Re: [76attendees] A Net Neutrality comment

[Jason Livingood <[e-mail]>]

+1 on this being an interesting study.

As for Richard's comment, speaking from personal experience, involvement in
application throttling at any level of scale and customer diversity can at
best be described as "problematic."  ;-)

Jason

On 11/9/09 9:21 AM, "Richard Barnes" <[e-mail]> wrote:

> Hey Fred,
> 
> Thanks for this little study.  Note that this doesn't necessarily
> argue for *application* throttling, as much as for *user* throttling.
> The network might want to prevent the bittorrent user from interfering
> with you, but the he can be left free to shut down his own VPN if he
> wants to.
> 
> --Richard
> 
> 
> 
> On Mon, Nov 9, 2009 at 9:05 AM, Fred Baker <[e-mail]> wrote:
>> A remark to those who take a militant libertarian view of Net Neutrality,
>> and those of ledbat and tcpm who have difficulty understanding why
>> transports should tune to the knee (just enough of data outstanding, aka
>> cwnd, to maintain the maximum goodput) as opposed to the cliff (the knee
>> plus the maximum depth of the bottleneck queue, at which point throughput
>> has not increased but loss has increased).
>> 
>> Saturday night, as I do many nights that I spend at hotels, I ran a ping
>> study to characterize the network. It was obviously massively
>> overprovisioned - it was difficult to register RTT variance in excess of a
>> millisecond trans-pacific between Japan and the US. I did this again last
>> night. The network behavior as measured from my room was equally stable
>> until about 11:58 PM; at that point, someone fired up something huge, my
>> guess being something that uses bittorrent, delay dramatically increased,
>> and my VPN went down within a couple of minutes. When this happens,
>> customers call ISPs and ISPs start throttling applications, because the
>> applications are doing horrible things to the ISPs' customers.
>> 
>> The attached are a case in point.