[AusNOG] What does an IPv6 broadband service look like?

[Kevin Littlejohn]

On 03/08/2008, at 20:28 , Mark Newton wrote:

>
> On 03/08/2008, at 7:32 PM, Adrian Chadd wrote:
>>
>> * Some kind of dhcp/ddns based name update scheme?
>
> I think it'll be this one.
>
> It's a fascinating issue that nobody has even started working
> on yet, isn't it? :-)

That, and billing - I missed the start of the conversation, but is  
anyone actually offering native IPv6 atm, and if so, are you billing  
per Mb for it (and if so, how? :)?

So far, the edge devices we deal with (SCE, PacketLogic, Packeteer,  
BlueCoat) don't, AFAIK, support IPv6.  The RADIUS-based ones (SSG/ISG/ 
802.1x etc), one would assume, will just continue to report bulk  
traffic across v6 the same way they do across v4 (I haven't tested  
that, tho).  Similarly anything pipe-centric like SNMP, if anyone's  
actually still using that for billing - but RADIUS and SNMP share the  
same problems for tracking _what_ the traffic is.

I also don't know how the app-layer inspection devices (see list above  
mainly) deal with IPv6 traffic - I'm fairly certain most of them don't  
grok it, therefore may not play nicely if you're filtering or proxying  
or shaping.

This would seem to be a particular issue also if you're billing, say,  
on-net versus off-net as the universities do (or peering), where you  
need to track different classes of traffic based on source/dest.

Vendors don't seem to be showing a huge interest in moving to solve  
this quite yet, and at that "every bit counts" level of speed they're  
going for with transparent inspection devices, I'd imagine it's not  
necessarily an easy fix for some of them.

KevinL