Re: Drop in IPv6 traffic

[Marco Hogewoning <marcoh () marcoh net>]

Hi Patrick ,

On 9 jul 2009, at 16:10, Patrick W. Gilmore wrote:

> On Jul 9, 2009, at 9:58 AM, michiel.muhlenbaumer () atratoip net wrote:
>
> > Hi Jeroen & others,
> >
> > Yep, looks like we are doing a great portion of AMSIX's IPv6  
> > traffic and
> > our (free) IPv6 service was affected because of an internal error  
> > last
> > night around 00.30 am.
>
> Michiel,
>
> Thank you for the information.  Could you let us know if XS4All's  
> free v6 news feed went to zero, or was just dropped by some  
> percentage?

Please note Michiel isn't from XS4ALL :)

We (XS4ALL) did not have any outage on the usenet service, everything  
is still running and passing traffic. The only thing we did was adjust  
some filters so if you are depend on more specifics and not announcing  
the /32 you might have been dropped from the table around 15:00 GMT.

The remaining traffic at ams-ix matches our internal graphs, so I  
guess I'm the only bulk sender at the moment.

> I ask because the AMS-IX is frequently used as an example that v6 is  
> being heavily adopted.  If it is all one source for one application,  
> that is important information to the people fighting for v6  
> adoption.  Going from peaks of 1.4 Gbps to 0.4 Gbps is impressive.   
> If that 0.4 Gbps still includes some of your traffic, it is very  
> impressive.

There are 3 open usenet servers at amsterdam that I know off:

- XS4ALL (at newszilla6.xs4all.nl)
- Highwinds/Eweka (news.ipv6.eweka.nl)
- XSnews

I think we, once again, have proof 99% of the ams-ix traffic. The  
remaining bit possibly is google and the traffic on this mailinglist  
(and ripe.net).

As an exercise we could see if spikes in nanog activity can be matched  
to spikes in the v6 traffic :)

Groet,

MarcoH