[AusNOG] News: Telstra to clamp down on peer-to-peer

Kris Price ausnog at punk.co.nz
Fri Feb 8 08:44:06 EST 2013


On 2/8/2013 9:56 AM, Phillip Grasso wrote:
>
> On 07/02/2013 1:20 PM, "Kris Price" <ausnog at punk.co.nz
> <mailto:ausnog at punk.co.nz>> wrote:
>  > When I buy a 100 Mbps service, I expect that I get to 100 Mbps. If
> the network upstream is congested, then I expect to get
> (my__paid_for_bw)/(num_cust*avg__paid_for_bw) share of that.
>
> I don't think what your asking for is either technically possible or
> economically possible at *scale*.
>

Asides from my example being overly simplified, can you explain what's 
not technically possible about that, say using H-QoS on a 7450/7750?

We do this often on broadband networks so I'm confused by your 
statement. Maybe we're talking about different points in the network 
we're talking about, I've assume we're talking at the broadband 
aggregation layer, and you're maybe thinking about in the core and 
transit -- where yes you basically provision enough and let stat muxing 
take care of the rest.

That aside, my example is in relation to my feelings on what it means to 
be neutral. You couldn't for example argue that Google Fiber was neutral 
if it's voice and video services were given any preferential QoS over 
other voice and video providers. (But Google Fiber is easy anyway, 
because it's not an open access network.)

In open access networks (e.g. NBN and UFB) this interests me, because 
it's not trivial to solve, but it does cause unfairness and other 
problems if a poor job is done of it.



More information about the AusNOG mailing list