Debian Codegraph @

rfc939.html

raw
1 <pre>Network Working Group National Research Council
2 Request for Comments: 939
3 February 1985
4
5 <span class="h1">Executive Summary</span>
6 <span class="h1">of the NRC Report on</span>
7 <span class="h1">Transport Protocols for</span>
8 Department of Defense
9 Data Networks
10
11
12 STATUS OF THIS MEMO
13
14 This RFC is distributed for information only. This RFC does not
15 establish any policy for the DARPA research community or the DDN
16 operational community. Distribution of this memo is unlimited.
17
18 INTRODUCTION
19
20 This RFC reproduces the material from the "front pages" of the
21 National Research Council report resulting from a study of the DOD
22 Internet Protocol (IP) and Transmission Control Protocol (TCP) in
23 comparison with the ISO Internet Protocol (ISO-IP) and Transport
24 Protocol level 4 (TP-4). The point of this RFC is to make the text
25 of the Executive Summary widely available in a timely way. The order
26 of presentation has been altered, and the pagination changed.
27
28 The title of the full report is:
29
30
31
32 Transport Protocols for
33 Department of Defense
34 Data Networks
35
36 Report to the Department of Defense
37 and the National Bureau of Standards
38
39 Committee on Computer-Computer Communication Protocols
40
41 Board on Telecommunications and Computer Applications Commission on
42 Engineering and Technical Systems
43 National Research Council
44
45 National Academy Press
46 Washington, D.C. February 1985
47
48
49
50
51
52
53
54 <span class="grey">National Research Council [Page 1]</span></pre>
55 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
56 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
57 Executive Summary of the NRC Report Transport on Protocols
58
59
60 OVERVIEW
61
62 The project that is the subject of this report was approved by the
63 Governing Board on the National Research Council, whose members are
64 drawn from the councils of the National Academy of Sciences, the
65 National Academy of Engineering, and the Institute of Medicine. The
66 members of the committee responsible for the report were chosen for
67 their special competences and with regard for appropriate balance.
68
69 This report has been reviewed by a group other than the authors,
70 according to procedures approved by a Report Review Committee
71 consisting of members of the National Academy of Sciences, the
72 National Academy of Engineering, and the Institute of Medicine.
73
74 The National Research Council was established by the National Academy
75 of Sciences in 1916 to associate the broad community of science and
76 technology with the Academy's purposes of furthering knowledge and of
77 advising the federal government. The Council operates in accordance
78 with general policies determined by the Academy under the authority
79 of its congressional charter of 1863, which establishes the Academy
80 as a private, nonprofit, self-governing membership corporation. The
81 Council has become the principal operating agency of both the
82 National Academy of Sciences and the National Academy of Engineering
83 in the conduct of their services to the government, the public, and
84 the scientific and engineering communities. It is administered
85 jointly by both Academies and the Institute of Medicine. The
86 National Academy of Engineering and the Institute of Medicine were
87 established in 1964 and 1970, respectively, under the charter of the
88 National Academy of Sciences.
89
90 This is a report of work supported by Contract No. DCA-83-C-0051
91 between the U.S. Defense Communications Agency and the National
92 Academy of Sciences, underwritten jointly by the Department of
93 Defense and the National Bureau of Standards.
94
95 Copies of the full report are available from:
96
97 Board on Telecommunications and Computer Applications Commission
98 on Engineering and Technical Systems
99 National Research Council
100 2101 Constitution Avenue, N.W.
101 Washington, D.C. 20418
102
103
104
105
106
107
108
109 <span class="grey">National Research Council [Page 2]</span></pre>
110 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
111 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
112 Executive Summary of the NRC Report Transport on Protocols
113
114
115 PREFACE
116
117 This is the final report of the National Research Council Committee
118 on Computer-Computer Communication Protocols. The committee was
119 established in May l983 at the request of the Department of Defense
120 (DOD) and the National Bureau of Standards (NBS), Department of
121 Commerce, to develop recommendations and guidelines for resolving
122 differences between the two agencies on a data communications
123 transport protocol standard.
124
125 Computer-based information and transaction-processing systems are
126 basic tools in modern industry and government. Over the past several
127 years there has been a growing demand to transfer and exchange
128 digitized data in these systems quickly and accurately. This demand
129 for data transfer and exchange has been both among the terminals and
130 computers within an organization and among those in different
131 organizations.
132
133 Rapid electronic transport of digitized data requires electronic
134 communication links that tie the elements together. These links are
135 established, organized, and maintained by means of a layered series
136 of procedures performing the many functions inherent in the
137 communications process. The successful movement of digitized data
138 depends upon the participants using identical or compatible
139 procedures, or protocols.
140
141 The DOD and NBS have each developed and promulgated a transport
142 protocol as standard. The two protocols, however, are dissimilar and
143 incompatible. The committee was called to resolve the differences
144 between these protocols.
145
146 The committee held its first meeting in August l983 at the National
147 Research Council in Washington, D.C. Following this two-day meeting
148 the committee held five more two-day meetings, a three-day meeting,
149 and a one-week workshop.
150
151 The committee was briefed by personnel from both agencies. In
152 addition, the committee heard from Jon Postel, University of Southern
153 California's Information Sciences Institute; Dave Oran, Digital
154 Equipment Corporation; Vinton Cerf, MCI; David Wood, The Mitre
155 Corporation; Clair Miller, Honeywell, and Robert Follett, IBM,
156 representing the Computer and Business Equipment Manufacturer's
157 Association; and John Newman, Ultimate Corporation. In most cases
158 the briefings were followed by discussion.
159
160 The committee wishes to thank Philip Selvaggi of the Department of
161 Defense and Robert Blanc of the NBS, Institute of Computer Sciences
162
163
164 <span class="grey">National Research Council [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
167 Executive Summary of the NRC Report Transport on Protocols
168
169
170 and Technology, for their cooperation as their agency's liaison
171 representatives to the committee. The committee appreciates the
172 contributions and support of Richard B. Marsten, Executive Director
173 of the Board on Telecommunications -- Computer Applications (BOTCAP),
174 and Jerome D. Rosenberg, BOTCAP Senior Staff Officer and the
175 committee Study Director. We also wish to thank Lois A. Leak for her
176 expert administrative and secretarial support.
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219 <span class="grey">National Research Council [Page 4]</span></pre>
220 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
221 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
222 Executive Summary of the NRC Report Transport on Protocols
223
224
225 EXECUTIVE SUMMARY
226
227 Computer communication networks have become a very important part of
228 military and commercial operations. Indeed, the nation is becoming
229 dependent upon their efficiency and reliability, and the recent
230 proliferation of networks and their widespread use have emphasized
231 the importance of developing uniform conventions, or protocols, for
232 communication between computer systems. The Department of Defense
233 (DOD) and the National Bureau of Standards (NBS) have been actively
234 engaged in activities related to protocol standardization. This
235 report is concerned primarily with recommendations on protocol
236 standardization within the Department of Defense.
237
238 Department of Defense's Transmission Protocol
239
240 The DOD's Defense Advanced Research Projects Agency (DARPA) has
241 been conducting and supporting research on computer networks for
242 over fifteen years (1). These efforts led to the development of
243 modern packet-switched network design concepts. Transmission
244 between computers is generally accomplished by packet switching
245 using strict protocols for the control and exchange of messages.
246 The Advanced Research Projects Agency network (ARPANET),
247 implemented in the early 1970s, provided a testing ground for
248 research on communications protocols. In 1978, after four years
249 of development, the DOD promulgated versions of its Transmission
250 Control Protocol (TCP) and an Internet Protocol (IP) and mandated
251 their use as standards within the DOD. TCP is now widely used and
252 accepted. These protocols meet the unique operational and
253 functional requirements of the DOD, and any changes in the
254 protocols are viewed with some trepidation by members of the
255 department. DOD representatives have stated that standardizing
256 TCP greatly increased the momentum within the DOD toward
257 establishing interoperability between networks within the DOD.
258
259 International Standards Organization's Transport Protocol
260
261 The NBS Institute for Computer Sciences and Technology (ICST), in
262 cooperation with the DOD, many industrial firms, and the
263 International Standards Organization (ISO), has developed a new
264 international standard
265
266 Transport Protocol (TP-4) and a new Internetwork Protocol (2).
267 These protocols will soon be available as commercial products.
268 Although in part derived from TCP, the new protocols are not
269 compatible with TCP (3). The U.S. standards organizations are
270
271
272
273
274 <span class="grey">National Research Council [Page 5]</span></pre>
275 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
276 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
277 Executive Summary of the NRC Report Transport on Protocols
278
279
280 supporting TP-4 in international operations, and the Department of
281 Commerce is proposing TP-4 as a Federal Information Processing
282 Standard (FIPS) for use by all federal agencies.
283
284 DOD OPERATIONAL AND TECHNICAL NEEDS
285
286 The DOD has unique needs that could be affected by the Transport
287 and Internet Protocol layers. Although all data networks must
288 have some of these capabilities, the DOD's needs for operational
289 readiness, mobilization, and war-fighting capabilities are
290 extreme. These needs include the following:
291
292 Survivability--Some networks must function, albeit at reduced
293 performance, after many nodes and links have been destroyed.
294
295 Security--Traffic patterns and data must be selectively
296 protected through encryption, access control, auditing, and
297 routing.
298
299 Precedence--Systems should adjust the quality of service on the
300 basis of priority of use; this includes a capability to preempt
301 services in cases of very high priority.
302
303 Robustness--The system must not fail or suffer much loss of
304 capability because of unpredicted situations, unexpected loads,
305 or misuse. An international crisis is the strongest test of
306 robustness, since the system must operate immediately and with
307 virtually full performance when an international situation
308 flares up unexpectedly.
309
310 Availability--Elements of the system needed for operational
311 readiness or fighting must be continuously available.
312
313 Interoperability--Different elements of the Department must be
314 able to "talk" to one another, often in unpredicted ways
315 between parties that had not planned to interoperate.
316
317 These operational needs reflect themselves into five technical or
318 managerial needs:
319
320 1. Functional and operational specifications (that is, will
321 the protocol designs meet the operational needs?);
322
323 2. Maximum interoperability;
324
325 3. Minimum procurement, development, and support costs;
326
327
328
329 <span class="grey">National Research Council [Page 6]</span></pre>
330 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
331 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
332 Executive Summary of the NRC Report Transport on Protocols
333
334
335 4. Ease of transition to new protocols; and
336
337 5. Manageability and responsiveness to changing DOD
338 requirements.
339
340 These are the criteria against which DOD options for using the ISO
341 transport and internet protocols should be evaluated.
342
343 Interoperability is a very important DOD need. Ideally, DOD
344 networks would permit operators at any terminal to access or be
345 accessed by applications in any computer. This would provide more
346 network power for users, integration of independently developed
347 systems, better use of resources, and increased survivability. To
348 increase interoperability, the Office of the Secretary of Defense
349 has mandated the use of TCP for the Defense Communication System's
350 Defense Data Network (DDN), unless waivers are granted. In
351 addition, the Defense Communication Agency (DCA) is establishing
352 standards for three higher-level "utility" protocols for file
353 transfer, terminal access, and electronic mail. Partly as a
354 result of these actions, it has become clear that there is growing
355 momentum toward accepting interoperability and a recognition that
356 it is an important operational need.
357
358 It is very important, however, to recognize that functional
359 interoperability is only achieved with full generality when two
360 communication nodes can interoperate at all protocol levels. For
361 the DOD the relevant levels are as follows:
362
363 1. Internet, using IP;
364
365 2. Transport, using TCP;
366
367 3. Utility, using file, terminal, or mail protocols; and
368
369 4. Specific applications that use the above protocols for
370 their particular purpose.
371
372 Accordingly, if a network is developed using one transport
373 protocol, it would generally not be able to interoperate
374 functionally with other networks using the same transport protocol
375 unless both networks were also using the higher-level utility and
376 application protocols. In evaluating whether or not to convert to
377 TP-4 and in developing a transition plan, the following factors
378 must be considered:
379
380 The DOD contains numerous communities of interest whose
381 principal need is to interoperate within their own members,
382
383
384 <span class="grey">National Research Council [Page 7]</span></pre>
385 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
386 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
387 Executive Summary of the NRC Report Transport on Protocols
388
389
390 independently. Such communities generally have a specific,
391 well-defined mission. The DOD Intelligence Information System
392 (DODIIS) and the World Wide Military Command and Control System
393 (WWMCCS) are examples. Interoperability is needed primarily
394 between the higher layer applications programs initially unique
395 to each community of interest.
396
397 There are many different kinds of operations needed between
398 communities of interest. Examples of such operations are
399 headquarters' need for access to several subordinate
400 communities and the communities' need for some minimum
401 functional interoperability with each other (such as mail
402 exchange).
403
404 The need for functional interoperability can arise,
405 unexpectedly and urgently, at a time of crisis or when improved
406 management opportunities are discovered. Widespread
407 standardization of TP-4 and higher-level protocols can readily
408 help to achieve these needs. Often, special development of
409 additional applications that cost time and money will be
410 necessary.
411
412 The DOD needs functional interoperability with many important
413 external agencies that are committed to ISO standards: The
414 North Atlantic Treaty Organization (NATO), some intelligence
415 and security agencies, and other parts of the federal
416 government.
417
418 The same objectives that have prompted the use of standardized
419 protocols at higher-level headquarters will lead to their use
420 by tactical groups in the field.
421
422 SOME COMPARISONS
423
424 A detailed comparison of the DOD Transmission Control Protocol and
425 the ISO Transport Protocol indicates they are functionally
426 equivalent and provide essentially similar services. Because it
427 is clear that a great deal of care and experience in protocol
428 development have gone into generating the specifications for TP-4,
429 the committee is confident that TP-4 will meet military
430 requirements.
431
432 Although there are differences between the two protocols, they do
433 not compromise DOD requirements. And, although in several areas,
434 including the data transfer interface, flow control, connection
435 establishment, and out-of-band, services are provided in different
436 ways by the two protocols, neither seems intrinsically superior.
437
438
439 <span class="grey">National Research Council [Page 8]</span></pre>
440 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
441 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
442 Executive Summary of the NRC Report Transport on Protocols
443
444
445 Thus, while existing applications may need to be modified somewhat
446 if moved from TCP to TP-4, new applications can be written to use
447 either protocol with a similar level of effort.
448
449 The TCP and TP-4 protocols are sufficiently equivalent in their
450 security-related properties in that there are no significant
451 technical points favoring the use of one over the other.
452
453 While TCP currently has the edge in maturity of implementation,
454 TP-4 is gaining rapidly due to the worldwide support for and
455 acceptance of the Open System Interconnection (OSI) international
456 standards. Experimental TCP implementations were completed in
457 1974 at Stanford University and BBN Communications Corporation.
458 Between 1974 and 1982 a large number of implementations were
459 produced. The Defense Advanced Research Projects Agency (ARPA)
460 network switched to a complete use of TCP in January 1983.
461 Operations have been satisfactory and its use is growing. A
462 number of TCP implementations are also in commercial use in
463 various private networks.
464
465 In contrast, TP-4 has not yet been implemented in any large
466 operational system. It has been tested experimentally, however,
467 and has received endorsement by many commercial vendors worldwide.
468 In addition, substantial portions of TP-4 have been demonstrated
469 at the National Computer Conference in July 1984.
470
471 The Internet Protocol (IP) part of the standards is not believed
472 to be a problem. The ISO IP is not as far along as TP-4, but it
473 is much less complex. The ISO IP, based very strongly on the DOD
474 IP, became a draft international standard in April 1984.
475
476 The rapidity of the progress in ISO and the results achieved over
477 the past two years have surprised even the supporters of
478 international standards. The reasons for this progress are
479 twofold: strong market demands stemming from the growing
480 integration of communications and data processing and the progress
481 in networking technology over the past years as the result of ARPA
482 and commercial developments.
483
484 Although the DOD networks have been a model upon which the ISO
485 transport standards have been built, the rest of the world is
486 adopting TP-4. Because the DOD represents a small fraction of the
487 market and because the United States supports the ISO standard, it
488 is not realistic to hope that TP-4 can be altered to conform with
489 TCP. This raises the question as to what action should be taken
490 by the DOD with respect to the ISO standard.
491
492
493
494 <span class="grey">National Research Council [Page 9]</span></pre>
495 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
496 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
497 Executive Summary of the NRC Report Transport on Protocols
498
499
500 SOME ECONOMIC CONSIDERATIONS
501
502 The DOD has a large and growing commitment in operational TCP
503 networks, and this will increase by 50 to 100 percent in the next
504 eighteen months. This rate of investment will probably continue
505 for the next five years for new systems and the upgrading of
506 current ones. The current Military Network (MILNET) and Movement
507 Information Network (MINET) systems are expanding and will shortly
508 be combined. The Strategic Air Command Digital Information
509 Network (SACDIN) and DODIIS are undergoing major upgrading. When
510 these changes are completed, there are plans to upgrade the WWMCCS
511 Intercomputer Network (WIN) and to add separate SECRET and TOP
512 SECRET networks. There are plans to combine these six networks in
513 the late 1980s, and they will become interoperable and multilevel
514 secure using an advanced technology now under development. If
515 these plans are implemented on schedule, a delay of several years
516 in moving to TP-4 would mean that the DOD networks in the late
517 1980s would be virtually all TCP-based. Subsequent conversion to
518 international standards would be very expensive if hastily
519 attempted in order to maintain established DOD interoperability
520 and gain interoperability with a large body of users.
521
522 As the Department of Defense policy recognizes, there are
523 significant advantages in using commercial vendor products if they
524 meet the department's operational needs. The major advantages are
525 as follows:
526
527 Costs to the DOD for development, production, and maintenance
528 are significantly lower because (1) vendors spread the cost
529 over a much larger user base, (2) commercial vendors are
530 generally more efficient in their operations, and (3) vendors
531 look for ways to improve their product to meet competition.
532
533 The department generally gets more effective products because
534 vendors integrate the protocol functions into their entire
535 software and hardware product line. Thus the DOD may be able
536 eventually to use commercial software products that are built
537 on top of, and thereby take advantage of, the transport
538 protocols.
539
540 By depending on industry to manage the development and
541 maintenance of products, the department can use its scarce
542 management and technical resources on activities unique to its
543 mission.
544
545 Because the costs of transport and internet protocol development
546 and maintenance are so intertwined with other factors, it is
547
548
549 <span class="grey">National Research Council [Page 10]</span></pre>
550 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
551 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
552 Executive Summary of the NRC Report Transport on Protocols
553
554
555 impossible to give a precise estimate of the savings that would be
556 achieved by using commercial products. Savings will vary in
557 individual cases. The marginal savings should range from 30 to 80
558 percent.
559
560 RECOMMENDATIONS
561
562 The ISO protocols are now well specified but will not generally be
563 commercially available for many months. Nevertheless, this
564 committee believes that the principles on which they are based are
565 well-established, and the protocols can be made to satisfy fully
566 DOD's needs. The committee recommends that the DOD move toward
567 adoption of TP-4 as costandard with TCP and toward exclusive use
568 of TP-4.
569
570 Transition to the use of the ISO standards, however, must be
571 managed in a manner that will maintain DOD's operational
572 capabilities and minimize risks. The timing of the transition is,
573 therefore, a major concern.
574
575 Descriptions of two options that take this requirement into
576 account follow. A majority of the committee recommends the first
577 option, while a minority favors the second. A third option--to
578 defer action--is also described but not recommended.
579
580 Option 1
581
582 The first option is for the DOD to immediately modify its
583 current transport policy statement to specify TP-4 as a
584 costandard along with TCP. In addition, the DOD would develop
585 a military specification for TP-4 that would also cover DOD
586 requirements for discretionary options allowed under the NBS
587 protocol specifications. Requests for proposals (RFPs) for new
588 networks or major upgrades of existing networks would specify
589 TP-4 as the preferred protocol. Contracts for TP-4 systems
590 would be awarded only to contractors providing commercial
591 products, except for unique cases.
592
593 Existing networks that use TCP and new networks firmly
594 committed to the use of TCP-based systems could continue to
595 acquire implementations of TCP. The DOD should carefully
596 review each case, however, to see whether it would be
597 advantageous to delay or modify some of these acquisitions in
598 order to use commercial TP-4 products. For each community of
599 users it should be decided when it is operationally or
600
601
602
603
604 <span class="grey">National Research Council [Page 11]</span></pre>
605 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
606 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
607 Executive Summary of the NRC Report Transport on Protocols
608
609
610 economically most advantageous to replace its current or
611 planned systems in order to conform to ISO standards without
612 excessively compromising continued operations.
613
614 United States government test facilities would be developed to
615 enable validation of TP-4 products (4). The Department of
616 Defense would either require that products be validated using
617 these test facilities or that they be certified by the vendor.
618 The test facilities could also be used to isolate multivendor
619 protocol compatibility problems. The existing NBS validation
620 tools should be used as the base for the DOD test facilities.
621
622 Because under this option networks based on both TCP and TP-4
623 would coexist for some time, several capabilities that
624 facilitate interoperability among networks would need to be
625 developed. The Department of Defense generally will not find
626 them commercially available. Examples are gateways among
627 networks or specialized hosts that provide services such as
628 electronic mail. The department would need to initiate or
629 modify development programs to provide these capabilities, and
630 a test and demonstration network would be required.
631
632 Option 2
633
634 Under Option 2 the Department of Defense would immediately
635 announce its intention to adopt TP-4 as a transport protocol
636 costandard with TCP after a satisfactory demonstration of its
637 suitability for use in military networks. A final commitment
638 would be deferred until the demonstration has been evaluated
639 and TP-4 is commercially available.
640
641 The demonstration should take at most eighteen months and
642 should involve development of TP-4 implementations and their
643 installation. This option differs from Option 1 primarily in
644 postponing the adoption of a TP-4 standard and, consequently,
645 the issuance of RFPs based on TP-4 until successful completion
646 of a demonstration. The department, however, should proceed
647 with those provisions of Option 1 that may be completed in
648 parallel with the demonstration. Early issuance of a TP-4
649 military specification, development of validation procedures,
650 and implementation of means for interoperability would be
651 particularly important in this regard.
652
653
654
655
656
657
658
659 <span class="grey">National Research Council [Page 12]</span></pre>
660 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
661 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
662 Executive Summary of the NRC Report Transport on Protocols
663
664
665 Option 3
666
667 Under the third option the DOD would continue using TCP as the
668 accepted transport standard and defer any decision on the use
669 of TP-4 indefinitely. The department would be expected to stay
670 well informed on the development and use of the new protocol in
671 the commercial and international arena and, with the National
672 Bureau of Standards, work on means to transfer data between the
673 two protocol systems. Testing and evaluation of TP-4 standards
674 by NBS would continue. The DOD might eventually accommodate
675 both protocol systems in an evolutionary conversion to TP-4.
676
677 Comparison of Options
678
679 The committee believes that all three options equally satisfy
680 the functional objectives of the DOD, including matters of
681 security. It believes the two protocols are sufficiently
682 similar and no significant differences in performance are to be
683 expected if the chosen protocol implementation is of equal
684 quality and is optimized for the given environment.
685
686 The primary motivation for recommending Option 1 is to obtain
687 the benefits of standard commercial products in the
688 communication protocol area at an early date. Benefits include
689 smaller development, procurement, and support costs; more
690 timely updates; and a wider product availability. By
691 immediately committing to TP-4 as a costandard for new systems,
692 Option 1 minimizes the number of systems that have to be
693 converted eventually from TCP. The ability to manage the
694 transition is better than with Option 2 since the number of
695 systems changed would be smaller and the time duration of mixed
696 TCP and TP-4 operation would be shorter. Interoperability with
697 external systems (NATO, government, commercial), which
698 presumably will also use TP-4, would be brought about more
699 quickly. Option 1 involves greater risk, however, since it
700 commits to a new approach without as complete a demonstration
701 of its viability.
702
703 As with Option 1, a primary benefit of following Option 2 would
704 be obtaining the use of standard commercial products. Unit
705 procurement costs probably would be lower than with Option 1
706 because the commercial market for TP-4 will have expanded
707 somewhat by the time DOD would begin to buy TP-4 products.
708 Risk is smaller, compared to Option 1, because testing and
709 demonstration of the suitability for military use will have
710 preceded the commitment to the ISO protocols. Transition and
711 support costs would be higher than for Option 1, however,
712
713
714 <span class="grey">National Research Council [Page 13]</span></pre>
715 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
716 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
717 Executive Summary of the NRC Report Transport on Protocols
718
719
720 because more networks and systems would already have been
721 implemented with TCP. Also this is perhaps the most difficult
722 option to manage since the largest number of system conversions
723 and the longest interval of mixed TCP and TP-4 operations would
724 occur. In addition, interoperability with external networks
725 through standardization would be delayed.
726
727 The principal benefit of exercising Option 3 would be the
728 elimination of transition cost and the risk of faulty system
729 behavior and delay. It would allow the most rapid achievement
730 of full internal interoperability among DOD systems.
731 Manageability should be good because only one set of protocols
732 would be in use (one with which the DOD already has much
733 experience), and because the DOD would be in complete control
734 of system evolution. Procurement costs for TCP systems would
735 remain high compared with standard ISO protocol products,
736 however, and availability of implementations for new systems
737 and releases would remain limited. External interoperability
738 with non-DOD systems would be limited and inefficient.
739
740 In summary, Option 1 provides the most rapid path toward the
741 use of commercial products and interoperability with external
742 systems. Option 2 reduces the risk but involves somewhat
743 greater delay and expense. Option 3 involves the least risk
744 and provides the quickest route to interoperability within the
745 Defense Department at the least short-term cost. These are,
746 however, accompanied by penalties of incompatibility with NATO
747 and other external systems and higher life-cycle costs.
748
749 NOTES:
750
751 (1) The Advanced Research Projects Agency (ARPA) was reorganized
752 and became the Defense Advanced Research Projects Agency
753 (DARPA) in 1973.
754
755 (2) The ISO Transport Protocol and ISO Internetwork Protocol
756 became Draft International Standards in September 1983 and
757 April 1984, respectively. Commercial vendors normally
758 consider Draft International Standards to be ready for
759 implementation.
760
761 (3) Except where noted, the abbreviation TCP generally refers to
762 both the DOD's Transmission Control Protocol and its Internet
763 Protocol. Similarly, the abbreviation TP-4 refers to both
764 the ISO Transport Protocol class 4 and its Internetwork
765 Protocol. (Transport Protocol classes 0 to 3 are used for
766 special purposes not related to those of this study.)
767
768
769 <span class="grey">National Research Council [Page 14]</span></pre>
770 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
771 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
772 Executive Summary of the NRC Report Transport on Protocols
773
774
775 (4) Validation means a systematic and thorough state-of-the-art
776 testing of the products to assure that all technical
777 specifications are being achieved.
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824 <span class="grey">National Research Council [Page 15]</span></pre>
825 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
826 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
827 Executive Summary of the NRC Report Transport on Protocols
828
829
830 CONTENTS OF THE FULL REPORT
831
832 PREFACE ......................................................... <a href="#page-ix">ix</a>
833
834 EXECUTIVE SUMMARY ............................................... <a href="#page-xi">xi</a>
835
836 <a href="#appendix-I">I</a> Introduction ............................................... <a href="#page-1">1</a>
837
838 II Review of NBS and DOD Objectives ........................... <a href="#page-3">3</a>
839
840 III Comparison of DOD and ISO Protocols ....................... <a href="#page-13">13</a>
841
842 IV Status of DOD and ISO Protocol
843 Implementations and Specifications ....................... <a href="#page-25">25</a>
844
845 <a href="#appendix-V">V</a> Markets ................................................... <a href="#page-31">31</a>
846
847 VI Development of Standard Commercial versus
848 Special Commercial Products ............................... <a href="#page-39">39</a>
849
850 VII Responsiveness of International Standards
851 Process to Change ......................................... <a href="#page-43">43</a>
852
853 VIII Options for DOD and NBS ................................... <a href="#page-45">45</a>
854
855 IX Cost Comparison of Options ............................... <a href="#page-47">47</a>
856
857 <a href="#appendix-X">X</a> Evaluation of Options ..................................... <a href="#page-53">53</a>
858
859 XI Recommendations ........................................... <a href="#page-61">61</a>
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879 <span class="grey">National Research Council [Page 16]</span></pre>
880 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
881 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
882 Executive Summary of the NRC Report Transport on Protocols
883
884
885 BOARD ON TELECOMMUNICATIONS -- COMPUTER APPLICATIONS
886 COMMITTEE ON COMPUTER-COMPUTER COMMUNICATION PROTOCOLS
887
888 Chairman
889
890 C. CHAPIN CUTLER, Professor of Applied Physics, Stanford
891 University, Stanford, California
892
893 Members
894
895 HERBERT D. BENINGTON, Technical Director, System Development
896 Corporation, McLean, Virginia
897
898 DONALD L. BOYD, Director, Honeywell Corporate Computer Sciences
899 Center, Honeywell Corporate Technology Center, Bloomington,
900 Minnesota
901
902 DAVID J. FARBER, Professor of Electrical Engineering and Professor
903 of Computer Science, Department of Electrical Engineering,
904 University of Delaware, Newark, Delaware
905
906 LAWRENCE H. LANDWEBER, Professor, Computer Sciences Department,
907 University of Wisconsin, Madison, Wisconsin
908
909 ANTHONY G. LAUCK, Manager, Distributed Systems Architecture and
910 Advanced Development, Digital Equipment Corporation, Tewksbury,
911 Massachusetts
912
913 KEITH A. LUCKE, General Manager of Control Data Technical
914 Standards, Control Data Corporation, Minneapolis, Minnesota
915
916 MISCHA SCHWARTZ, Professor of Electrical Engineering and Computer
917 Science, Columbia University, New York, New York
918
919 ROBERT F. STEEN, Director of Architecture, Communication Products
920 Division IBM Corporation, Research Triangle Park, North Carolina
921
922 CARL A. SUNSHINE, Principal Engineer, Sytek, Incorporated, Los
923 Angeles Operation, Culver City, California
924
925 DANIEL J. FINK, (Ex-officio), President, D.J. Fink Associates,
926 Inc., Arlington, Virginia
927
928 JAMES L. FLANAGAN, (CETS LIAISON MEMBER), Head, Acoustics Research
929 Department, AT&amp;T Bell Laboratories, Murray Hill, New Jersey
930
931
932
933
934 <span class="grey">National Research Council [Page 17]</span></pre>
935 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
936 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
937 Executive Summary of the NRC Report Transport on Protocols
938
939
940 Staff
941
942 RICHARD B. MARSTEN, Executive Director
943 JEROME D. ROSENBERG, Senior Staff Officer and Study Director
944 LOIS A. LEAK, Administrative Secretary
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989 <span class="grey">National Research Council [Page 18]</span></pre>
990 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
991 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
992 Executive Summary of the NRC Report Transport on Protocols
993
994
995 COMMISSION ON ENGINEERING AND TECHNICAL SYSTEMS
996 BOARD ON TELECOMMUNICATIONS -- COMPUTER APPLICATIONS
997
998 Chairman
999
1000 DANIEL J. FINK, President, D.J. Fink Associates, Inc., Arlington,
1001 Virginia
1002
1003 Past Chairman
1004
1005 BROCKWAY MCMILLAN, Vice President (Retired), Bell Laboratories,
1006 Sedgwick, Maine
1007
1008 Members
1009
1010 ARTHUR G. ANDERSON, Vice President (Retired), IBM Corporation, San
1011 Jose, California
1012
1013 DANIEL BELL, Henry Ford II Professor of Social Sciences,
1014 Department of Sociology, Harvard University, Cambridge,
1015 Massachusetts
1016
1017 HERBERT D. BENINGTON, Technical Director, System Development
1018 Corporation, McLean, Virginia
1019
1020 ELWYN R. BERLEKAMP, Professor of Mathematics, Department of
1021 Mathematics, University of California, Berkeley, California
1022
1023 ANTHONY J. DEMARIA, Assistant Director of Research for Electronics
1024 and Electro-Optics Technology, United Technologies Research
1025 Center, East Hartford, Connecticut
1026
1027 GERALD P. DINNEEN, Vice President, Science and Technology,
1028 Honeywell Incorporated, Minneapolis, Minnesota
1029
1030 GEORGE GERBNER, Professor and Dean, The Annenberg School of
1031 Communications, University of Pennsylvania, Philadelphia,
1032 Pennsylvania
1033
1034 ANNE P. JONES, Partner, Sutherland, Asbill and Brennan,
1035 Washington, D.C.
1036
1037 ADRIAN M. MCDONOUGH, Professor of Management and Decision Sciences
1038 (Retired), The Wharton School, University of Pennsylvania,
1039 Havertown, Pennsylvania
1040
1041
1042
1043
1044 <span class="grey">National Research Council [Page 19]</span></pre>
1045 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
1046 <span class="grey"><a href="./rfc939">RFC 939</a> February 1985</span>
1047 Executive Summary of the NRC Report Transport on Protocols
1048
1049
1050 WILBUR L. PRITCHARD, President, Satellite Systems Engineering,
1051 Inc., Bethesda, Maryland
1052
1053 MICHAEL B. PURSLEY, Professor of Electrical Engineering,
1054 University of Illinois, Urbana, Illinois
1055
1056 IVAN SELIN, Chairman of the Board, American Management Systems,
1057 Inc., Arlington, Virginia
1058
1059 MISCHA SCHWARTZ, Professor of Electrical Engineering and Computer
1060 Science, Columbia University, New York, New York
1061
1062 ERIC E. SUMNER, Vice President, Operations System and Network
1063 Planning, AT&amp;T Bell Laboratories, Holmdel, New Jersey
1064
1065 KEITH W. UNCAPHER, Executive Director, USC-Information Sciences
1066 Institute Associate Dean, School of Engineering, University of
1067 Southern California, Marina del Rey, California
1068
1069 JAMES L. FLANAGAN, (CETS LIAISON MEMBER), Head, Acoustics Research
1070 Department, AT&amp;T Bell Laboratories, Murray Hill, New Jersey
1071
1072 Staff
1073
1074 Richard B. Marsten, Executive Director
1075 Jerome D. Rosenberg, Senior Staff Officer
1076 Karen Laughlin, Administrative Coordinator
1077 Carmen A. Ruby, Administrative Assistant
1078 Lois A. Leak, Administrative Secretary
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099 National Research Council [Page 20]
1100 </pre>