[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: draft-ietf-v6ops-addr-select-ps-02



-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Brian, I'm very perlexed by how you come to that conclusion. Maybe you can fill me in.

The text of rule 9 is:

   Rule 9:  Use longest matching prefix.
   When DA and DB belong to the same address family (both are IPv6 or
   both are IPv4): If CommonPrefixLen(DA, Source(DA)) >
   CommonPrefixLen(DB, Source(DB)), then prefer DA.  Similarly, if
   CommonPrefixLen(DA, Source(DA)) < CommonPrefixLen(DB, Source(DB)),
   then prefer DB.

In the Internet, regardless of whether the prefixes are PA or PI, there is little if any reason to believe that any two customers chosen at random have the same provider. I have appended four traceroutes for your (anecdotal) consideration in support of the point - I am served by Cox, but you use google mail (which apparently has multiple ISPs), Florian is reached through interscholz.net, ops.ietf.org is reached through NTT, and ietf.org is reached through AT&T. I would encourage you to spend an hour picking random domain names out of your mail spool and traceroute to them. If your view of the world is at all similar to mine, you will find your path going through your network, your provider, their provider, and their network, and potentially multiple networks in between the two providers.

If the two communicating networks have different upstream providers, what on God's Green Earth leads you to believe that they would have a common PA prefix? If they don't have a common prefix, exactly how does rule 9 help?




On Jan 27, 2008, at 12:39 PM, Brian E Carpenter wrote:

On 2008-01-28 00:26, Florian Weimer wrote:
I've also noticed that the draft fails to mention that Rule 9, when
applied by most of the client population on the Internet, results in
worse performance than no destination address sorting at all.

Rule 9 might offer a significant advantage in private deployments which use non-IANA address allocation. In other cases, routing is typically
not hierarchical at all, so that the length of the matching address
prefix is meaningless.

Not true if the sites concerned are both using PA space and
happen to have the same ISP. We still have hopes that PA space will
be predominant in IPv6, despite the PI heresy.

Can you describe cases in which this rule is actively harmful?

   Brian



[Fred-Bakers-Computer-3:~] fred% traceroute mail.google.com
traceroute: Warning: mail.google.com has multiple addresses; using 209.85.199.18 traceroute to googlemail.l.google.com (209.85.199.18), 64 hops max, 40 byte packets
 1  10.32.244.217 (10.32.244.217)  1.570 ms  1.153 ms  1.104 ms
2 wsip-98-173-193-1.sb.sd.cox.net (98.173.193.1) 87.274 ms 49.568 ms 48.215 ms
 3  68.6.13.101 (68.6.13.101)  57.624 ms  47.359 ms  38.188 ms
 4  68.6.13.45 (68.6.13.45)  33.014 ms  32.715 ms  32.097 ms
5 paltbbrj02-ae0.0.r2.pt.cox.net (68.1.0.235) 51.490 ms 52.489 ms 41.434 ms 6 209.85.130.2 (209.85.130.2) 125.613 ms 122.277 ms 209.85.130.6 (209.85.130.6) 129.134 ms
 7  66.249.95.135 (66.249.95.135)  110.462 ms  124.537 ms  116.237 ms
 8  209.85.250.144 (209.85.250.144)  128.569 ms  130.422 ms  134.411 ms
9 64.233.174.131 (64.233.174.131) 145.547 ms 64.233.174.129 (64.233.174.129) 132.366 ms 118.576 ms
10  * 209.85.248.254 (209.85.248.254)  125.765 ms  138.378 ms
11  rv-in-f18.google.com (209.85.199.18)  118.186 ms  109.112 ms *
[Fred-Bakers-Computer-3:~] fred% traceroute
[Fred-Bakers-Computer-3:~] fred% nslookup 209.85.130.2
Server:         68.105.28.13
Address:        68.105.28.13#53

Non-authoritative answer:
*** Can't find 2.130.85.209.in-addr.arpa.: No answer

Authoritative answers can be found from:

[Fred-Bakers-Computer-3:~] fred% traceroute deneb.enyo.de
traceroute to deneb.enyo.de (212.9.189.171), 64 hops max, 40 byte packets
 1  10.32.244.217 (10.32.244.217)  1.528 ms  1.131 ms  1.126 ms
2 wsip-98-173-193-1.sb.sd.cox.net (98.173.193.1) 12.350 ms 10.628 ms 10.509 ms
 3  68.6.13.101 (68.6.13.101)  14.043 ms  13.115 ms  12.642 ms
 4  68.6.13.45 (68.6.13.45)  18.518 ms  17.599 ms  16.974 ms
5 paltbbrj02-ae0.0.r2.pt.cox.net (68.1.0.235) 33.845 ms 31.246 ms 31.796 ms 6 so-1-2-2.mpr2.sjc2.us.above.net (64.125.29.126) 31.804 ms 32.963 ms 29.991 ms 7 so-1-0-0.mpr1.lga5.us.above.net (64.125.26.142) 106.799 ms 104.833 ms 106.786 ms 8 so-0-0-0.mpr1.lga5.us.above.net (64.125.27.237) 107.908 ms 109.646 ms 107.201 ms 9 so-1-0-0.mpr1.ams1.nl.above.net (64.125.27.186) 192.188 ms 190.569 ms 192.356 ms 10 pos-9-1.mpr2.fra1.de.above.net (64.125.23.253) 206.446 ms 201.130 ms 201.177 ms 11 ge-9-7.er2b.fra1.de.above.net (64.125.23.210) 200.881 ms ge-9-4.er2b.fra1.de.above.net (64.125.23.206) 199.369 ms ge-9-7.er2b.fra1.de.above.net (64.125.23.210) 202.015 ms 12 c12gsr-5-02.intx.fra.interscholz.net (62.93.193.62) 201.813 ms 208.735 ms 198.146 ms 13 c12gsr-3-02.z10a.stg.interscholz.net (85.236.200.229) 205.121 ms 207.205 ms 204.092 ms 14 ge-knd-lfnet.z10.stg.interscholz.net (85.236.201.130) 204.035 ms 201.747 ms 209.954 ms
15  dsl-gw.ispeg.de (212.9.161.26)  202.581 ms  201.286 ms  200.045 ms
16  dsl.enyo.de (213.178.172.64)  227.865 ms  237.674 ms  228.145 ms
17  * *^C
[Fred-Bakers-Computer-3:~] fred% traceroute ops.ietf.org
traceroute to ops.ietf.org (147.28.0.62), 64 hops max, 40 byte packets
 1  10.32.244.217 (10.32.244.217)  1.320 ms  1.063 ms  1.030 ms
2 wsip-98-173-193-1.sb.sd.cox.net (98.173.193.1) 15.270 ms 14.536 ms 12.443 ms
 3  68.6.13.117 (68.6.13.117)  17.136 ms  13.310 ms  13.317 ms
 4  68.6.13.49 (68.6.13.49)  17.679 ms  22.767 ms  19.230 ms
5 paltbbrj01-so000.0.r2.pt.cox.net (68.1.0.199) 43.538 ms 41.442 ms 43.887 ms 6 ae-1.r20.snjsca04.us.bb.gin.ntt.net (129.250.5.33) 48.245 ms 42.201 ms 47.455 ms 7 p64-0-0-0.r20.sttlwa01.us.bb.gin.ntt.net (129.250.4.22) 72.543 ms 66.362 ms 81.955 ms 8 p16-0-0-0.r05.sttlwa01.us.bb.gin.ntt.net (129.250.3.11) 66.021 ms 72.685 ms 65.238 ms 9 p1-0.psg.sttlwa01.us.bb.gin.ntt.net (129.250.11.42) 69.071 ms 64.404 ms 64.297 ms
10  * *^C
[Fred-Bakers-Computer-3:~] fred% traceroute ietf.org
traceroute: Warning: ietf.org has multiple addresses; using 209.173.53.180
traceroute to ietf.org (209.173.53.180), 64 hops max, 40 byte packets
 1  10.32.244.217 (10.32.244.217)  1.581 ms  1.118 ms  1.095 ms
2 wsip-98-173-193-1.sb.sd.cox.net (98.173.193.1) 18.068 ms 12.854 ms 14.602 ms
 3  68.6.13.101 (68.6.13.101)  12.794 ms  11.780 ms  14.710 ms
 4  68.6.13.45 (68.6.13.45)  20.709 ms  28.882 ms  27.386 ms
5 fed1dsrj02-so010.0.rd.sd.cox.net (68.6.8.13) 18.876 ms 17.096 ms 22.599 ms 6 tbr1.la2ca.ip.att.net (12.122.104.38) 79.486 ms 79.698 ms 79.613 ms 7 tbr2.la2ca.ip.att.net (12.122.9.146) 80.148 ms 80.635 ms 84.841 ms 8 tbr2.sl9mo.ip.att.net (12.122.10.13) 82.377 ms 82.078 ms 96.295 ms 9 tbr1.sl9mo.ip.att.net (12.122.9.141) 91.041 ms 81.589 ms 78.615 ms 10 tbr1.wswdc.ip.att.net (12.122.10.29) 80.846 ms 76.828 ms 78.812 ms
11  12.123.8.193 (12.123.8.193)  91.766 ms  77.840 ms  77.159 ms
12  12.118.28.10 (12.118.28.10)  79.630 ms  82.614 ms  79.309 ms


-----BEGIN PGP SIGNATURE-----

iD8DBQFHnZH+bjEdbHIsm0MRAozqAKCCXuLKmV+8JiXtUCpxnKOITCxnVgCfYtud
i89+S4Fxu8roS5m85TwYw7o=
=7HEA
-----END PGP SIGNATURE-----