On Tue, Jun 28, 2005 at 10:38:06AM -0700, Marc Singer wrote:
> On Tue, Jun 28, 2005 at 12:05:58PM +0200, Jeroen Massar wrote:
> > On Mon, 2005-06-27 at 15:27 -0400, Michael Richardson wrote:
> >
> > > Yes, that's right.
> > > If you can't get native IPv6, 6to4 is better than tunnels.
> >
> > Don't forget that 6to4 is also a tunnel ;)
> > Actually the most significant difference is that 6to4 makes an automatic
> > tunnel to the remote 6to4 site, one huge problem though, you can't
> > easily reach RIR space because you need an upstream out of 2002::/16 for
> > that.
> I thought that is what the endpoint was for, 192.88.99.1. At least,
> that's how Craig Small describes it.
> <http://people.debian.org/~csmall/ipv6/setup624.html>
The problem with 192.88.99.1 is it depends on someone nearby offering
up a 6to4 gateway, and advertising it, and the advertisements getting
to you.
From my machines in Canberra, Australia, the nearest anycast node
on 192.88.99.1 is in Switzerland, 300ms away.
So I run 6to4 for talking to other 6to4 machines, but run a tunnel
(and use _it_ in DNS as it's staticly addressed) to broker.aarnet.net.au,
which is good if you're in Australia.
My old ADSL connection through Comindico used to have 300ms to the
tunnel broker (hence my addition of 6to4) but now I get 40ms through
my new TransACT TalkWEB going out via MCI.
Anyway, below are various traceroutes, if you're interested. ^_^
Trace from Yurika (6to4 only, connected via Comindico) to my home
machine (MCI)'s tunnel address.
yurika:~# tracepath6 2001:388:f000::2d
1?: [LOCALHOST] pmtu 1472
1: swiIX1.switch.ch 323.440ms
2: swiEZ2-10GE-1-3.switch.ch 322.697ms
3: swiLS2-10GE-1-1.switch.ch 326.904ms
4: swiCE2-10GE-1-3.switch.ch 326.615ms
5: switch.ch1.ch.geant.net asymm 6 327.496ms
6: ch.fr1.fr.geant.net asymm 7 335.113ms
7: fr.uk1.uk.geant.net asymm 5 377.503ms
8: uk.ny1.ny.geant.net asymm 6 447.485ms
9: nycmng-geant.abilene.ucaid.edu asymm 6 359.103ms
10: chinng-nycmng.abilene.ucaid.edu asymm 5 329.737ms
11: iplsng-chinng.abilene.ucaid.edu asymm 4 331.258ms
12: kscyng-iplsng.abilene.ucaid.edu asymm 5 341.049ms
13: dnvrng-kscyng.abilene.ucaid.edu asymm 6 350.189ms
14: snvang-dnvrng.abilene.ucaid.edu asymm 7 374.841ms
15: 3ffe:80a::b1 asymm 4 326.175ms
16: gigether0-2-7.bb1.a.syd.aarnet.net.au asymm 3 344.748ms
17: broker1.a.syd.aarnet.net.au asymm 4 327.108ms
18: broker1.a.syd.aarnet.net.au asymm 4 326.244ms pmtu 1280
18: 2001:388:f000::2d asymm 1 392.101ms reached
Resume: pmtu 1280 hops 18 back 1
Trace from Yurika (6to4 only, connected via Comindico) to my
home machine (MCI)'s 6to4 address.
yurika:~# tracepath6 2002:ca06:27d1:f000::1
1?: [LOCALHOST] pmtu 1472
1: 2002:ca06:27d1:f000::1 59.259ms reached
Resume: pmtu 1472 hops 1 back 1
As you can see, the difference is staggering (and before I went to 6to4, I had
to use ssh -4 to avoid frustration when typing).
Here's a traceroute from keitarou (the home machine) to 192.88.99.1
tbble@keitarou:~$ traceroute 192.88.99.1
traceroute to 192.88.99.1 (192.88.99.1), 30 hops max, 38 byte packets
1 10.100.100.254 (10.100.100.254) 13.554 ms 11.512 ms 10.841 ms
2 10.100.100.97 (10.100.100.97) 27.980 ms 12.429 ms 11.989 ms
3 10.10.0.1 (10.10.0.1) 16.075 ms 11.989 ms 13.440 ms
4 fastethernet1-0.hsrp77.bdr2.nort-cbr.officelink.net.au (203.10.77.4) 11.913 ms 12.227 ms 12.479 ms
5 static-146-81.transact.net.au (202.55.146.81) 13.968 ms 13.233 ms 12.924 ms
6 static-144-17.transact.net.au (202.55.144.17) 13.944 ms 12.777 ms 14.843 ms
7 502.ATM2-0-0.GW2.CBR2.ALTER.NET (210.80.144.249) 14.448 ms 15.791 ms 13.755 ms
8 322.AT-3-0-0.XR2.CBR2.ALTER.NET (210.80.32.205) 15.275 ms 19.211 ms 15.351 ms
9 0.so-0-0-0.XR2.SYD2.ALTER.NET (210.80.33.49) 18.168 ms 36.084 ms 25.517 ms
10 0.SO-6-2-0.TR2.SYD2.ALTER.NET (210.80.51.109) 22.033 ms 21.984 ms 26.283 ms
11 0.so-3-1-0.IR2.LAX12.Alter.Net (210.80.49.162) 223.702 ms 192.603 ms 199.159 ms
12 POS3-0.IR2.LAX9.ALTER.NET (137.39.31.242) 178.101 ms 174.524 ms 177.255 ms
13 0.so-5-2-0.TL2.LAX9.ALTER.NET (152.63.0.150) 182.744 ms 188.158 ms 185.949 ms
14 0.so-1-0-0.XL2.LAX9.ALTER.NET (152.63.115.149) 186.133 ms 184.306 ms 188.492 ms
15 POS7-0.BR3.LAX9.ALTER.NET (152.63.115.5) 182.189 ms 197.560 ms 199.407 ms
16 so3-0-0-622M.ar1.LAX2.gblx.net (208.51.134.1) 273.343 ms 182.909 ms 181.928 ms
17 so4-0-0-2488M.ar2.CDG2.gblx.net (67.17.65.90) 326.416 ms 327.085 ms 337.697 ms
18 swiCE2-switch-ch.so-0-1-0.ar2.CDG2.gblx.net (64.212.70.62) 353.985 ms 332.120 ms 327.864 ms
19 swiZH2-10GE-1-1.switch.ch (130.59.36.2) 358.043 ms 346.236 ms 351.654 ms
20 swiIX1-10GE-1-3.switch.ch (130.59.36.129) 338.936 ms 337.319 ms *
And from Yurika:
tbble@yurika:~$ traceroute 192.88.99.1
traceroute to 192.88.99.1 (192.88.99.1), 30 hops max, 38 byte packets
1 203.221.207.65 (203.221.207.65) 0.567 ms 0.495 ms 0.548 ms
2 fe2-0.wsr01-nort-cbr.comindico.com.au (203.220.47.65) 231.026 ms 258.791 ms 1.839 ms
3 ge3-0.bdr02-kent-syd.comindico.com.au (203.194.1.121) 5.621 ms 5.771 ms 5.273 ms
4 134.159.123.245 (134.159.123.245) 7.100 ms 6.041 ms 6.369 ms
5 i-5-0.syd-core02.net.reach.com (202.84.144.249) 9.373 ms 7.381 ms 7.298 ms
6 i-0-0.wil-core02.net.reach.com (202.84.144.101) 155.798 ms 156.700 ms 156.479 ms
7 202.84.251.166 (202.84.251.166) 156.819 ms 156.571 ms 155.961 ms
8 so-6-1.LosAngeles1.Level3.net (4.68.127.237) 162.146 ms 161.519 ms 162.412 ms
9 ae-1-55.bbr1.LosAngeles1.Level3.net (4.68.102.129) 164.755 ms 162.360 ms 162.258 ms
10 as-1-0.bbr1.Frankfurt1.Level3.net (212.187.128.62) 314.156 ms 314.303 ms 313.746 ms
11 so-1-0-0.mpls2.Geneva1.Level3.net (212.187.128.245) 323.843 ms so-3-0-0.mpls1.Geneva1.Level3.net (212.187.128.241) 324.569 ms so-1-0-0.mpls2.Geneva1.Level3.net (212.187.128.245) 323.865 ms
12 ge-5-0.hsa2.Geneva1.Level3.net (213.242.72.20) 323.825 ms ge-4-0.hsa2.Geneva1.Level3.net (213.242.72.4) 324.344 ms ge-5-0.hsa2.Geneva1.Level3.net (213.242.72.20) 322.920 ms
13 213.242.73.138 (213.242.73.138) 323.407 ms 323.118 ms 324.618 ms
14 swiLS2-10GE-1-3.switch.ch (130.59.37.2) 325.540 ms 326.328 ms 324.342 ms
15 swiEZ2-10GE-1-1.switch.ch (130.59.36.206) 319.935 ms 320.510 ms 319.857 ms
16 swiIX1-10GE-1-1.switch.ch (130.59.36.250) 320.541 ms 320.018 ms *
So I hope this demonstrates that the best speed comes from running 6to4, and
your default route (2000::/3) should either be 6to4 to 192.88.99.1 or a tunnel,
depending on relative distances. Unless your upstream provides IPv6
connectivity, in which case you can hope _they_ do the above. ^_^
--
-----------------------------------------------------------
Paul "TBBle" Hampson, MCSE
8th year CompSci/Asian Studies student, ANU
The Boss, Bubblesworth Pty Ltd (ABN: 51 095 284 361)
Paul.Hampson@Anu.edu.au
"No survivors? Then where do the stories come from I wonder?"
-- Capt. Jack Sparrow, "Pirates of the Caribbean"
This email is licensed to the recipient for non-commercial
use, duplication and distribution.
-----------------------------------------------------------
Attachment:
pgpIePhq93uBL.pgp
Description: PGP signature