RFC 4692 - Considerations on the IPv6 Host Density Metric(2)

时间:2006-11-02 来源: 作者: 点击:
addressconsumptionofapproximatelyoneorderofmagnitude.Thisis aheavy-taildistribution,whereasmallproportionoflargeaddress allocationssignificantlyimpactthetotaladdressconsumptionrate. AlteringtheHD-Rat
  
   address consumption of approximately one order of magnitude.  This is
   a heavy-tail distribution, where a small proportion of large address
   allocations significantly impact the total address consumption rate.
   Altering the HD-Ratio will have little impact on more than 95% of the
   IPv6 allocations but will generate significant variance within the
   largest 2% of these allocations, which, in turn, will have a
   significant impact on total address consumption rates.

7.  Considerations

   The HD-Ratio with a value of 0.8 as a model of network address
   utilization efficiency produces extremely low efficiency outcomes for
   networks spanning of the order of 10**6 end customers and larger.

   The HD-Ratio with a 0.8 value makes the assumption that as the
   address allocation block increases in size, the network within which
   the addresses will be deployed adds additional levels of hierarchical
   structure.  This increasing depth of hierarchical structure to
   arbitrarily deep hierarchies is not a commonly observed feature of
   public IP network deployments.

   The fixed efficiency model, as used in the IPv4 address allocation
   policy, uses the assumption that as the allocation block becomes
   larger, the network structure remains at a fixed level of levels; if
   the number of levels is increased, then efficiency achieved at each
   level increases significantly.  There is little evidence to suggest
   that increasing a number of levels in a network hierarchy increases
   the efficiency at each level.

   It is evident that neither of these models accurately encompass IP
   network infrastructure models and the associated requirements of
   address deployment.  The fixed efficiency model places an excessive
   burden on the network operator to achieve very high levels of
   utilization at each level in the network hierarchy, leading to either
   customer renumbering or deployment of technologies such as Network
   Address Translation (NAT) to meet the target efficiency value in a

   hierarchically structured network.  The HD-Ratio model using a value
   of 0.8 specifies an extremely low address efficiency target for
   larger networks, and while this places no particular stress on
   network architects in terms of forced renumbering, there is the
   concern that this represents an extremely inefficient use of address
   resources.  If the objective of IPv6 is to encompass a number of
   decades of deployment, and to span a public network that ultimately
   encompasses many billions of end customers and a very high range and
   number of end use devices and components, then there is legitimate
   cause for concern that the HD-Ratio value of 0.8 may be setting too
   conservative a target for address efficiency, in that the total
   address consumption targets may be achieved too early.

   This study concludes that consideration should be given to the
   viability of specifying a higher HD-Ratio value as representing a
   more relevant model of internal network structure, internal routing,
   and internal address aggregation structures in the context of IPv6
   network deployment.

8.  Security Considerations

   Considerations of various forms of host density metrics create no new
   threats to the security of the Internet.

9.  Acknowledgements

   The document was reviewed by Kurt Lindqvist, Thomas Narten, Paul
   Wilson, David Kessens, Bob Hinden, Brian Haberman, and Marcelo
   Bagnulo.

10. References

10.1.  Normative References

   [RFC1715]   Huitema, C., "The H Ratio for Address Assignment
               Efficiency", RFC 1715, November 1994.

   [RFC3177]   IAB and IESG, "IAB/IESG Recommendations on IPv6 Address
               Allocations to Sites", RFC 3177, September 2001.

   [RFC3194]   Durand, A. and C. Huitema, "The H-Density Ratio for
               Address Assignment Efficiency An Update on the H ratio",
               RFC 3194, November 2001.

   [RFC4291]   Hinden, R. and S. Deering, "IP Version 6 Addressing
               Architecture", RFC 4291, February 2006.

10.2.  Informative References

   [RIR-Data]  RIRs, "RIR Delegation Records", February 2005,
               <ftp://ftp.apnic.net/pub/stats/>.

Appendix A.  Comparison Tables

   The first table compares the threshold number of /48 end user
   allocations that would be performed for a given assigned address
   block in order to consider that the utilization has achieved its
   threshold utilization level.

   Fixed Efficiency Value  0.8
   HD-Ratio Value          0.8

                               Number of /48 allocations to fill the
                                address block to the threshold level

   Prefix          Size              Fixed Efficiency      HD-Ratio
                                       0.8                     0.8

   /48                 1                 1 100%              1  100%
   /47                 2                 2 100%              2   87%
   /46                 4                 4 100%              3   76%
   /45                 8                 7  88%              5   66%
   /44                16                13  81%              9   57%
   /43                32                26  81%             16   50%
   /42                64                52  81%             28   44%
   /41               128               103  80%             49   38%
   /40               256               205  80%             84   33%
   /39               512               410  80%            147   29%
   /38             1,024               820  80%            256   25%
   /37             2,048             1,639  80%            446   22%
   /36             4,096             3,277  80%            776   19%
   /35             8,192             6,554  80%          1,351   16%
   /34            16,384            13,108  80%          2,353   14%
   /33            32,768            26,215  80%          4,096   13%
   /32            65,536            52,429  80%          7,132   11%
   /31           131,072           104,858  80%         12,417    9%
   /30           262,144           209,716  80%         21,619    8%
   /29           524,288           419,431  80%         37,641    7%
   /28         1,048,576           838,861  80%         65,536    6%
   /27         2,097,152         1,677,722  80%        114,105    5%
   /26         4,194,304         3,355,444  80%        198,668    5%
   /25         8,388,608         6,710,887  80%        345,901    4%
   /24        16,777,216        13,421,773  80%        602,249    4%
   /23        33,554,432        26,843,546  80%      1,048,576    3%
   /22        67,108,864        53,687,092  80%      1,825,677    3%
   /21       134,217,728       107,374,180  80%      3,178,688    2%
   /20       268,435,456       214,748,365  80%      5,534,417    2%
   /19       536,870,912       429,496,730  80%      9,635,980    2%
   /18     1,073,741,824       858,993,460  80%     16,777,216    2%
   /17     2,147,483,648     1,717,986,919  80%     29,210,830    1%

   /16     4,294,967,296     3,435,973,837  80%     50,859,008    1%
   /15     8,589,934,592     6,871,947,674  80%     88,550,677    1%
   /14    17,179,869,184    13,743,895,348  80%    154,175,683    1%
   /13    34,359,738,368    27,487,790,695  80%    268,435,456    1%
   /12    68,719,476,736    54,975,581,389  80%    467,373,275    1%
   /11   137,438,953,472   109,951,162,778  80%    813,744,135    1%
   /10   274,877,906,944   219,902,325,556  80%  1,416,810,831    1%
   /9    549,755,813,888   439,804,651,111  80%  2,466,810,934    0%
   /8  1,099,511,627,776   879,609,302,221  80%  4,294,967,296    0%
   /7  2,199,023,255,552 1,759,218,604,442  80%  7,477,972,398    0%
   /6  4,398,046,511,104 3,518,437,208,884  80% 13,019,906,166    0%
   /5  8,796,093,022,208 7,036,874,417,767  80% 22,668,973,294    0%

           Table 1.  Comparison of Fixed Efficiency Threshold vs
                     HD-Ratio Threshold

                                 Figure 7

   One possible assumption behind the HD-Ratio is that the
   inefficiencies that are a consequence of large-scale deployments are
   an outcome of an increased number of levels of hierarchical structure
   within the network.  The following table calculates the depth of the
   hierarchy in order to achieve a 0.8 HD-Ratio, assuming a 0.8
   utilization efficiency at each level in the hierarchy.

   Prefix          Size              0.8 Structure
                                HD-Ratio    Levels
   /48                 1               1         1
   /47                 2               2         1
   /46                 4               3         2
   /45                 8               5         2
   /44                16               9         3
   /43                32              16         4
   /42                64              28         4
   /41               128              49         5
   /40               256              84         5
   /39               512             147         6
   /38             1,024             256         7
   /37             2,048             446         7
   /36             4,096             776         8
   /35             8,192           1,351         9
   /34            16,384           2,353         9
   /33            32,768           4,096        10
   /32            65,536           7,132        10
   /31           131,072          12,417        11
   /30           262,144          21,619        12
   /29           524,288          37,641        12
   /28         1,048,576          65,536        13

   /27         2,097,152         114,105        14
   /26         4,194,304         198,668        14
   /25         8,388,608         345,901        15
   /24        16,777,216         602,249        15
   /23        33,554,432       1,048,576        16
   /22        67,108,864       1,825,677        17
   /21       134,217,728       3,178,688        17
   /20       268,435,456       5,534,417        18
   /19       536,870,912       9,635,980        19
   /18     1,073,741,824      16,777,216        19
   /17     2,147,483,648      29,210,830        20
   /16     4,294,967,296      50,859,008        20
   /15     8,589,934,592      88,550,677        21
   /14    17,179,869,184     154,175,683        22
   /13    34,359,738,368     268,435,456        22
   /12    68,719,476,736     467,373,275        23
   /11   137,438,953,472     813,744,135        23
   /10   274,877,906,944   1,416,810,831        24
   /9    549,755,813,888   2,466,810,934        25
   /8  1,099,511,627,776   4,294,967,296        25

          Table 2: Number of Structure Levels Assumed by HD-Ratio

                                 Figure 8

   An alternative approach is to use a model of network deployment where
   the number of levels of hierarchy increases at a lower rate than that
   indicated in a 0.8 HD-Ratio model.  One such model is indicated in
   the following table.  This is compared to using an HD-Ratio value of
   0.94.

   Per-Level Target Efficiency: 0.75

   Prefix           Size Stepped      Stepped Efficiency      HD-Ratio
                         Levels          0.75                   0.94

   /48                 1  1                1 100%                 1 100%
   /47                 2  1                2 100%                 2 100%
   /46                 4  1                3  75%                 4 100%
   /45                 8  1                6  75%                 7  88%
   /44                16  1               12  75%                13  81%
   /43                32  1               24  75%                25  78%
   /42                64  1               48  75%                48  75%
   /41               128  1               96  75%                92  72%
   /40               256  1              192  75%               177  69%
   /39               512  2              384  75%               338  66%
   /38             1,024  2              576  56%               649  63%
   /37             2,048  2            1,152  56%             1,244  61%

   /36             4,096  2            2,304  56%             2,386  58%
   /35             8,192  2            4,608  56%             4,577  56%
   /34            16,384  2            9,216  56%             8,780  54%
   /33            32,768  2           18,432  56%            16,845  51%
   /32            65,536  2           36,864  56%            32,317  49%
   /31           131,072  3           73,728  56%            62,001  47%
   /30           262,144  3          110,592  42%           118,951  45%
   /29           524,288  3          221,184  42%           228,210  44%
   /28         1,048,576  3          442,368  42%           437,827  42%
   /27         2,097,152  3          884,736  42%           839,983  40%
   /26         4,194,304  3        1,769,472  42%         1,611,531  38%
   /25         8,388,608  3        3,538,944  42%         3,091,767  37%
   /24        16,777,216  3        7,077,888  42%         5,931,642  35%
   /23        33,554,432  4       14,155,776  42%        11,380,022  34%
   /22        67,108,864  4       21,233,664  32%        21,832,894  33%
   /21       134,217,728  4       42,467,328  32%        41,887,023  31%
   /20       268,435,456  4       84,934,656  32%        80,361,436  30%
   /19       536,870,912  4      169,869,312  32%       154,175,684  29%
   /18     1,073,741,824  4      339,738,624  32%       295,790,403  28%
   /17     2,147,483,648  4      679,477,248  32%       567,482,240  26%
   /16     4,294,967,296  4    1,358,954,496  32%     1,088,730,702  25%
   /15     8,589,934,592  5    2,717,908,992  32%     2,088,760,595  24%
   /14    17,179,869,184  5    4,076,863,488  24%     4,007,346,185  23%
   /13    34,359,738,368  5    8,153,726,976  24%     7,688,206,818  22%
   /12    68,719,476,736  5   16,307,453,952  24%    14,750,041,884  21%
   /11   137,438,953,472  5   32,614,907,904  24%    28,298,371,876  21%
   /10   274,877,906,944  5   65,229,815,808  24%    54,291,225,552  20%
   /9    549,755,813,888  5  130,459,631,616  24%   104,159,249,331  19%
   /8  1,099,511,627,776  5  260,919,263,232  24%   199,832,461,158  18%

                   Table 3: Limited Levels of Structure

                                 Figure 9

Author’s Address

   Geoff Huston
   APNIC

   EMail: gih@apnic.net

Full Copyright Statement

   Copyright (C) The Internet Society (2006).

   This document is subject to the rights, licenses and restrictions
   contained in BCP 78, and except as set forth therein, the authors
   retain all their rights.

   This document and the information contained herein are provided on an
   "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
   OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET
   ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED,
   INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE
   INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
   WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in BCP 78 and BCP 79.

   Copies of IPR disclosures made to the IETF Secretariat and any
   assurances of licenses to be made available, or the result of an
   attempt made to obtain a general license or permission for the use of
   such proprietary rights by implementers or users of this
   specification can be obtained from the IETF on-line IPR repository at
   http://www.ietf.org/ipr.

   The IETF invites any interested party to bring to its attention any
   copyrights, patents or patent applications, or other proprietary
   rights that may cover technology that may be required to implement
   this standard.  Please address the information to the IETF at
   ietf-ipr@ietf.org.

Acknowledgement

   Funding for the RFC Editor function is provided by the IETF
   Administrative Support Activity (IASA).
------分隔线----------------------------
顶一下
(1)
100%
踩一下
(0)
0%
------分隔线----------------------------
最新评论 查看所有评论
发表评论 查看所有评论
请自觉遵守互联网相关的政策法规,严禁发布色情、暴力、反动的言论。
评价:
表情:
用户名: 密码: 验证码:
推荐内容