Re: [NNagain] FCC NOI due dec 1 on broadband speed standards

[Sebastian Moeller <moeller0@gmx.de>]

Hi Jeremy,


> On Nov 14, 2023, at 12:58, Jeremy Austin via Nnagain <nnagain@lists.bufferbloat.net> wrote:
> 
> 
> 
> On Tue, Nov 14, 2023 at 6:46 AM Livingood, Jason via Nnagain <nnagain@lists.bufferbloat.net> wrote:
> On the subject of how much bandwidth does one household need, here's a fun stat for you.
> 
>  
> 
> At the IETF’s 118th meeting last week (Nov 4 – 10, 2023), there were over 1,000 engineers in attendance. At peak there were 870 devices connected to the WiFi network. Peak bandwidth usage:
> 
> 	• Downstream peak ~750 Mbps
> 	• Upstream ~250 Mbps
> 
> How was this calculated? That's an unusually high ratio of up to down, so my suspicion is that they aren't time correlated; they're also not /normal/ or /evening/ peaks, I'm expecting.

	[SM] Given that this is from a conference network, I think it is expected that the pattern does not match typical end-user traffic, no?


> 
> There's a big difference between individual peaks of upload and aggregate peaks of upload; most people aren't streaming high symmetric bandwidth simultaneously.

	[SM] Which is good, at current rates of over-subscription (or under-provisioning for the glass half empty folks) such a shift in usage behavior likely would not result in happiness all around?


> Consequently a peak busy hour online load, I'm finding, is still much more like 8:1 over all users (idle and active), in Preseem's data set.

	[SM] But this is end users that have been trained over decades to operate on heavily asymmetric links and that adjusted their usage patterns to match the "possible" while we might expect the network experts at IETF to have different expectations? (Then again these folks likely also are users of normal home internet links).


> 
> In addition to speed tests being, like democracy, the worst form of government except for all the others that have been tried, it would be instructive both for end users and ISPs to choose, agree on and understand specific percentiles of expected performance at idle and at peak busy hour.

	[SM] That will not be easy to achieve.


> Has anyone solved the math problem of distinguishing (from outside) a constraint in supply from a reduction in demand?

	[SM] Keep in mind that the former can cause the latter, if connectivity/responsiveness is too bad, people might shift to do other things there y reducing the measurable load, but not the conceptual demand (as they might prefer a workable internet access).

Regards
	Sebastian


> 
> Jeremy
> 
> 
> _______________________________________________
> Nnagain mailing list
> Nnagain@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/nnagain