Debian Codegraph @

rfc1017.html

raw
1 <pre>Network Working Group Barry M. Leiner
2 Request for Comments: 1017 RIACS
3 August 1987
4
5 <span class="h1">Network Requirements for Scientific Research</span>
6
7 Internet Task Force on Scientific Computing
8
9 STATUS OF THIS MEMO
10
11 This RFC identifies the requirements on communication networks for
12 supporting scientific research. It proposes some specific areas for
13 near term work, as well as some long term goals. This is an "idea"
14 paper and discussion is strongly encouraged. Distribution of this
15 memo is unlimited.
16
17 INTRODUCTION
18
19 Computer networks are critical to scientific research. They are
20 currently being used by portions of the scientific community to
21 support access to remote resources (such as supercomputers and data
22 at collaborator's sites) and collaborative work through such
23 facilities as electronic mail and shared databases. There is
24 considerable movement in the direction of providing these
25 capabilities to the broad scientific community in a unified manner,
26 as evidence by this workshop. In the future, these capabilities will
27 even be required in space, as the Space Station becomes a reality as
28 a scientific research resource.
29
30 The purpose of this paper is to identify the range of requirements
31 for networks that are to support scientific research. These
32 requirements include the basic connectivity provided by the links and
33 switches of the network through the basic network functions to the
34 user services that need to be provided to allow effective use of the
35 interconnected network. The paper has four sections. The first
36 section discusses the functions a user requires of a network. The
37 second section discusses the requirements for the underlying link and
38 node infrastructure while the third proposes a set of specifications
39 to achieve the functions on an end-to-end basis. The fourth section
40 discusses a number of network-oriented user services that are needed
41 in addition to the network itself. In each section, the discussion
42 is broken into two categories. The first addresses near term
43 requirements: those capabilities and functions that are needed today
44 and for which technology is available to perform the function. The
45 second category concerns long term goals: those capabilities for
46 which additional research is needed.
47
48 This RFC was produced by the IAB Task force a Scientific Computing,
49
50
51
52 <span class="grey">Leiner [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
55
56
57 which is chartered to investigate advanced networking requirements
58 that result from scientific applications. Work reported herein was
59 supported in part by Cooperative Agreement NCC 2-387 from the
60 National Aeronautics and Space Administration (NASA) to the
61 Universities Space Research Association (USRA).
62
63 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. NETWORK FUNCTIONS</span>
64
65 This section addresses the functions and capabilities that networks
66 and particularly internetworks should be expected to support in the
67 near term future.
68
69 Near Term Requirements
70
71 There are many functions that are currently available to subsets of
72 the user community. These functions should be made available to the
73 broad scientific community.
74
75 User/Resource Connectivity
76
77 Undoubtedly the first order of business in networking is to provide
78 interconnectivity of users and the resources they need. The goal in
79 the near term for internetworking should be to extend the
80 connectivity as widely as possible, i.e. to provide ubiquitous
81 connectivity among users and between users and resources. Note that
82 the existence of a network path between sites does not necessarily
83 imply interoperability between communities and or resources using
84 non-compatible protocol suites. However, a minimal set of functions
85 should be provided across the entire user community, independent of
86 the protocol suite being used. These typically include electronic
87 mail at a minimum, file transfer and remote login capabilities must
88 also be provided.
89
90 Home Usage
91
92 One condition that could enhance current scientific computing would
93 be to extend to the home the same level of network support that the
94 scientist has available in his office environment. As network access
95 becomes increasingly widespread, the extension to the home will allow
96 the user to continue his computing at home without dramatic changes
97 in his work habits, based on limited access.
98
99 Charging
100
101 The scientific user should not have to worry about the costs of data
102 communications any more than he worries about voice communications
103 (his office telephone), so that data communications becomes an
104 integral and low-cost part of our national infrastructure. This
105
106
107
108 <span class="grey">Leiner [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
111
112
113 implies that charges for network services must NOT be volume
114 sensitive and must NOT be charged back to the individual. Either of
115 these conditions forces the user to consider network resources as
116 scarce and therefore requiring his individual attention to conserve
117 them. Such attention to extraneous details not only detracts from
118 the research, but fundamentally impacts the use and benefit that
119 networking is intended to supply. This does not require that
120 networking usage is free. It should be either be low enough cost
121 that the individual does not have to be accountable for "normal"
122 usage or managed in such a manner that the individual does not have
123 to be concerned with it on a daily basis.
124
125 Applications
126
127 Most applications, in the near term, which must be supported in an
128 internetwork environment are essentially extensions of current ones.
129 Particularly:
130
131 Electronic Mail
132
133 Electronic mail will increase in value as the extended
134 interconnectivity provided by internetworking provides a much
135 greater reachability of users.
136
137 Multimedia Mail
138
139 An enhancement to text based mail which includes capabilities
140 such as figures, diagrams, graphs, and digitized voice.
141
142 Multimedia Conferencing
143
144 Network conferencing is communication among multiple people
145 simultaneously. Conferencing may or may not be done in "real
146 time", that is all participants may not be required to be on-
147 line at the same time. The multimedia supported may include
148 text, voice, video, graphics, and possibly other capabilities.
149
150 File Transfer
151
152 The ability to transfer data files.
153
154 Bulk Transfer
155
156 The ability to stream large quantities of data.
157
158 Interactive Remote Login
159
160 The ability to perform remote terminal connections to hosts.
161
162
163
164 <span class="grey">Leiner [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
167
168
169 Remote Job Entry
170
171 The ability to submit batch jobs for processing to remote hosts
172 and receive output.
173
174 Applications which need support in the near term but are NOT
175 extensions of currently supported applications include:
176
177 Remote Instrument Control
178
179 This normally presumes to have a human in the "control loop".
180 This condition relaxes the requirements on the (inter)network
181 somewhat as to response times and reliability. Timing would be
182 presumed to be commensurate with human reactions and
183 reliability would not be as stringent as that required for
184 completely automatic control.
185
186 Remote Data Acquisition
187
188 This supports the collection of experimental data where the
189 experiment is remotely located from the collection center.
190 This requirement can only be satisfied when the bandwidth,
191 reliability, and predictability of network response are
192 sufficient. This cannot be supported in the general sense
193 because of the enormous bandwidth, very high reliability,
194 and/or guaranteed short response time required for many
195 experiments.
196
197 These last two requirements are especially crucial when one considers
198 remote experimentation such as will be performed on the Space
199 Station.
200
201 Capabilities
202
203 The above applications could be best supported on a network with
204 infinite bandwidth, zero delay, and perfect reliability.
205 Unfortunately, even currently feasible approximations to these levels
206 of capabilities can be very expensive. Therefore, it can be expected
207 that compromises will be made for each capability and between them,
208 with different balances struck between different networks. Because
209 of this, the user must be given an opportunity to declare which
210 capability or capabilities is/are of most interest-most likely
211 through a "type-of-service" required declaration. Some examples of
212 possible trade-offs: File Transport Normally requires high
213 reliability primarily and high bandwidth secondarily. Delay is not as
214 important.
215
216
217
218
219
220 <span class="grey">Leiner [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
223
224
225 Bulk Transport
226
227 Some applications such as digitized video might require high
228 bandwidth as the most important capability. Depending on the
229 application, delay would be second, and reliability of lesser
230 importance. Image transfers of scientific data sometimes will
231 invert the latter two requirements.
232
233 Interactive Traffic
234
235 This normally requires low delay as a primary consideration.
236 Reliability may be secondary depending on the application.
237 Bandwidth would usually be of least importance.
238
239 Standards
240
241 The use of standards in networking is directed toward
242 interoperability and availability of commercial equipment. However,
243 as stated earlier, full interoperability across the entire
244 scientific community is probably not a reasonable goal for
245 internetworking in the near term because of the protocol mix now
246 present. That is not to say, though, that the use of standards
247 should not be pursued on the path to full user interoperability.
248 Standards, in the context of near term goal support, include:
249
250 Media Exchange Standards
251
252 Would allow the interchange of equations, graphics, images, and data
253 bases as well as text.
254
255 Commercially Available Standards
256
257 Plug compatible, commercially available standards will allow a degree
258 of interoperability prior to the widespread availability of the ISO
259 standard protocols.
260
261 Long Term Goals
262
263 In the future, the internetwork should be transparent communications
264 between users and resources, and provide the additional network
265 services required to make use of that communications. A user should
266 be able to access whatever resources are available just as if the
267 resource is in the office. The same high level of service should
268 exist independent of which network one happens to be on. In fact,
269 one should not even be able to tell that the network is there!
270
271 It is also important that people be able to work effectively while at
272 home or when traveling. Wherever one may happen to be, it should be
273
274
275
276 <span class="grey">Leiner [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
279
280
281 possible to "plug into" the internetwork and read mail, access files,
282 control remote instruments, and have the same kind of environment one
283 is used to at the office.
284
285 Services to locate required facilities and take advantage of them
286 must also be available on the network. These range from the basic
287 "white" and "yellow" pages, providing network locations (addresses)
288 for users and capabilities, through to distributed data bases and
289 computing facilities. Eventually, this conglomeration of computers,
290 workstations, networks, and other computing resources will become one
291 gigantic distributed "world computer" with a very large number of
292 processing nodes all over the world.
293
294 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. NETWORK CONNECTIVITY</span>
295
296 By network connectivity, we mean the ability to move packets from one
297 point to another.
298
299 Note that an implicit assumption in this paper is that packet
300 switched networks are the preferred technology for providing a
301 scientific computer network. This is due to the ability of such
302 networks to share the available link resources to provide
303 interconnection between numerous sites and their ability to
304 effectively handle the "bursty" computer communication requirement.
305
306 Note that this need not mean functional interoperability, since the
307 endpoints may be using incompatible protocols. Thus, in this
308 section, we will be addressing the use of shared links and
309 interconnected networks to provide a possible path. In the next
310 section, the exploitation of these paths to achieve functional
311 connectivity will be addressed.
312
313 In this section, we discuss the need for providing these network
314 paths to a wide set of users and resources, and the characteristics
315 of those paths. As in other sections, this discussion is broken into
316 two major categories. The first category are those goals which we
317 believe to be achievable with currently available technology and
318 implementations. The second category are those for which further
319 research is required.
320
321 Near Term Objectives
322
323 Currently, there are a large number of networks serving the
324 scientific community, including Arpanet, MFEnet, SPAN, NASnet, and
325 the NSFnet backbone. While there is some loose correlation between
326 the networks and the disciplines they serve, these networks are
327 organized more based on Federal funding. Furthermore, while there is
328 significant interconnectivity between a number of the networks, there
329
330
331
332 <span class="grey">Leiner [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
335
336
337 is considerable room for more sharing of these resources.
338
339 In the near term, therefore, there are two major requirement areas;
340 providing for connectivity based on discipline and user community,
341 and providing for the effective use of adequate networking resources.
342
343 Discipline Connectivity
344
345 Scientists in a particular community/discipline need to have access
346 to many common resources as well as communicate with each other. For
347 example, the quantum physics research community obtains funding from
348 a number of Federal sources, but carries out its research within the
349 context of a scientific discourse. Furthermore, this discourse often
350 overlaps several disciplines. Because networks are generally
351 oriented based on the source of funding, this required connectivity
352 has in the past been inhibited. NSFnet is a major step towards
353 satisfying this requirement, because of its underlying philosophy of
354 acting as an interconnectivity network between supercomputer centers
355 and between state, regional, and therefore campus networks. This
356 move towards a set of networks that are interconnected, at least at
357 the packet transport level, must be continued so that a scientist can
358 obtain connectivity between his/her local computing equipment and the
359 computing and other resources that are needed, independently of the
360 source of funds.
361
362 Obviously, actual use of those resources will depend on obtaining
363 access permission from the appropriate controlling organization. For
364 example, use of a supercomputer will require permission and some
365 allocation of computing resources. The lack of network access should
366 not, however, be the limiting factor for resource utilization.
367
368 Communication Resource Sharing
369
370 The scientific community is always going to suffer from a lack of
371 adequate communication bandwidth and connections. There are
372 requirements (e.g. graphic animation from supercomputers) that
373 stretch the capabilities of even the most advanced long-haul
374 networks. In addition, as more and more scientists require
375 connection into networks, the ability to provide those connections on
376 a network-centric basis will become more and more difficult.
377
378 However, the communication links (e.g. leased lines and satellite
379 channels) providing the underlying topology of the various networks
380 span in aggregate a very broad range of the scientific community
381 sites. If, therefore, the networks could share these links in an
382 effective manner, two objectives could be achieved:
383
384 The need to add links just to support a particular network
385
386
387
388 <span class="grey">Leiner [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
391
392
393 topology change would be decreased, and
394
395 New user sites could be connected more readily.
396
397 Existing technology (namely the DARPA-developed gateway system based
398 on the Internet Protocol, IP) provides an effective method for
399 accomplishing this sharing. By using IP gateways to connect the
400 various networks, and by arranging for suitable cost-sharing, the
401 underlying connectivity would be greatly expanded and both of the
402 above objectives achieved.
403
404 Expansion of Physical Structure
405
406 Unfortunately, the mere interconnectivity of the various networks
407 does not increase the bandwidth available. While it may allow for
408 more effective use of that available bandwidth, a sufficient number
409 of links with adequate bandwidth must be provided to avoid network
410 congestion. This problem has already occurred in the Arpanet, where
411 the expansion of the use of the network without a concurrent
412 expansion in the trunking and topology has resulted in congestion and
413 consequent degradation in performance.
414
415 Thus, it is necessary to augment the current physical structure
416 (links and switches) both by increasing the bandwidth of the current
417 configuration and by adding additional links and switches where
418 appropriate.
419
420 Network Engineering
421
422 One of the major deficiencies in the current system of networks is
423 the lack of overall engineering. While each of the various networks
424 generally is well supported, there is woefully little engineering of
425 the overall system. As the networks are interconnected into a larger
426 system, this need will become more severe. Examples of the areas
427 where engineering is needed are:
428
429 Topology engineering-deciding where links and switches should be
430 installed or upgraded. If the interconnection of the networks is
431 achieved, this will often involve a decision as to which networks
432 need to be upgraded as well as deciding where in the network those
433 upgrades should take place.
434
435 Connection Engineering-when a user site desires to be connected,
436 deciding which node of which network is the best for that site,
437 considering such issues as existing node locations, available
438 bandwidth, and expected traffic patterns to/from that site.
439
440 Operations and Maintenance-monitoring the operation of the overall
441
442
443
444 <span class="grey">Leiner [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
447
448
449 system and identifying corrective actions when failures occur.
450
451 Support of Different Types of Service
452
453 Several different end user applications are currently in place, and
454 these put different demands on the underlying structure. For
455 example, interactive remote login requires low delay, while file
456 transfer requires high bandwidth. It is important in the
457 installation of additional links and switches that care be given to
458 providing a mix of link characteristics. For example, high bandwidth
459 satellite channels may be appropriate to support broadcast
460 applications or graphics, while low delay will be required to support
461 interactive applications.
462
463 Future Goals
464
465 Significant expansion of the underlying transport mechanisms will be
466 required to support future scientific networking. These expansions
467 will be both in size and performance.
468
469 Bandwidth
470
471 Bandwidth requirements are being driven higher by advances in
472 computer technology as well as the proliferation of that technology.
473 As high performance graphics workstations work cooperatively with
474 supercomputers, and as real-time remote robotics and experimental
475 control become a reality, the bandwidth requirements will continue to
476 grow. In addition, as the number of sites on the networks increase,
477 so will the aggregate bandwidth requirement. However, at the same
478 time, the underlying bandwidth capabilities are also increasing.
479 Satellite bandwidths of tens of megabits are available, and fiber
480 optics technologies are providing extremely high bandwidths (in the
481 range of gigabits). It is therefore essential that the underlying
482 connectivity take advantage of these advances in communications to
483 increase the available end-to-end bandwidth.
484
485 Expressway Routing
486
487 As higher levels of internet connectivity occur there will be a new
488 set of problems related to lowest hop count and lowest delay routing
489 metrics. The assumed internet connectivity can easily present
490 situations where the highest speed, lowest delay route between two
491 nodes on the same net is via a route on another network. Consider
492 two sites one either end of the country, but both on the same
493 multipoint internet, where their network also is gatewayed to some
494 other network with high speed transcontinental links. The routing
495 algorithms must be able to handle these situations gracefully, and
496 they become of increased importance in handling global type-of-
497
498
499
500 <span class="grey">Leiner [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
503
504
505 service routing.
506
507 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. NETWORK SPECIFICATIONS</span>
508
509 To achieve the end-to-end user functions discussed in <a href="#section-2">section 2</a>, it
510 is not adequate to simply provide the underlying connectivity
511 described in the previous section. The network must provide a
512 certain set of capabilities on an end-to-end basis. In this
513 section, we discuss the specifications on the network that are
514 required.
515
516 Near Term Specifications
517
518 In the near term, the requirements on the networks are two-fold.
519 First is to provide those functions that will permit full
520 interoperability, and second the internetwork must address the
521 additional requirements that arise in the connection of networks,
522 users, and resources.
523
524 Interoperability
525
526 A first-order requirement for scientific computer networks (and
527 computer networks in general) is that they be interoperable with each
528 other, as discussed in the above section on connectivity. A first
529 step to accomplish this is to use IP. The use of IP will allow
530 individual networks built by differing agencies to combine resources
531 and minimize cost by avoiding the needless duplication of network
532 resources and their management. However, use of IP does not provide
533 end-to-end interoperability. There must also be compatibility of
534 higher level functions and protocols. At a minimum, while commonly
535 agreed upon standards (such as the ISO developments) are proceeding,
536 methods for interoperability between different protocol suites must
537 be developed. This would provide interoperability of certain
538 functions, such as file transfer, electronic mail and remote login.
539 The emphasis, however, should be on developing agreement within the
540 scientific community on use of a standard set of protocols.
541
542 Access Control
543
544 The design of the network should include adequate methods for
545 controlling access to the network by unauthorized personnel. This
546 especially includes access to network capabilities that are reachable
547 via the commercial phone network and public data nets. For example,
548 terminal servers that allow users to dial up via commercial phone
549 lines should have adequate authentication mechanisms in place to
550 prevent access by unauthorized individuals. However, it should be
551 noted that most hosts that are reachable via such networks are also
552 reachable via other "non-network" means, such as directly dialing
553
554
555
556 <span class="grey">Leiner [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
559
560
561 over commercial phone lines. The purpose of network access control
562 is not to insure isolation of hosts from unauthorized users, and
563 hosts should not expect the network itself to protect them from
564 "hackers".
565
566 Privacy
567
568 The network should provide protection of data that traverses it in a
569 way that is commensurate with the sensitivity of that data. It is
570 judged that the scientific requirements for privacy of data traveling
571 on networks does not warrant a large expenditure of resources in this
572 area. However, nothing in the network design should preclude the use
573 of link level or end-to-end encryption, or other such methods that
574 can be added at a later time. An example of this kind of capability
575 would be use of KG-84A link encryptors on MILNET or the Fig Leaf
576 DES-based end-to-end encryption box developed by DARPA.
577
578 Accounting
579
580 The network should provide adequate accounting procedures to track
581 the consumption of network resources. Accounting of network
582 resources is also important for the management of the network, and
583 particularly the management of interconnections with other networks.
584 Proper use of the accounting database should allow network management
585 personnel to determine the "flows" of data on the network, and the
586 identification of bottlenecks in network resources. This capability
587 also has secondary value in tracking down intrusions of the network,
588 and to provide an audit trail if malicious abuse should occur. In
589 addition, accounting of higher level network services (such as
590 terminal serving) should be kept track of for the same reasons.
591
592 Type of Service Routing
593
594 Type of service routing is necessary since not all elements of
595 network activity require the same resources, and the opportunities
596 for minimizing use of costly network resources are large. For
597 example, interactive traffic such as remote login requires low delay
598 so the network will not be a bottleneck to the user attempting to do
599 work. Yet the bandwidth of interactive traffic can be quite small
600 compared to the requirements for file transfer and mail service which
601 are not response time critical. Without type of service routing,
602 network resources must sized according to the largest user, and have
603 characteristics that are pleasing to the most finicky user. This has
604 major cost implications for the network design, as high-delay links,
605 such as satellite links, cannot be used for interactive traffic
606 despite the significant cost savings they represent over terrestrial
607 links. With type of service routing in place in the network
608 gateways, and proper software in the hosts to make use of such
609
610
611
612 <span class="grey">Leiner [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
615
616
617 capabilities, overall network performance can be enhanced, and
618 sizable cost savings realized. Since the IP protocol already has
619 provisions for such routing, such changes to existing implementations
620 does not require a major change in the underlying protocol
621 implementations.
622
623 Administration of Address Space
624
625 Local administration of network address space is essential to provide
626 for prompt addition of hosts to the network, and to minimize the load
627 on backbone network administrators. Further, a distributed name to
628 address translation service also has similar advantages. The DARPA
629 Name Domain system currently in use on the Internet is a suitable
630 implementation of such a name to address translation system.
631
632 Remote Procedure Call Libraries
633
634 In order to provide a standard library interface so that distributed
635 network utilities can easily communicate with each other in a
636 standard way, a standard Remote Procedure Call (RPC) library must be
637 deployed. The computer industry has lead the research community in
638 developing RPC implementations, and current implementations tend to
639 be compatible within the same type of operating system, but not
640 across operating systems. Nonetheless, a portable RPC implementation
641 that can be standardized can provide a substantial boost in present
642 capability to write operating system independent network utilities.
643 If a new RPC mechanism is to be designed from scratch, then it must
644 have enough capabilities to lure implementors away from current
645 standards. Otherwise, modification of an existing standard that is
646 close to the mark in capabilities seems to be in order, with the
647 cooperation of vendors in the field to assure implementations will
648 exist for all major operating systems in use on the network.
649
650 Remote Job Entry (RJE)
651
652 The capabilities of standard network RJE implementations are
653 inadequate, and are implemented prolifically among major operating
654 systems. While the notion of RJE evokes memories of dated
655 technologies such as punch cards, the concept is still valid, and is
656 favored as a means of interaction with supercomputers by science
657 users. All major supercomputer manufacturers support RJE access in
658 their operating systems, but many do not generalize well into the
659 Internet domain. That is, a RJE standard that is designed for 2400
660 baud modem access from a card reader may not be easily modifiable for
661 use on the Internet. Nonetheless, the capability for a network user
662 to submit a job from a host and have its output delivered on a
663 printer attached to a different host would be welcomed by most
664 science users. Further, having this capability interoperate with
665
666
667
668 <span class="grey">Leiner [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
671
672
673 existing RJE packages would add a large amount of flexibility to the
674 whole system.
675
676 Multiple Virtual Connections
677
678 The capability to have multiple network connections open from a
679 user's workstation to remote network hosts is an invaluable tool that
680 greatly increases user productivity. The network design should not
681 place limits (procedural or otherwise) on this capability.
682
683 Network Operation and Management Tools
684
685 The present state of internet technology requires the use of
686 personnel who are, in the vernacular of the trade, called network
687 "wizards," for the proper operation and management of networks.
688 These people are a scarce resource to begin with, and squandering
689 them on day to day operational issues detracts from progress in the
690 more developmental areas of networking. The cause of this problem is
691 that a good part of the knowledge for operating and managing a
692 network has never been written down in any sort of concise fashion,
693 and the reason for that is because networks of this type in the past
694 were primarily used as a research tool, not as an operational
695 resource. While the usage of these networks has changed, the
696 technology has not adjusted to the new reality that a wizard may not
697 be nearby when a problem arises. To insure that the network can
698 flexibly expand in the future, new tools must be developed that allow
699 non-wizards to monitor network performance, determine trouble spots,
700 and implement repairs or 'work-arounds'.
701
702 Future Goals
703
704 The networks of the future must be able to support transparent access
705 to distributed resources of a variety of different kinds. These
706 resources will include supercomputer facilities, remote observing
707 facilities, distributed archives and databases, and other network
708 services. Access to these resources is to be made widely available
709 to scientists, other researchers, and support personnel located at
710 remote sites over a variety of internetted connections. Different
711 modes of access must be supported that are consonant with the sorts
712 of resources that are being accessed, the data bandwidths required
713 and the type of interaction demanded by the application.
714
715 Network protocol enhancements will be required to support this
716 expansion in functionality; mere increases in bandwidth are not
717 sufficient. The number of end nodes to be connected is in the
718 hundreds of thousands, driven by increasing use of microprocessors
719 and workstations throughout the community. Fundamentally different
720 sorts of services from those now offered are anticipated, and dynamic
721
722
723
724 <span class="grey">Leiner [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
727
728
729 bandwidth selection and allocation will be required to support the
730 different access modes. Large-scale internet connections among
731 several agency size internets will require new approaches to routing
732 and naming paradigms. All of this must be planned so as to
733 facilitate transition to the ISO/OSI standards as these mature and
734 robust implementations are placed in service and tuned for
735 performance.
736
737 Several specific areas are identified as being of critical importance
738 in support of future network requirements, listed in no particular
739 order:
740
741 Standards and Interface Abstractions
742
743 As more and different services are made available on these
744 various networks it will become increasingly important to
745 identify interface standards and suitable application
746 abstractions to support remote resource access. These
747 abstractions may be applicable at several levels in the
748 protocol hierarchy and can serve to enhance both applications
749 functionality and portability. Examples are transport or
750 connection layer abstractions that support applications
751 independence from lower level network realizations or interface
752 abstractions that provide a data description language that can
753 handle a full range of abstract data type definitions.
754 Applications or connection level abstractions can provide means
755 of bridging across different protocol suites as well as helping
756 with protocol transition.
757
758 OSI Transition and Enhancements
759
760 Further evolution of the OSI network protocols and realization
761 of large-scale networks so that some of the real protocol and
762 tuning issues can be dealt with must be anticipated. It is
763 only when such networks have been created that these issues can
764 be approached and resolved. Type-of-service and Expressway
765 routing and related routing issues must be resolved before a
766 real transition can be contemplated. Using the interface
767 abstraction approach just described will allow definition now
768 of applications that can transition as the lower layer networks
769 are implemented. Applications gateways and relay functions
770 will be a part of this transition strategy, along with dual
771 mode gateways and protocol translation layers.
772
773 Processor Count Expansion
774
775 Increases in the numbers of nodes and host sites and the
776 expected growth in use of micro-computers, super-micro
777
778
779
780 <span class="grey">Leiner [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
783
784
785 workstations, and other modest cost but high power computing
786 solutions will drive the development of different network and
787 interconnect strategies as well as the infrastructure for
788 managing this increased name space. Hierarchical name
789 management (as in domain based naming) and suitable transport
790 layer realizations will be required to build networks that are
791 robust and functional in the face of the anticipated
792 expansions.
793
794 Dynamic Binding of Names to Addresses
795
796 Increased processor counts and increased usage of portable
797 units, mobile units and lap-top micros will make dynamic
798 management of the name/address space a must. Units must have
799 fixed designations that can be re-bound to physical addresses
800 as required or expedient.
801
802 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. USER SERVICES</span>
803
804 The user services of the network are a key aspect of making the
805 network directly useful to the scientist. Without the right user
806 services, network users separate into artificial subclasses based on
807 their degree of sophistication in acquiring skill in the use of the
808 network. Flexible information dissemination equalizes the
809 effectiveness of the network for different kinds of users.
810
811 Near Term Requirements
812
813 In the near term, the focus is on providing the services that allow
814 users to take advantage of the functions that the interconnected
815 network provides.
816
817 Directory services
818
819 Much of the information necessary in the use of the network is for
820 directory purposes. The user needs to access resources available on
821 the network, and needs to obtain a name or address.
822
823 White Pages
824
825 The network needs to provide mechanisms for looking up names and
826 addresses of people and hosts on the network. Flexible searches
827 should be possible on multiple aspects of the directory listing.
828 Some of these services are normally transparent to the user/host name
829 to address translation for example.
830
831
832
833
834
835
836 <span class="grey">Leiner [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
839
840
841 Yellow Pages
842
843 Other kinds of information lookup are based on cataloging and
844 classification of information about resources on the networks.
845
846 Information Sharing Services
847
848 Bulletin Boards
849
850 The service of the electronic bulletin board is the one-to-many
851 analog of the one-to-one service of electronic mail. A
852 bulletin board provides a forum for discussion and interchange
853 of information. Accessibility is network-wide depending on the
854 definition of the particular bulletin board. Currently the
855 SMTP and UUCP protocols are used in the transport of postings
856 for many bulletin boards, but any similar electronic mail
857 transport can be substituted without affecting the underlying
858 concept. An effectively open-ended recipient list is specified
859 as the recipient of a message, which then constitutes a
860 bulletin board posting. A convention exists as to what
861 transport protocols are utilized for a particular set of
862 bulletin boards. The user agent used to access the Bulletin
863 Board may vary from host to host. Some number of host
864 resources on the network provide the service of progressively
865 expanding the symbolic mail address of the Bulletin Board into
866 its constituent parts, as well as relaying postings as a
867 service to the network. Associated with this service is the
868 maintenance of the lists used in distributing the postings.
869 This maintenance includes responding to requests from Bulletin
870 Board readers and host Bulletin Board managers, as well as
871 drawing the appropriate conclusions from recurring
872 automatically generated or error messages in response to
873 distribution attempts.
874
875 Community Archiving
876
877 Much information can be shared over the network. At some point
878 each particular information item reaches the stage where it is
879 no longer appropriately kept online and accessible. When
880 moving a file of information to offline storage, a network can
881 provide its hosts a considerable economy if information of
882 interest to several of them need only be stored offline once.
883 Procedures then exist for querying and retrieving from the set
884 of offline stored files.
885
886 Shared/distributed file system
887
888 It should be possible for a user on the network to look at a
889
890
891
892 <span class="grey">Leiner [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
895
896
897 broadly defined collection of information on the network as one
898 useful whole. To this end, standards for accessing files
899 remotely are necessary. These standards should include means
900 for random access to remote files, similar to the generally
901 employed on a single computer system.
902
903 Distributed Databases and Archives
904
905 As more scientific disciplines computerize their data archives
906 and catalogs, mechanisms will have to be provided to support
907 distributed access to these resources. Fundamentally new kins
908 of collaborative research will become possible when such
909 resources and access mechanisms are widely available.
910
911 Resource Sharing Services
912
913 In sharing the resources or services available on the network,
914 certain ancillary services are needed depending on the
915 resource.
916
917 Access Control
918
919 Identification and authorization is needed for individuals, hosts or
920 subnetworks permitted to make use of a resource available via the
921 network. There should be consistency of procedure for obtaining and
922 utilizing permission for use of shared resources. The identification
923 scheme used for access to the network should be available for use by
924 resources as well. In some cases, this will serve as sufficient
925 access control, and in other cases it will be a useful adjunct to
926 resource-specific controls. The information on the current network
927 location of the user should be available along with information on
928 user identification to permit added flexibility for resources. For
929 example, it should be possible to verify that an access attempt is
930 coming from within a state. A state agency might then grant public
931 access to its services only for users within the state. Attributes
932 of individuals should be codifiable within the access control
933 database, for example membership in a given professional society.
934
935 Privacy
936
937 Users of a resource have a right to expect that they have control
938 over the release of the information they generate. Resources should
939 allow classifying information according to degree of access, i.e.
940 none, access to read, access according to criteria specified in the
941 data itself, ability to change or add information. The full range of
942 identification information described under access control should be
943 available to the user when specifying access. Access could be
944 granted to all fellow members of a professional society, for example.
945
946
947
948 <span class="grey">Leiner [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
951
952
953 Accounting
954
955 To permit auditing of usage, accounting information should be
956 provided for those resources for which it is deemed necessary. This
957 would include identity of the user of the resource and the
958 corresponding volume of resource components.
959
960 Legalities of Interagency Research Internet
961
962 To make the multiply-sponsored internetwork feasible, the federal
963 budget will have to recognize that some usage outside a particular
964 budget category may occur. This will permit the cross-utilization of
965 agency funded resources. For example, NSFnet researchers would be
966 able to access supercomputers over NASnet. In return for this, the
967 total cost to the government will be significantly reduced because of
968 the benefits of sharing network and other resources, rather than
969 duplicating them.
970
971 Standards
972
973 In order for the networking needs of scientific computing to be met,
974 new standards are going to evolve. It is important that they be
975 tested under actual use conditions, and that feedback be used to
976 refine them. Since the standards for scientific communication and
977 networking are to be experimented with, they are more dynamic than
978 those in other electronic communication fields. It is critical that
979 the resources of the network be expended to promulgate experimental
980 standards and maximize the range of the community utilizing them. To
981 this end, the sharing of results of the testing is important.
982
983 User-oriented Documentation
984
985 The functionality of the network should be available widely without
986 the costly need to refer requests to experts for formulation. A
987 basic information facility in the network should therefore be
988 developed. The network should be self-documenting via online help
989 files, interactive tutorials, and good design. In addition, concise,
990 well-indexed and complete printed documentation should be available.
991
992 Future Goals
993
994 The goal for the future should be to provide the advanced user
995 services that allow full advantage to be taken of the interconnection
996 of users, computing resources, data bases, and experimental
997 facilities. One major goal would be the creation of a national
998 knowledge bank. Such a knowledge bank would capture and organize
999 computer-based knowledge in various scientific fields that is
1000 currently available only in written/printed form, or in the minds of
1001
1002
1003
1004 <span class="grey">Leiner [Page 18]</span></pre>
1005 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
1006 <span class="grey"><a href="./rfc1017">RFC 1017</a> Requirements for Scientific Research August 1987</span>
1007
1008
1009 experts or experienced workers in the field. This knowledge would be
1010 stored in knowledge banks which will be accessible over the network
1011 to individual researchers and their programs. The result will be a
1012 codification of scientific understanding and technical know-how in a
1013 series of knowledge based systems which would become increasingly
1014 capable over time.
1015
1016 CONCLUSION
1017
1018 In this paper, we have tried to describe the functions required of
1019 the interconnected national network to support scientific research.
1020 These functions range from basic connectivity through to the
1021 provision for powerful distributed user services.
1022
1023 Many of the goals described in this paper are achievable with current
1024 technology. They require coordination of the various networking
1025 activities, agreement to share costs and technologies, and agreement
1026 to use common protocols and standards in the provision of those
1027 functions. Other goals require further research, where the
1028 coordination of the efforts and sharing of results will be key to
1029 making those results available to the scientific user.
1030
1031 For these reasons, we welcome the initiative represented by this
1032 workshop to have the government agencies join forces in providing the
1033 best network facilities possible in support of scientific research.
1034
1035 APPENDIX
1036
1037 Internet Task Force on Scientific Computing
1038
1039
1040 Rick Adrion University of Massachusetts
1041 Ron Bailey NASA Ames Research Center
1042 Rick Bogart Stanford University
1043 Bob Brown RIACS
1044 Dave Farber University of Delaware
1045 Alan Katz USC Information Science Institute
1046 Jim Leighton Lawrence Livermore Laboratories
1047 Keith Lantz Stanford University
1048 Barry Leiner (chair) RIACS
1049 Milo Medin NASA Ames Research Center
1050 Mike Muuss US Army Ballistics Research Laboratory
1051 Harvey Newman California Institute of Technology
1052 David Roode Intellicorp
1053 Ari Ollikainen General Electric
1054 Peter Shames Space Telescope Science Institute
1055 Phil Scherrer Stanford University
1056
1057
1058
1059
1060 Leiner [Page 19]
1061 </pre>