[141389] in North American Network Operators' Group
Re: IPv6 day fun is beginning!
daemon@ATHENA.MIT.EDU (Jared Mauch)
Tue Jun 7 23:47:23 2011
From: Jared Mauch <jared@puck.nether.net>
In-Reply-To: <BANLkTin4yeOBtX3c4ZBHHTe0+iv_oNa7PQ@mail.gmail.com>
Date: Tue, 7 Jun 2011 23:46:25 -0400
To: Jorge Amodio <jmamodio@gmail.com>
Cc: NANOG list <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
On Jun 7, 2011, at 11:31 PM, Jorge Amodio wrote:
> Thanks for the link Jared.
>=20
> I wonder how many eye-balls are really enabled to reach the IPv6
> sites. Akamai's site doesn't show very impressive numbers, trying to
> figure why 300ms latency and >4% packet loss ?
My guess is it's over the entire set of akamai properties hosted there, =
so cisco, bing, etc.. that all point to edgesuite and their related =
domains.
The latency is likely due to suboptimal tunneling vs native. The =
density of IPv6 peering likely doesn't fully match the rest of the world =
either, sometimes you have to go across the country because someone =
can't do v6 on the local port.
I do also find it interesting there's not a significant spike at the =
AMSIX IPv6 sFlow page either.
http://www.ams-ix.net/sflow-stats/ipv6/
We have seen a traffic increase but nothing like what I was expecting, =
nay hoping to see. (i.e.: gigs and gigs of traffic - it does look like =
~2x to me in an unscientific eye-look at a chart).
Some of this may just be due to the methods used by the various sites to =
enable IPv6. (e.g.: main site only, not sub-sites, and not things like =
fbcdn etc).
There are people listed on the ISOC site that are not serving up AAAA =
records either, so perhaps they are doing last minute testing and we =
will see an increase as a result. It's still early to measure a final =
result obviously, but the observation part is quite interesting for me =
now. I do hope to see more traffic over the next 12-24 hours. Maybe =
the "asia peak" time will be most interesting=85.
- Jared=
