The technologies of interest here are in protocol chips and processors, such as RISC, pipeline, or parallel design. Nets have very large numbers of states; that is, they are complex machines with mostly complex software. Breakthroughs are needed in software specification, software testing, and proof of correctness for protocols. Much of this technology is similar to that reported in Chapter 4 (on software) and will not be repeated here. However, network software has the added complexity and benefit of permitting communal development via a net. The United States is a leader in design, manufacturing, and testing of protocols because of its longer experience and its widely available network testbeds. This is a perishable lead that can be maintained by continued support from government and industry, and resolution of export licensing difficulties for such noncritical technologies as DOD protocol stacks, Tow-grade encryption, and network commodities. Many network products are rapidly reaching commodity status. Modem sales in the United States should top 10 minion annually in 1991 as described earlier. LAN components are also commodi- ties. A typical Ethernet LAN consists of a 10-Mbits/s cable system, cable taps, CSMA/CD level 1 protocols and X.25 level 2 protocols packaged on PC boards, and level 3 and above protocols packaged as PC software products. For mainframes and high-performance workstations, the protocol layers 1 through 4 are packaged as stand- alone network front ends (NFEs). LAN NFEs are rapidly replacing modems as preferred network interfaces, particularly in office and campus applications. The PC board products are rivaling modem prices, but with 10 to 100 times the performance. LEADING INDUSTRY PLAYERS All the worId's a stage for the network business. Key players are ISO, the International Consultative Committee on Telephone and Telegraphy (CCITT), multinational telephone companies, and defense departments of the United States and other nations, as well as computer and communications companies in the United States, France, England, Japan, West Germany, and Italy. The Soviets are players through ISO.

OCR for page 107
COMPUTER NETWORKS 119 Between 1982 and 1987 the world telecommunications market grew 5.4 percent per year: Europe, 4.1 percent per year; Asia, 8.1 percent per year; South America, 4.2 percent per year; and world remainder, 5.8 percent per year (CBEMA, 1987~. At the same time, the world market for computer, business, and telecommunications equipment grew 8.6 percent per year. These figures hint at the growth in international activity and interest in computer networking, some of which is highlighted in other sections of this chapter. PROTECTABILITY It is hard to protect a technology so dependent on international cooperation in specification and development of protocol standards. Furthermore, by the time one developer has the products, so do others. Consequently, out-producing the competition may be prefer- able to restricting product export. The U.S. lead in testing complex protocols can be a significant competitive edge. Also, production technology (e.g., protocol testing and verification technology) may be more effectively controlled with limited adverse impact on indus- try as compared with restricting product exports. Nevertheless, some communications products will continue to be regarded as critical to national security and, therefore, to be controlled. Computer Network Security Security is an emerging technology rapidly becoming a neces- sary requirement for a complex, democratic society in a world with a growing number of computer-sophisticated governments. The ba- sic problem is misplaced trust; humans trust their computers and networks with sensitive assets when there is often no basis for that trust. These are complex systems, rarely flawless, and the flaws permit exploitation of users and their sensitive codes and data. Through DOD research, several technologies such as trust en- gineering and cryptography are emerging to provide a sound basis for building secure systems. The United States has a significant lead through DOD applications, but export control policies make it difficult for U.S. industry to be more competitive. Basic Vulnerability Networks are subject to human exploitation of flaws inherent in the complexity of modern computer systems containing millions

OCR for page 107
120 GL OBAL TRENDS IN COMP UTER TECHNOL O G Y of instructions and data items. An unauthorized value in any one instruction or data item could lead to a violation of security. Modern computer hardware is designed for easy logic board maintenance and replacement. Unauthorized boards with bogus logic can easily sub- stitute for original parts if physical machine access is uncontrolled. Software production technology provides years of opportunity and dozens of methods for substituting malicious code for original code (so caned Trojan Horse, virus, or trap-door software) in the end-user computer system. The trusting, but possibly naive, employee risks assets to a flawed computer system in the unsuspecting belief that it is safe for holding sensitive assets. Even relatively sophisticated com- puter users are vulnerable, as the recent virus/worm attack on the ARPANET illustrates (Markoff, 1988~. Providing assurance that the risk is small is the special task of computer security. Security assur- ance forces trust engineering to design defenses and countermeasures to threats that reduce risk to acceptable limits.4 Legal and Institutional Mends In the United States, the National Institute of Standards and Technology (NIST) (formerly, the National Bureau of Stan(lards) and the National Security Agency (NSA) have played the largest roles in providing technical measures to improve the trustworthiness of computer systems. N[ST has led in government and commercial security standards; most noteworthy are risk assesment, audit, dis- aster recovery, and the data encryption standard (DES). NSA has led in DOD efforts to establish security standards for military-grade encryption and trusted computer systems and nets. Security Technical Standards Trusted Computer System Evaluation Criteria (TCSEC) are de- fined for DOD application by the National Computer Security Center (NCSC) of the NSA (DOD 5200.28-STD). These criteria are having an influence on secure systems beyond the DOD, including domestic 4 Most industrialized countries have begun to recognize the problem. Initial con- cerns focused on the threat of omnipotent government computers interfering with the privacy of citizens. One outcome was laws protecting individual privacy in mandatory government record keeping. Citizen privacy continues to be a concern. In addition, government and private systems now need protection from hackers, criminals, political opponents, and spies.

OCR for page 107
COMP UTER NETWORKS 121 commercial systems, and on international military and commercial applications. Essentially, the TCSEC seeks a "security triad" in the design and implementation of trusted computers and networks- security policy, enforcement mechanism, and assurance evidence. AD three triad elements must be present and in balance to achieve a trust rating. There are seven trust ratings: A1 (highest), B3, B2, B1, C2, C1, D (no trust).5 NSA has stated in public that A1 and B3 rated systems are subject to export control because such systems will be employed most heavily in classified applications. However, classified systems wiB require export license regardless of their use of security technology, and it may be asked what is gained by control- ling export of commercial products at any TCSEC rated level when DOD systems already require export control as classified military weapons. The whole computer and communications security export policy is an area warranting further study of the national security and competitiveness trade-off. Cryptography By law (Title 22 Code of Federal Regulations Part 121, Subchap- ter M, International Traffic in Arms Regulations [ITAR]), crypto- graphic products are subject to export control. Fromits founding by President Truman's Presidential Memorandum of October 1952, NSA has had a virtual monopoly on control of U.S. encryption tech- nology (Jelen, 1985~. That exclusivity was weakened somewhat in the past decade by commercial and academic nontraditional applica- tions of cryptography to computers and nets. The traditional uses of cryptography are to maintain data confidentiality in point-to-point communications. Newer technologies are coming on-line that provide other services. These include: ~ Sender identification/authentication: implied by possession of the encryption key. . Receiver identification/authentication: implied by possession of the decryption key. . Integrity: use cryptography to protect the integrity of data by an unforgeable integrity code (e.g., a "cryptoseal"~. 5 C-level trust serves most commercial applications, whereas the A and B levels employ mandated security controls favored by the DOD. Some highly sensitive non-DOD applications also require B-level systems, for example, the Treasury, DEA, Justice, and the stock market.

OCR for page 107
122 GLOBAL TRENDS IN COMPUTER TECHNOLOGY ~ Certification authority: a trusted third party provides or ex- changes cryptosealed credentials among a cooperating group. The credentials can include predefined and negotiated data: identifica- tion, passwords, credentials, electronic signatures, dates, time, and so on. Non-repudiation: an electronic signature and message receipt are recorded (e.g., certified mail) by a trusted third party for later verification of the transaction, transaction initiator, and transaction receiver. These applications are growing in the commercial market and will become more important to U.S. competitiveness. This is an- other area that warrants further study of the national security and competitiveness trade-off. Data Encryption Standard The National Bureau of Standards (now NIST) developed in the late 1970s a data encryption-aIgorithm standard (DES) for U.S. government and domestic commercial use. It has been openly pub- lished and copied around the world ever since, and has recently been endorsed by NIST for another five years. The ITAR require export license for U.S. products using DES, and this appears to be a case of unnecessary export control diminishing U.S. competitiveness. DES or comparable cryptography is available from multiple international vendors. Public Key Cryptography Public key cryptography (PKC) is a two-key encryption system one key for encryption and a different key for decryption devel- oped by researchers at Stanford and MIT. PKC is based on factoring theory and modulus arithmetic multiplying and dividing very large primes. PKC may be used to simplify key management because the public keys only require integrity protection (not confidentiality) while being stored or transferred. PKC systems have become quite popular in commercial applications worldwide. CCITT X.509 is a proposed international standard, and its security is based on PKC. Smart Cards: An Example of a Lost Network Security Business Among the growth applications of these security techniques is the smart card, a credit card containing memory and computer logic.

OCR for page 107
COMPUTER NETWORKS 123 VISA International has 140 minion bank cards in circulation in Eu- rope, and many of them are being upgraded to smart cards. Toshiba and Casio are pursuing dramatic advances in smart cards, with func- tionality rivaling a watch, calculator, credit card, and address book combined. In Europe, Thomson, Bull, Phillips, Siemens, and V OEST (Austria) are prominent in smart card production, with BuD and Phillips garnering 15 percent of their revenue from smart cards and their chips.6 The United States has essentially ignored and lost this growth market. DOD Security Requirements at Odds with ISO The Department of Defense has always held that security is its main reason for resisting ISO standards. Although efforts are under way by NIST to merge these standards, security is a wedge keep- ing them apart. DOD is active in network security though various classified voice and data programs (e.g., STU 2, STU 3, BLACKER, and SDNS). Of significance is that security affects protocols, driv- ing them away from standards if the standards do not satisfy their often classified requirements. Given the threat requirements that drive them, export of these security protocols is tightly controlled. Security protocols are unlikely to be merged with ISO, and this is a policy deadlock without clear signs of solution. It may be wise to formalize this de facto duality to protect both national security and competitiveness. U.S. Industry Resistance With literaBy billions of dollars required to manufacture, main- ta~n, and market major computer components, industry will resist having serious government restrictions imposed on its markets. There was a rush to build NSA C-rated secure operating systems compat- ible with commercial offerings, and such systems are now successful and not export controlled. There has been resistance by industry to upgrading or building new products having the higher ratings. After five years, there have been no products added to the NSA Evaluated Products List (EPL) at ratings B2 or above. 6Typical card configurations include 1.2-micron CMOS 8-bit CPU, 4k ROM, 2k EPROM, 256 bytes RAM, DES, and perform user identification, authentication, encrypt tion, data processing, and secure writing. ISO STD 78 defines the physical and electrical characteristics. IFIP TC11 is working on computer and data standards.

OCR for page 107
124 GLOBAL TRENDS IN COMPUTER TECHNOLOGY Another example is the Toss of the encryption market to Japan and Western Europe attributable to the ITAR and NSA's monopoly on domestic encryption technology. NIST supports recertification of DES for another five years. NSA does not concur with the recertifica- tion of DES but NSA favors new standards based on its Commercial COMSEC Endorsement Program (CCEP). Via CCEP, NSA is build- ing new encryption standards for both classified (Type 1 devices) and unclassified sensitive government and industry use (Type 2 devices). It favors general industrial use of Data Standard DS-72, a CCEP Type 2 standard in lieu of DES. Industry is reTuct ant to abandon a known technology with products available and compatible with overseas equipment, even if it is not manufactured in the United States. l CONCLUSIONS Computer networking epitomizes the dual-use nature of com- puter technology. It serves as the backbone of modern military com- mand and control systems as well as commercial office and factory automation systems and other civilian applications. Standards, particularly international standards, increasingly drive the development of computer network products. In the United States, product development is affected by a split between DOD- favored standards and international, commercial standards. This split has adversely affected U.S. company positions in computer net- work markets. The United States is a leader in design, manufacturing, and testing of protocols because of its longer experience and its widely available computer network testbeds. But this is a perishable lead that can only be maintained by continued support from government and industry and by resolution of export licensing difficulties for such noncritical technologies as DOD protocol stacks, Tow-grade encryp- tion, and computer network commodities. Further study is needed on a host of computer-network-related issues, including security and control of access to U.S. and inter- national research computer networks, transborder flows of computer and communications technologies via computer networks, and special trade regulations. For example:

OCR for page 107
COMPUTER NETWORKS 125 ~ Transborder computer network access is rampant among in- dividuals and multinational corporations via private and public com- puter networks. How is export control of "soft" technologies software, algorithms, specifications, end reports to tee controlled on such computer networks? ~ Is it in the best national security interest of the United States to permit CMEA access to commercial and university computer net- works, both directly while in the United States and through remote t e ~ e c o m m u n i c a t i o n s ? ~ Are the existing ITAR necessary or relevant to modern com- mercia] (nonmilitary) security needs encryption and trusted systems in banking and computer network retailing? They may retard U.S. competitiveness as non-CoCom sources grow stronger.