Dual stack statistics?

[Tim Chown]

On 28 Jan 2013, at 14:58, Tore Anderson <tore at fud.no> wrote:
> 
> Using Chrome by any chance? If so there's a perfectly good explanation
> for it, map46.cisco.com's IPv6 connectivity quite simply sucks. This is
> on native v6 in Norway:
> 
> tore at envy:~$ nping -4 --tcp-connect -c 5 6lab.cisco.com | grep rtt
> Max rtt: 47.510ms | Min rtt: 41.501ms | Avg rtt: 44.245ms
> tore at envy:~$ nping -6 --tcp-connect -c 5 6lab.cisco.com | grep rtt
> Max rtt: 370.014ms | Min rtt: 368.259ms | Avg rtt: 368.913ms

Ouch.  A trace from here indicates similar. It doesn't respond to pings.

$ traceroute6 6lab.cisco.com
traceroute to 6lab.cisco.com (2001:420:81:101:0:c:15c0:4664), 30 hops max, 40 byte packets
 1  2001:630:d0:f102::2 (2001:630:d0:f102::2)  0.372 ms  0.450 ms  0.503 ms
 2  zaphod.6core.ecs.soton.ac.uk (2001:630:d0:f001::1)  0.136 ms  0.125 ms  0.113 ms
 3  internal-router.6core.ecs.soton.ac.uk (2001:630:d0:f000::2)  4753.037 ms  4753.043 ms  4753.037 ms
 4  2001:630:d0:f000::3 (2001:630:d0:f000::3)  94.469 ms  95.777 ms  102.027 ms
 5  2001:630:c1:113::1 (2001:630:c1:113::1)  102.010 ms  101.994 ms  101.977 ms
 6  2001:630:c1:4b::a (2001:630:c1:4b::a)  103.152 ms  104.486 ms  105.779 ms
 7  so1-1-0.bris-sbr1.ja.net (2001:630:0:8041::5)  107.773 ms  107.819 ms  109.194 ms
 8  so1-1-0.warr-sbr1.ja.net (2001:630:0:10::28d)  198.367 ms  120.025 ms  119.988 ms
 9  * * *
10  * ae0.lond-gw-ixp4.ja.net (2001:630:0:10::152)  122.062 ms  121.954 ms
11  10gigabitethernet1-1.core1.lon1.he.net (2001:7f8:4::1b1b:1)  129.714 ms  126.693 ms  123.809 ms
12  10gigabitethernet7-4.core1.nyc4.he.net (2001:470:0:128::1)  190.271 ms  178.730 ms  175.673 ms
13  10gigabitethernet8-3.core1.chi1.he.net (2001:470:0:1c6::2)  193.944 ms  190.353 ms  190.406 ms
14  10gigabitethernet3-2.core1.den1.he.net (2001:470:0:1af::2)  219.056 ms  219.041 ms  218.436 ms
15  10gigabitethernet11-4.core1.sjc2.he.net (2001:470:0:1b4::1)  252.044 ms  241.729 ms  238.265 ms
16  10gigabitethernet3-2.core1.pao1.he.net (2001:470:0:32::2)  250.943 ms * 10gigabitethernet3-2.core1.pao1.he.net (2001:470:0:32::2)  257.704 ms
17  * * *
18  2001:420:80:8::2 (2001:420:80:8::2)  259.873 ms  226.644 ms  229.099 ms
19  2001:420:81:100::2 (2001:420:81:100::2)  412.994 ms  408.821 ms  423.204 ms
20  map46.cisco.com (2001:420:81:101:0:c:15c0:4664)  428.284 ms  426.115 ms  378.611 ms

> 300ms is AFAIK the cut-off time when Chrome's Happy Eyeballs
> implementation decides it's not worth waiting for an IPv6 connection to
> establish any longer.

Whatever Chrome's threshold is, it's probably no more than 300+ ms.

The IPv4 instance appears to be in Europe based on its responses.

Tim