| 1 |
<pre>Network Working Group Z. Wang |
| 2 |
Request for Comments: 1335 J. Crowcroft |
| 3 |
University College London |
| 4 |
May 1992 |
| 5 |
|
| 6 |
|
| 7 |
<span class="h1">A Two-Tier Address Structure for the Internet:</span> |
| 8 |
<span class="h1">A Solution to the Problem of Address Space Exhaustion</span> |
| 9 |
|
| 10 |
Status of this Memo |
| 11 |
|
| 12 |
This memo provides information for the Internet community. It does |
| 13 |
not specify an Internet standard. Distribution of this memo is |
| 14 |
unlimited. |
| 15 |
|
| 16 |
Abstract |
| 17 |
|
| 18 |
This RFC presents a solution to problem of address space exhaustion |
| 19 |
in the Internet. It proposes a two-tier address structure for the |
| 20 |
Internet. This is an "idea" paper and discussion is strongly |
| 21 |
encouraged. |
| 22 |
|
| 23 |
Introduction |
| 24 |
|
| 25 |
Address space exhaustion is one of the most serious and immediate |
| 26 |
problems that the Internet faces today [<a href="#ref-1" title=""The IP Addressing Issue"">1</a>,<a href="#ref-2" title=""Towards the Future Architecture"">2</a>]. The current Internet |
| 27 |
address space is 32-bit. Each Internet address is divided into two |
| 28 |
parts: a network portion and a host portion. This division |
| 29 |
corresponds the three primary Internet address classes: Class A, |
| 30 |
Class B and Class C. Table 1 lists the network number statistics as |
| 31 |
of April 1992. |
| 32 |
|
| 33 |
Total Allocated Allocated (%) |
| 34 |
Class A 126 48 54% |
| 35 |
Class B 16383 7006 43% |
| 36 |
Class C 2097151 40724 2% |
| 37 |
|
| 38 |
Table 1: Network Number Statistics (April 1992) |
| 39 |
|
| 40 |
If recent trends of exponential growth continue, the network numbers |
| 41 |
in Class B will soon run out [<a href="#ref-1" title=""The IP Addressing Issue"">1</a>,<a href="#ref-2" title=""Towards the Future Architecture"">2</a>]. There are over 2 million Class C |
| 42 |
network numbers and only 2% have been allocated. However, a Class C |
| 43 |
network number can only accommodate 254 host numbers which is too |
| 44 |
small for most networks. With the rapid expansion of the Internet |
| 45 |
and drastic increase in personal computers, the time when the 32-bit |
| 46 |
address space is exhausted altogether is also not too distant [1-3]. |
| 47 |
|
| 48 |
Recently several proposals have been put forward to deal with the |
| 49 |
|
| 50 |
|
| 51 |
|
| 52 |
<span class="grey">Wang & Crowcroft [Page 1]</span></pre> |
| 53 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span> |
| 54 |
<span class="grey"><a href="./rfc1335">RFC 1335</a> Two-Tier Address Structure for the Internet May 1992</span> |
| 55 |
|
| 56 |
|
| 57 |
immediate problem [1-4]. The Supernetting and C-sharp schemes |
| 58 |
attempt to make the Class C numbers more usable by re-defining the |
| 59 |
way in which Class C network numbers are classified and assigned |
| 60 |
[<a href="#ref-3" title=""A Revision to IP Address Classifications"">3</a>,<a href="#ref-4" title=""Supernetting: an Address Assignment and Aggregation Strategy"">4</a>]. Both schemes require modifications to the exterior routing |
| 61 |
algorithms and global coordination across the Internet may be |
| 62 |
required for the deployment. The two schemes do not expand the total |
| 63 |
number of addresses available to the Internet and therefore can only |
| 64 |
be used as a short-term fix for next two or three years. Schemes |
| 65 |
have also been put forwarded in which the 32-bit address field is |
| 66 |
replaced with a field of the same size but with different meaning and |
| 67 |
the gateways on the boundary re-write the address when the packet |
| 68 |
crossed the boundary [<a href="#ref-1" title=""The IP Addressing Issue"">1</a>,<a href="#ref-2" title=""Towards the Future Architecture"">2</a>,<a href="#ref-5" title=""The IP Network Address Translator"">5</a>]. Such schemes, however, requires |
| 69 |
substantial changes to the gateways and the exterior routing |
| 70 |
algorithm. |
| 71 |
|
| 72 |
In this paper, we present an alternative solution to the problem of |
| 73 |
address space exhaustion. The "Dual Network Addressing (DNA)" scheme |
| 74 |
proposed here is based on a two-tier address structure and sharing of |
| 75 |
addresses. It requires no modifications to the exterior routing |
| 76 |
algorithms and any networks can adopt the scheme individually at any |
| 77 |
time without affecting other networks. |
| 78 |
|
| 79 |
The Scheme |
| 80 |
|
| 81 |
The DNA scheme attempts to reduce the waste in using the Internet |
| 82 |
addresses. A useful analogy to our scheme is the extension system |
| 83 |
used in the telephone system. Many large organizations usually have |
| 84 |
extensive private telephone networks for internal use and at the mean |
| 85 |
time hire a limited number of external lines for communications with |
| 86 |
the outside world. In such a telephone system, important offices may |
| 87 |
have direct external lines and telephones in the public areas may be |
| 88 |
restricted to internal calls only. The majority of the telephones |
| 89 |
can usually make both internal calls and external calls. But they |
| 90 |
must share a limited number of external lines. When an external call |
| 91 |
is being made, a pre-defined digit has to be pressed so that an |
| 92 |
external line can be allocated from the poll of external lines. |
| 93 |
|
| 94 |
In the DNA scheme, there are two types of Internet addresses: |
| 95 |
Internal addresses and External addresses. An internal address is an |
| 96 |
Internet address only used within one network and is unique only |
| 97 |
within that network. An interface with an internal address can only |
| 98 |
communicate with another interface with an internal address in the |
| 99 |
same network. An external address is unique in the entire Internet |
| 100 |
and an interface with an external address can communicate directly to |
| 101 |
another interface with an external address over the Internet. All |
| 102 |
current Internet addresses are external addresses. |
| 103 |
|
| 104 |
In effect, the external addresses form one global Internet and the |
| 105 |
|
| 106 |
|
| 107 |
|
| 108 |
<span class="grey">Wang & Crowcroft [Page 2]</span></pre> |
| 109 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span> |
| 110 |
<span class="grey"><a href="./rfc1335">RFC 1335</a> Two-Tier Address Structure for the Internet May 1992</span> |
| 111 |
|
| 112 |
|
| 113 |
internal addresses form many private Internets. Within one network, |
| 114 |
the external addresses are only used for inter-network communications |
| 115 |
and internal addresses for intra-network communications. An External |
| 116 |
Address Sharing Service (EASS) is needed to manage the sharing of |
| 117 |
external addresses. An EASS server reserves a number of external |
| 118 |
addresses. When a machine that only has an internal address wants to |
| 119 |
communicate a machine with an external address in other networks, it |
| 120 |
can send a request to an EASS server to obtain a temporary external |
| 121 |
address. After the use, the machine can return the external address |
| 122 |
to the EASS server. |
| 123 |
|
| 124 |
We believe that, with the DNA scheme, a network can operate with a |
| 125 |
limited number of external addresses. The reasons are as follows: |
| 126 |
|
| 127 |
* In most networks, the majority of the traffic is confined to |
| 128 |
its local area networks. This is due the nature of |
| 129 |
networking applications and the bandwidth constraints on |
| 130 |
inter-network links. |
| 131 |
|
| 132 |
* The number of machines which act as Internet servers, i.e., |
| 133 |
running programs waiting to be called by machines in other |
| 134 |
networks, is often limited and certainly much smaller than |
| 135 |
the total number of machines. These machines include mail |
| 136 |
servers, domain name servers, ftp archive servers, directory |
| 137 |
servers, etc. |
| 138 |
|
| 139 |
* There are an increasingly large number of personal machines |
| 140 |
entering the Internet. The use of these machines is |
| 141 |
primarily limited to their local environment. They may also |
| 142 |
be used as "clients" such as ftp and telnet to access other |
| 143 |
machines. |
| 144 |
|
| 145 |
* For security reasons, many large organizations, such as banks, |
| 146 |
government departments, military institution and some |
| 147 |
companies, may only allow a very limited number of their |
| 148 |
machines to have access to the global Internet. The majority |
| 149 |
of their machines are purely for internal use. |
| 150 |
|
| 151 |
In the DNA scheme, all machines in a network are assigned a permanent |
| 152 |
internal address and can communicate with any machines within the |
| 153 |
same network. The allocation of external addresses depends on the |
| 154 |
functions of the machines and as a result it creates three-level |
| 155 |
privileges: |
| 156 |
|
| 157 |
* machines which act as servers or used as central computing |
| 158 |
infrastructure are likely to have frequent communications |
| 159 |
with other networks therefore they may require external |
| 160 |
addresses all the time. These machines are allocated |
| 161 |
|
| 162 |
|
| 163 |
|
| 164 |
<span class="grey">Wang & Crowcroft [Page 3]</span></pre> |
| 165 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span> |
| 166 |
<span class="grey"><a href="./rfc1335">RFC 1335</a> Two-Tier Address Structure for the Internet May 1992</span> |
| 167 |
|
| 168 |
|
| 169 |
permanent external addresses. |
| 170 |
|
| 171 |
* machines which are not allowed to communicate with other |
| 172 |
networks have no external addresses and can only communicate |
| 173 |
with machines within their own network. |
| 174 |
|
| 175 |
* the rest of the machines share a number of external |
| 176 |
addresses. The external addresses are allocated by |
| 177 |
the EASS server on request. These machines can only |
| 178 |
used as clients to call machines in other networks, |
| 179 |
i.e., they can not be called by machines in other networks. |
| 180 |
|
| 181 |
A network can choose any network number other than its external |
| 182 |
network number as its internal network number. Different networks |
| 183 |
can use the same network number as their internal number. We propose |
| 184 |
to reserve one Class A network number as the well-known network |
| 185 |
number for internal use. |
| 186 |
|
| 187 |
The Advantages |
| 188 |
|
| 189 |
The DNA scheme attempts to tackle the problem from the bottom of the |
| 190 |
Internet, i.e., each individual network, while other schemes |
| 191 |
described in the first section deal with the problem from the top of |
| 192 |
the Internet, i.e., gateways and exterior routing algorithms. These |
| 193 |
schemes, however, do not need to be consider as mutually exclusive. |
| 194 |
The DNA scheme has several advantages: |
| 195 |
|
| 196 |
* The DNA scheme takes an evolutionary approach towards the |
| 197 |
changes. Different networks can individually choose to |
| 198 |
adopt the scheme at any time only when necessary. |
| 199 |
There is no need for global coordination between different |
| 200 |
networks for their deployment. The effects of the deployment |
| 201 |
are confined to the network in which the scheme is being |
| 202 |
implemented, and are invisible to exterior routing |
| 203 |
algorithms and external networks. |
| 204 |
|
| 205 |
* With the DNA scheme, it is possible for a medium size organization |
| 206 |
to use a Class C network number with 254 external addresses. |
| 207 |
The scheme allows the current Internet to expand to over 2 million |
| 208 |
networks and each network to have more than 16 million hosts. |
| 209 |
This will allow considerable time for a long-term solution to |
| 210 |
be developed and fully tested. |
| 211 |
|
| 212 |
* The DNA scheme requires modifications to the host software. |
| 213 |
However, the modifications are needed only in those networks |
| 214 |
which adopt the DNA scheme. Since all existing Class A and B |
| 215 |
networks usually have sufficient external addresses for all their |
| 216 |
machines, they do not need to adopt the DNA scheme, and therefore |
| 217 |
|
| 218 |
|
| 219 |
|
| 220 |
<span class="grey">Wang & Crowcroft [Page 4]</span></pre> |
| 221 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span> |
| 222 |
<span class="grey"><a href="./rfc1335">RFC 1335</a> Two-Tier Address Structure for the Internet May 1992</span> |
| 223 |
|
| 224 |
|
| 225 |
need no modifications at all to their software. The networks |
| 226 |
which need to use the DNA scheme are those new networks which are |
| 227 |
set up after the Class A and B numbers run out and have to |
| 228 |
use a Class C number. |
| 229 |
|
| 230 |
* The DNA scheme makes it possible to develop to a new addressing |
| 231 |
scheme without expanding the 32-bit address length to 64-bit. |
| 232 |
With the two-tier address structure, the current 32-bit space |
| 233 |
can accommodate over 4 billion hosts in the global Internet and |
| 234 |
100 million hosts in each individual network. When we move to a |
| 235 |
classless multi-hierarchic addressing scheme, the use of external |
| 236 |
addresses can be more efficient and less wasteful and the |
| 237 |
32-bit space can be adequate for the external addresses. |
| 238 |
|
| 239 |
* When a new addressing scheme has been developed, all current |
| 240 |
Internet addresses have to be changed. The DNA scheme will make |
| 241 |
such a undertaking much easier and smoother, since only the |
| 242 |
EASS servers and those have permanent external addresses will |
| 243 |
be affected, and communications within the network will not |
| 244 |
be interrupted. |
| 245 |
|
| 246 |
The Modifications |
| 247 |
|
| 248 |
The major modifications to the host software is in the network |
| 249 |
interface code. The DNA scheme requires each machine to have at |
| 250 |
least two addresses. But most of the host software currently does |
| 251 |
not allow us to bind two addresses to one physical interface. This |
| 252 |
problem can be solved by using two network interfaces on each |
| 253 |
machine. But this option is too expensive. Note the two interfaces |
| 254 |
are actually connected to the same physical network. Therefore, if |
| 255 |
we modify the interface code to allow two logical interfaces to be |
| 256 |
mapped onto one single physical interface, the machine can then use |
| 257 |
both the external address and the internal address with one physical |
| 258 |
interface as if it has two physical interfaces. In effect, two |
| 259 |
logical IP networks operate over the same physical network. |
| 260 |
|
| 261 |
The DNA scheme also has implications to the DNS service. Many |
| 262 |
machines will have two entries in the local name server. The DNS |
| 263 |
server must examine the source address of the request and decide |
| 264 |
which entry to use. If the source address matches the well-known |
| 265 |
internal network number, it passes the internal address of the domain |
| 266 |
name. Otherwise, the name server passes the external address. |
| 267 |
|
| 268 |
An EASS server is required to manage the sharing of the external |
| 269 |
addresses, i.e., to allocate and de-allocate external addresses to |
| 270 |
the machines which do not have permanent external addresses. This |
| 271 |
service can be provided by using the "Dynamic Host Configuration |
| 272 |
Protocol (DHCP)" [<a href="#ref-6" title=""Dynamic Host Configuration Protocol"">6</a>]. |
| 273 |
|
| 274 |
|
| 275 |
|
| 276 |
<span class="grey">Wang & Crowcroft [Page 5]</span></pre> |
| 277 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span> |
| 278 |
<span class="grey"><a href="./rfc1335">RFC 1335</a> Two-Tier Address Structure for the Internet May 1992</span> |
| 279 |
|
| 280 |
|
| 281 |
Many hosts do an inverse lookup of incoming connections. Therefore, |
| 282 |
it is desirable the entry in the DNS server be updated whenever a new |
| 283 |
external address is allocated. This will also allow an machine which |
| 284 |
currently has a temporary external address to be called by other |
| 285 |
machines. The updating of the entry in the DNS server can be done |
| 286 |
more easily if the EASS server and DNS server are co-located. |
| 287 |
|
| 288 |
Acknowledgements |
| 289 |
|
| 290 |
We would like to thank J. K. Reynolds for the network statistics, and |
| 291 |
V. Cerf, C. Topolcic, K. McCloghrie, R. Ullmann and K. Carlberg for |
| 292 |
their useful comments and discussion. |
| 293 |
|
| 294 |
References |
| 295 |
|
| 296 |
[<a id="ref-1">1</a>] Chiappa, N., "The IP Addressing Issue", work in progress, |
| 297 |
October 1990. |
| 298 |
|
| 299 |
[<a id="ref-2">2</a>] Clark, D., Chapin, L., Cerf, V., Braden, R., and R. Hobby, |
| 300 |
"Towards the Future Architecture", <a href="./rfc1287">RFC 1287</a>, MIT, BBN, CNRI, |
| 301 |
ISI, UC Davis, December 1991. |
| 302 |
|
| 303 |
[<a id="ref-3">3</a>] Solensky, F., and F. Kastenholz, "A Revision to IP Address |
| 304 |
Classifications", work in progress, March 1992. |
| 305 |
|
| 306 |
[<a id="ref-4">4</a>] Fuller, V., Li, T., Yu, J., and K. Varadhan, "Supernetting: |
| 307 |
an Address Assignment and Aggregation Strategy", work in |
| 308 |
progress, March 1992. |
| 309 |
|
| 310 |
[<a id="ref-5">5</a>] Tsuchiya, P., "The IP Network Address Translator", work in |
| 311 |
progress, March 1991. |
| 312 |
|
| 313 |
[<a id="ref-6">6</a>] Droms, R., "Dynamic Host Configuration Protocol", work in |
| 314 |
progress, March 1992. |
| 315 |
|
| 316 |
|
| 317 |
|
| 318 |
|
| 319 |
|
| 320 |
|
| 321 |
|
| 322 |
|
| 323 |
|
| 324 |
|
| 325 |
|
| 326 |
|
| 327 |
|
| 328 |
|
| 329 |
|
| 330 |
|
| 331 |
|
| 332 |
<span class="grey">Wang & Crowcroft [Page 6]</span></pre> |
| 333 |
<hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span> |
| 334 |
<span class="grey"><a href="./rfc1335">RFC 1335</a> Two-Tier Address Structure for the Internet May 1992</span> |
| 335 |
|
| 336 |
|
| 337 |
Security Considerations |
| 338 |
|
| 339 |
Security issues are not discussed in this memo. |
| 340 |
|
| 341 |
Authors' Addresses |
| 342 |
|
| 343 |
Zheng Wang |
| 344 |
Dept. of Computer Science |
| 345 |
University College London |
| 346 |
London WC1E 6BT, UK |
| 347 |
|
| 348 |
EMail: [email protected] |
| 349 |
|
| 350 |
|
| 351 |
Jon Crowcroft |
| 352 |
Dept. of Computer Science |
| 353 |
University College London |
| 354 |
London WC1E 6BT, UK |
| 355 |
|
| 356 |
EMail: [email protected] |
| 357 |
|
| 358 |
|
| 359 |
|
| 360 |
|
| 361 |
|
| 362 |
|
| 363 |
|
| 364 |
|
| 365 |
|
| 366 |
|
| 367 |
|
| 368 |
|
| 369 |
|
| 370 |
|
| 371 |
|
| 372 |
|
| 373 |
|
| 374 |
|
| 375 |
|
| 376 |
|
| 377 |
|
| 378 |
|
| 379 |
|
| 380 |
|
| 381 |
|
| 382 |
|
| 383 |
|
| 384 |
|
| 385 |
|
| 386 |
|
| 387 |
|
| 388 |
Wang & Crowcroft [Page 7] |
| 389 |
</pre> |