Debian Codegraph @

rfc1291.html

raw
1 <pre>Network Working Group V. Aggarwal
2 Request for Comments: 1291 JvNCnet Computer Network
3 December 1991
4
5
6 <span class="h1">Mid-Level Networks</span>
7 <span class="h1">Potential Technical Services</span>
8
9 Status of this Memo
10
11 This RFC provides information for the Internet community. It does not
12 specify an Internet standard. Distribution of this memo is unlimited.
13
14 Abstract
15
16 This document proposes a set of technical services that each Internet
17 mid-level network can offer within the mid-level network itself and
18 and to its peer networks. The term "mid-level" is used as a generic
19 term to represent all regional and similar networks, which, due to
20 continuous evolutions and transitions, can no longer be termed
21 "regional" [<a href="#ref-MAN" title="&quot;The Strategic Future of the Mid-Level Networks&quot;">MAN</a>]. It discusses the pros and cons of offering these
22 services, as well as areas in which mid-level networks can work
23 together.
24
25 A large portion of the ideas stem from discussions at the IETF
26 Operational Statistics (OPstat), User Connectivity Problems (UCP) and
27 Network Joint Management (NJM) working groups.
28
29 Table of Contents
30
31 <a href="#section-1">1</a>. Introduction.................................................. <a href="#page-2">2</a>
32 <a href="#section-2">2</a>. The Generic Model............................................. <a href="#page-2">2</a>
33 <a href="#section-3">3</a>. Technical Services............................................ <a href="#page-3">3</a>
34 <a href="#section-3.1">3.1</a> Domain Name Service......................................... <a href="#page-3">3</a>
35 <a href="#section-3.2">3.2</a> Public Domain Software...................................... <a href="#page-4">4</a>
36 <a href="#section-3.3">3.3</a> Network Time................................................ <a href="#page-5">5</a>
37 <a href="#section-3.4">3.4</a> Network News................................................ <a href="#page-5">5</a>
38 <a href="#section-3.5">3.5</a> Mailing Lists............................................... <a href="#page-6">6</a>
39 <a href="#section-4">4</a>. Experimental Testbeds......................................... <a href="#page-6">6</a>
40 <a href="#section-5">5</a>. Network Information Services.................................. <a href="#page-7">7</a>
41 <a href="#section-6">6</a>. Network Operations............................................ <a href="#page-7">7</a>
42 <a href="#section-7">7</a>. References.................................................... <a href="#page-8">8</a>
43 <a href="#section-8">8</a>. Security Considerations....................................... <a href="#page-9">9</a>
44 <a href="#section-9">9</a>. Author's Address.............................................. <a href="#page-9">9</a>
45 <a href="#appendix-A">Appendix A</a> Mailing Lists......................................... <a href="#page-10">10</a>
46 <a href="#appendix-B">Appendix B</a> DNS Architecture Strategy............................. <a href="#page-10">10</a>
47
48
49
50
51
52 <span class="grey">Aggarwal [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
55
56
57 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Introduction</span>
58
59 Over the past few years, the Internet has grown to be a very large
60 entity and its dependability is critical to its users. Furthermore,
61 due to the size and nature of the network, the trend has been to
62 decentralize as many network functions (such as domain name-service,
63 whois, etc.) as possible. Efforts are being made in resource
64 discovery [<a href="#ref-SHHH90" title="&quot;Supporting Resource Discovery Among Public Internet Archives&quot;">SHHH90</a>] so that the work of researchers is not lost in the
65 volumes of data that is available on the Internet.
66
67 A side result of this growth has been the logical structure imposed
68 in the Internet of networks classified by function. Tangible examples
69 in the present state are the NSFnet national backbone, the mid-
70 level/regional networks and campus networks. Each of these can be
71 viewed as hierarchies within an organization, each serving a slightly
72 different function than the other (campus LANs providing access to
73 local resources, mid-level networks providing access to remote
74 resources, etc.). The functions of each hierarchy then become the
75 "services" offered to the organizational layer below it, who in turn
76 depend on these services.
77
78 This document proposes a set of basic technical services that could
79 be offered by a mid-level network. These services would not only
80 increase the robustness of the mid-level network itself, but would
81 also serve to structure the distribution of resources and services
82 within the Internet. It also proposes a uniform naming convention for
83 locating the hosts offering these services.
84
85 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. The Generic Model</span>
86
87 The Internet model that is used as the basis for this document is a
88 graph of mid-level networks connected to one another, each in turn
89 connecting the campus/organization networks and with the end users
90 attached to the campus networks. The model assumes that the mid-level
91 networks constitute the highest level of functional division within
92 the Internet hierarchy described above (this could change in the
93 unforeseen future). With this model in perspective, this document
94 addresses the objectives of minimizing unnecessary traffic within the
95 Internet as well as making the entire structure as robust as
96 possible.
97
98 The proposed structure is a derived extension of organizational LANs
99 where certain services are offered within the organizational LAN
100 itself, such as nameservice, mail, shared files, single or
101 hierarchical points of contact for problems, etc.
102
103 The following are the services that are discussed as possible
104 functions of a mid-level network:
105
106
107
108 <span class="grey">Aggarwal [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
111
112
113 o Technical services
114
115 o Experimental sites for testing and dissemination of new
116 software and technology to end sites on the network
117
118 In addition, the following services are mentioned briefly which are
119 discussed in detail elsewhere [<a href="#ref-SSM91" title="&quot;Building a Network Information Services Infrastructure&quot;">SSM91</a>, <a href="#ref-ML91" title="&quot;User Connectivity Problems Working Group&quot;">ML91</a>]:
120
121 o Network Operation services and the interaction between
122 different mid-level networks in this area
123
124 o Network Information services
125
126 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Technical Services</span>
127
128 The Internet has grown to be an essential entity because of the
129 services that it offers to its end users. The list of services is
130 long and growing, but some services are more widely used and deployed
131 than others. This section attempts to list and discuss those
132 technical services that could help a mid-level network provide robust
133 and improved services to its end sites.
134
135 <span class="h3"><a class="selflink" id="section-3.1" href="#section-3.1">3.1</a> Domain Name Service</span>
136
137 According to the NSFnet traffic statistics collected for May 1991,
138 about 7% of the packets on the NSFnet backbone were domain nameserver
139 (DNS) packets. This is a significant amount of traffic, and since
140 most of the other network applications depend on this service, a
141 robust DNS service is critical to any Internet site.
142
143 Proper location of secondary nameservers so that they are located on
144 different physical networks can increase the reliability of this
145 service to a large extent [<a href="#ref-MOC87a" title="&quot;Domain Names - Implementation and Specification&quot;">MOC87a</a>, <a href="#ref-MOC87b" title="&quot;Domain Names - Concepts and Facilities&quot;">MOC87b</a>]. However, the nature of
146 the service requires that the nameservers for the next highest level
147 be available in order to resolve names outline-mode side of one's
148 domain. Thus, for "foo.princeton.edu" to resolve "a.mid.net", the
149 root nameservers which point to mid.net's nameservers have to be
150 reachable.
151
152 To make the service more reliable, the mid-level network could have
153 at least one nameserver that is able to resolve nameserver queries
154 for all domains directly connected to it. Thus, in the event that the
155 entire mid-level network becomes isolated from the rest of the
156 Internet, applications can still resolve queries for sites directly
157 connected to the mid-level network. Without this functionality, there
158 is no way of resolving a name if the root (or higher level)
159 nameservers become unreachable, even if the query is for a site that
160 is directly connected and reachable.
161
162
163
164 <span class="grey">Aggarwal [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
167
168
169 Strategies for implementing this architecture are discussed in
170 <a href="#appendix-B">appendix B</a>.
171
172 To locate such a "meta-domain" server within a mid-level network, it
173 is proposed that a nameserver entry for "meta-dns" exist within the
174 mid-level network's domain.
175
176 <span class="h3"><a class="selflink" id="section-3.2" href="#section-3.2">3.2</a> Public Domain Software</span>
177
178 File transfer traffic constituted 23% of the NSFnet backbone traffic
179 for May 1991. Public shareware is a very valuable resource within the
180 Internet and a considerable amount of effort is being put into
181 developing applications to track all available resources in the
182 public archives [<a href="#ref-SHHH90" title="&quot;Supporting Resource Discovery Among Public Internet Archives&quot;">SHHH90</a>].
183
184 It would be difficult, if not impossible to create an up-to-date
185 repository for every public domain package available on the Internet,
186 simply because of the volume of software and the rate at which new
187 software is being developed every day. Some hosts have gained
188 popularity as good public archives (such as uunet.uu.net, sumex-
189 aim.stanford.edu, wuarchive.wustl.edu) and new developers tend to
190 distribute the software to these sites as distribution points. The
191 economics of maintaining centralized archives is another deterrent to
192 centralization (the UUnet archives at uunet.uu.net take up roughly
193 1GB of disk storage).
194
195 Recently however, a number of methods for resource discovery have
196 been developed and are available on the Internet ("ftp-list" file
197 compiled by John Granose - [email protected], Archie at
198 archie.cs.mcgill.ca and Prospero [<a href="#ref-NEU" title="&quot;The Virtual System Model: A Scalable Approach to Organizing Large Systems&quot;">NEU</a>]).
199
200 It is desirable that the mid-level networks be able to provide up-
201 to-date pointers to the distribution hosts for available public
202 software archives. Coordinating the distribution of a static list is
203 difficult (though not impossible) and the use of automated resource
204 discovery mechanisms such as Archie and Prospero is recommended.
205 Under ideal conditions, any software that is popular and significant
206 (e.g., X11, TeX, RFC's) could be archived and distributed within the
207 mid-level network, but measuring "popularity" and "significance" are
208 debatable and left for further evaluation. Furthermore, a nameserver
209 entry for host "swdist" within the domain can provide information on
210 the various available alternatives for software distribution and
211 discovery (static file location, pointers to Archie servers, etc.) --
212 this nameserver entry can be an alias for a CNAME or a TXT entry.
213
214
215
216
217
218
219
220 <span class="grey">Aggarwal [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
223
224
225 <span class="h3"><a class="selflink" id="section-3.3" href="#section-3.3">3.3</a> Network Time</span>
226
227 An important feature of any computer network providing distributed
228 services is the capability to synchronize the local clocks on the
229 various systems in the network. Ideally, the clocks of all the
230 reference sources would be synchronized to national standards by wire
231 or radio. The importance and immense popularity of this service makes
232 Network Time a very useful potential service that can be provided by
233 a mid-level network. No specific protocol for maintaining time is
234 proposed, and any available protocol that maintains time with
235 reasonable accuracy could be used.
236
237 Network Time Protocol (NTP) traffic constituted 1% of the NSFnet
238 traffic during May 1991. The traffic might seem insignificant, but
239 there have been instances where a particular stratum-1 timeserver
240 (e.g., one of the stratum-1 servers at University of Delaware) has
241 reached a point of overload with too many different sites trying to
242 peer with it.
243
244 It is proposed that at least one stratum-1 and two stratum-2 servers
245 be located within a mid-level network (the selection of three servers
246 is based on the NTP standards documentation [<a href="#ref-MIL89" title="&quot;Network Time Protocol&quot;">MIL89</a>]). Note that the
247 servers can be located at any of the directly connected sites in the
248 network as long as they are publicly accessible. All sites connected
249 to the mid-level network can then coordinate their system times with
250 the servers within the mid-level network itself. Besides increasing
251 the reliability of the timekeeping network, this approach would also
252 limit the load on each timeserver.
253
254 For locating the network time servers within a domain, nameserver
255 entries for "timekeeper-x" (where x= 1,2,3..) can be made within the
256 domain. The servers are numbered in order of preference and accuracy.
257 Thus, "timekeeper-1.foo.net" would be the primary timekeeper and
258 "timekeeper-2.foo.net" would be additional (possibly secondary)
259 timekeepers within domain "foo.net". If such hosts are not available
260 within a domain, a TXT entry pointing to other recommended time-
261 servers could be provided instead.
262
263 <span class="h3"><a class="selflink" id="section-3.4" href="#section-3.4">3.4</a> Network News</span>
264
265 Network News (or Usenet News) constituted 14% of the NSFnet traffic
266 in May 1991. Netnews is an expensive service, both in terms of disk
267 and CPU power, as well as network bandwidth consumed.
268
269 The present structure of Network News consists of several hub sites
270 which are distributed over the Internet. End sites get news feeds
271 from other sites, and an article gets injected into the news stream
272 by sending it to the nearest "upstream" site, which then forwards it
273
274
275
276 <span class="grey">Aggarwal [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
279
280
281 to its connected news sites, and so on. There is no preset norm for
282 finding a site willing to provide a news feed, and it usually ends up
283 being a site with whom the site administrator happens to be
284 acquainted. However, this could easily result in some sites not being
285 able to get an economical news feed from within the mid-level network
286 and actually having to derive the feed from a site located on another
287 mid-level network.
288
289 A mid-level network could alleviate such occurrences by being able to
290 provide a newsfeed to any or all of its directly connected end sites.
291 Though an expensive resource, some of the costs can be moderated by
292 acting as a transit news feeder so that the news needn't be stored
293 for a long time on disk. The software for providing the news feed is
294 not specific and depends entirely on the newsfeed provider.
295
296 <span class="h3"><a class="selflink" id="section-3.5" href="#section-3.5">3.5</a> Mailing Lists</span>
297
298 Internet mailing lists are another popular source of information in
299 parallel to Network News. However, like public software, there is no
300 central repository of all the possible mailing lists available on the
301 Internet, and it would require considerable effort to compile one (at
302 the time of writing this document, a fairly comprehensive list is
303 available on the Internet and mentioned in <a href="#appendix-A">appendix A</a>.
304
305 At this time, there is no clear strategy for distributing or
306 maintaining mailing lists. However, it can be very expensive for a
307 site to distribute mail to all individual end users directly, and if
308 a clear strategy for maintaining a list of mailing-lists can be
309 devised, then mail exploders can be set up at the mid-level networks,
310 each of which forwards the mail to exploders at the end sites. This
311 mechanism would reduce the load on the originating systems, and
312 provides a clean path for tracking down mailer problems. Also, in
313 order to prevent bounced mail from propagating back to the originator
314 of the message, the mailing lists should be set up in a way so that
315 bounced mail goes to the the "owner" of the list and not to the
316 originator of the mail message.
317
318 A list of major mailing lists for the services discussed in this
319 document are listed in <a href="#appendix-A">appendix A</a>.
320
321 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Experimental Testbeds</span>
322
323 Due to the working relationships that they have with their end sites
324 and peer networks, the mid-level networks are very good media for
325 distribution of new ideas and technology. Examples of this function
326 are the White Pages pilot project [<a href="#ref-RS90" title="&quot;The NYSERnet White Pages Pilot Project&quot;">RS90</a>] established by NYSERnet, the
327 NSAP routing schema for OSI transitioning [<a href="#ref-CGC91" title="and R. Callon">CGC91</a>], etc.
328
329
330
331
332 <span class="grey">Aggarwal [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
335
336
337 The mid-level networks could establish cooperative experimental
338 testbeds for testing and deployment of new technologies similar to
339 the ones mentioned above. Besides deployment and testing of new
340 technology, this could also serve to provide a "help" service to the
341 end-sites and to get them started with the new software.
342
343 The exact interaction between the mid-level networks in this area is
344 not very clear. It is complicated by competition for members between
345 the mid-level networks and needs to be discussed further.
346
347 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Network Information Services</span>
348
349 There are a variety of new and useful user services available on the
350 Internet that are difficult to document and provide a comprehensive
351 list of. Some attempt has been made at documenting such resources
352 [<a href="#ref-NNS" title="&quot;Internet Resource Guide&quot;">NNS</a>] and a mid-level network can be the initial point of contact for
353 distribution of such information on a wide basis. The information can
354 be disseminated in a more controlled and complete manner using this
355 hierarchical approach if each mid-level network maintains up-to-date
356 information about its directly connected sites. Network Information
357 services (NIC) also make the network easier and more attractive to
358 end users. Examples of these services are:
359
360 o provide information resources
361
362 - security advisory messages
363
364 - list of library catalogs [<a href="#ref-GL91" title="&quot;Internet Accessible Library Catalogs &amp; Databases&quot;">GL91</a>]
365
366 - geographical information servers
367
368 - password generators
369
370 o resolve end user problems (user support)
371
372 These services are NIC related and discussed in detail elsewhere
373 [<a href="#ref-SSM91" title="&quot;Building a Network Information Services Infrastructure&quot;">SSM91</a>]. For accessibility information, an entry for "nic" could
374 exist in the DNS for the domain (this could be a TXT entry listing
375 email or phone number information for users or other NIC's).
376
377 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Network Operations</span>
378
379 The Network Operation Center's (NOC's) at the mid-level networks need
380 to cooperate with each other to resolve network problems. In the
381 event of a network problem between two mid-level networks or if an
382 end-site has trouble getting to any host, the mid-level network NOCs
383 can serve to be the initial point of contact. The procedures for
384 interaction among NOCs and the formats for exchange of trouble-
385
386
387
388 <span class="grey">Aggarwal [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
391
392
393 tickets between the NOCs are described elsewhere [<a href="#ref-JOH91" title="&quot;NOC TT Requirements&quot;">JOH91</a>, <a href="#ref-ML91" title="&quot;User Connectivity Problems Working Group&quot;">ML91</a>].
394
395 It is important for cooperating NOCs to have contact information for
396 their directly connected campus/organizational sites and also about
397 their peer mid-level networks. A distributed mechanism for
398 maintaining contact information could be implemented by using a
399 nameserver TXT entry for "noc" or by maintaining "finger" information
400 for user "noc@domain" or "[email protected]". A NOC "phonebook" listing
401 the contact information for the various NOCs can be used as a static
402 non-distributed mechanism (it is understood that the phonebook can
403 contain outdated information, but the distributed mechanisms can
404 provide correct and updated NOC information provided that the hosts
405 are reachable at the desired time). If it is undesirable to publish
406 the phone number or email address of the NOC for any reason, an entry
407 saying "unpublished" (or words to that effect) could exist in the
408 nameserver or "finger" entry instead.
409
410 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. References</span>
411
412 [<a id="ref-BOG">BOG</a>] Dunlap, K., and M. Karels, "Nameserver Operations Guide
413 for Bind Release 4.8", CSRG, Department of Electrical
414 Engineering and Computer Sciences, University of
415 California, Berkeley, California.
416
417 [<a id="ref-CCI88">CCI88</a>] CCITT Blue Book, "X.500 Series Recommendations", ITU,
418 1989.
419
420 [<a id="ref-CGC91">CGC91</a>] Collela, R., Gardner, E., and R. Callon, "Guidelines for
421 OSI NSAP Allocation in the Internet'', <a href="./rfc1237">RFC 1237</a>,
422 NIST, Mitre, DEC, July 1991.
423
424 [<a id="ref-SSM91">SSM91</a>] Sitzler, D., Smith, P., and A. Marine, "Building a Network
425 Information Services Infrastructure", RFC in
426 preparation.
427
428 [<a id="ref-GL91">GL91</a>] George, A., and R. Larsen, "Internet Accessible Library
429 Catalogs &amp; Databases", Aug 1991.
430 Available via anonymous FTP from ariel.unm.edu.
431
432 [<a id="ref-JOH91">JOH91</a>] Johnson, D., "NOC TT Requirements", RFC in
433 preparation.
434
435 [<a id="ref-MAN">MAN</a>] Mandelbaum, R., and P. Mandelbaum, "The Strategic Future
436 of the Mid-Level Networks", University of Rochester,
437 NY, 1991.
438
439 [<a id="ref-MOC87a">MOC87a</a>] Mockapetris, P., "Domain Names - Implementation and
440 Specification", <a href="./rfc1035">RFC 1035</a>, USC Information Sciences
441
442
443
444 <span class="grey">Aggarwal [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
447
448
449 Institute, November 1987.
450
451 [<a id="ref-MOC87b">MOC87b</a>] Mockapetris, P., "Domain Names - Concepts and
452 Facilities", <a href="./rfc1034">RFC 1034</a>, USC Information Sciences
453 Institute, November 1987.
454
455 [<a id="ref-MIL89">MIL89</a>] Mills, D., "Network Time Protocol", <a href="./rfc1129">RFC 1129</a>, UDel,
456 October 1989.
457
458 [<a id="ref-ML91">ML91</a>] Mathis, M., and D. Long, "User Connectivity Problems
459 Working Group", RFC in preparation.
460
461 [<a id="ref-NEU">NEU</a>] Neuman, B., "The Virtual System Model: A Scalable
462 Approach to Organizing Large Systems", Department of
463 Computer Science, University of Washington, FR-35,
464 Seattle, WA, May 1990.
465
466 [<a id="ref-NNS">NNS</a>] NSF Network Service Center, "Internet Resource Guide",
467 Cambridge, MA.
468 Available via anonymous FTP from nnsc.nsf.net.
469
470 [<a id="ref-RS90">RS90</a>] Rose, M., and M. Schoffstall, "The NYSERnet White Pages
471 Pilot Project", NYSERnet, Inc., Mar 1990.
472
473 [<a id="ref-SHHH90">SHHH90</a>] Schwartz, M., Hardy, D., Heinzman, W., and G.
474 Hirschowitz, "Supporting Resource Discovery Among
475 Public Internet Archives", Department of Computer
476 Science, University of Colorado, Boulder, CO.,
477 September 1990.
478
479 <span class="h2"><a class="selflink" id="section-8" href="#section-8">8</a>. Security Considerations</span>
480
481 Security issues are not discussed in this memo.
482
483 <span class="h2"><a class="selflink" id="section-9" href="#section-9">9</a>. Author's Address</span>
484
485 Vikas Aggarwal
486 JvNCnet
487 6 von Neumann Hall
488 Princeton University
489 Princeton, NJ 08544
490
491 Phone: +1-609-258-2403
492 Email: [email protected]
493
494
495
496
497
498
499
500 <span class="grey">Aggarwal [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc1291">RFC 1291</a> Potential Technical Services December 1991</span>
503
504
505 Appendix A - Mailing Lists
506
507 The following is a list of popular mailing lists for the services
508 listed in this document. To subscribe to a particular mailing list,
509 send a request to "mailing-list-request" (do not send a request to
510 the entire mailing list).
511
512 o [email protected]: The general mailing list for the Internet
513 Engineering Task Force. This group is concerned with the evolution
514 and development of Internet related protocols and standards. Old
515 mail is archived at "venera.isi.edu" in directory ftp/irg/ietf.
516
517 o [email protected]: For discussions on the Network Time
518 Protocol (NTP).
519
520 o [email protected]: Mailing list for discussions on DNS
521 topics. Old mail is archived at "nic.ddn.mil".
522
523 At the time of writing this document, a list of mailing lists on the
524 Internet is available via anonymous FTP from host "ftp.nisc.sri.com"
525 in the file "netinfo/interest-groups".
526
527 Appendix B - DNS Architecture Strategy
528
529 This section discusses practical strategies for implementing a
530 nameserver architecture within a mid-level network, so that it can
531 resolve nameserver queries for all domains directly attached to it.
532
533 In order to resolve queries for all directly connected networks, a
534 host that is authoritative for all directly attached domains will
535 need to exist within the mid-level network. Nameservers at the end
536 sites would then treat this "group-of-domains" nameserver as a
537 forwarding server to resolve all non-local queries.
538
539 This can be done by adding a line to the named.boot file on the end
540 site nameservers such as:
541
542 forwarders 128.121.50.7 128.32.0.4
543
544 This method has the added advantage that the forwarding server builds
545 up a very rich cache of data [<a href="#ref-BOG" title="&quot;Nameserver Operations Guide for Bind Release 4.8&quot;">BOG</a>] and acts like a metacache that all
546 hosts can benefit from. Note that the forwarding server is queried
547 only if the end-site server cannot service a query locally -- hence
548 the "meta-domain" server is not overloaded with queries for all
549 nameserver lookups.
550
551
552
553
554
555
556 Aggarwal [Page 10]
557 </pre>