Re: quality of v6 connections [Re: unmanaged scope comments]

[Tim Chown <tjc@ecs.soton.ac.uk>]

On Tue, Sep 24, 2002 at 10:19:18AM -0400, Marc Blanchet wrote:
> 
> sorry, but in the networks I'm controlling or involved, there is a clear
> distinction between the two:
> - production = 2001::/16, reliable connections, v6 native if possible
> - test/experiment = 3ffe::/16, experimental code from vendors, testing new
> mechanisms, etc....

and how do you distinguish these prefixes in terms of peering and transit
policies?   

performance isn't just down to which prefix is used, e.g. i am here in
tilab's network in turin and have a 2001: route all the way back to
my home network in the uk:

>tracert6 www.ipv6.ecs.soton.ac.uk

Tracing route to seven.ecs.soton.ac.uk [2001:630:d0:200:a00:20ff:feb5:ef1e]
from 2001:6b8:10:1600:8157:1234:8972:6034 over a maximum of 30 hops:

  1        3 ms     3 ms     3 ms  2001:6b8:10:1600:207:85ff:fea8:6f61
  2        4 ms    60 ms     4 ms  2001:6b8:10:1700:201:2ff:fed8:3b0b
  3        *        *        *     Request timed out.
  4        8 ms     8 ms     7 ms  6bone-gw4.ipv6.cselt.it [2001:6b8:0:100::6]
  5        8 ms     9 ms     8 ms  6bone-gw1.ipv6.cselt.it [2001:6b8:0:100::2]
  6      230 ms   297 ms   200 ms  nl-ams-re-02-t-7.ipv6.aorta.net [2001:730::1:
8]
  7      446 ms   349 ms   326 ms  2001:630:0:1::1e
  8      294 ms   256 ms   402 ms  2001:630:d0:200:a00:20ff:feb5:ef1e

[excuse the broken reverse dns, which i guess should work on a
"production" network]

the response time via the v4 path is much better.   

it depends what the operators are doing with tunnels so i think pekka's
point is the mesh mess, which is annoying when you try to use ipv6 for
day to day use, especially for interactive ssh access.

tim