[186965] in North American Network Operators' Group
Re: Binge On! - get your umbrellas out, stuff's hitting the fan.
daemon@ATHENA.MIT.EDU (Owen DeLong)
Mon Jan 11 13:42:20 2016
X-Original-To: nanog@nanog.org
From: Owen DeLong <owen@delong.com>
In-Reply-To: <CABNB40U+QeUiBQqyyZ6EwBfDsHXMFJ+OOh8B+GinJApc9=YGew@mail.gmail.com>
Date: Mon, 11 Jan 2016 10:40:09 -0800
To: Jeremy Austin <jhaustin@gmail.com>
Cc: North American Network Operators' Group <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org
> On Jan 11, 2016, at 10:31 , Jeremy Austin <jhaustin@gmail.com> wrote:
>=20
>=20
>=20
> On Mon, Jan 11, 2016 at 9:15 AM, Owen DeLong <owen@delong.com =
<mailto:owen@delong.com>> wrote:
>=20
>>=20
>>=20
>>=20
>> This is similar to Hughesnet's FAP (unfortunately named Fair Access =
Policy).
>>=20
>> I've had some consumer success with this model. There are other =
fairness models that can augment it, however; it's not my favorite.
>=20
> What is your favorite?
>=20
> Does a dog have the Buddha nature?
>=20
> My favorite is actually having enough bandwidth to meet demand. What a =
concept. Ought to work for terrestrial; where we run out of =
spectrum/bandwidth is in shared-medium last-mile.=20
That=E2=80=99s not a billing model=E2=80=A6 We were talking about =
billing models.
What=E2=80=99s your favorite billing model?
> Pre-Title II classification, I had excellent success with per-flow =
equalization/fairness, but this is expensive and makes bandwidth =
guarantees difficult to manage.=20
>=20
> After, I've also had success with a) maintaining sane oversubscription =
ratios and b) using per-customer-class fairness balancing, and c) some =
experimentation with FQ-CODEL, although this is less neutral and still a =
gray area =E2=80=94 at least until I understand it better.
Again, we are apparently talking apples and oranges. I=E2=80=99m talking =
about billing models and you=E2=80=99re talking about service delivery =
techniques.
> However, as I said, I consider everything to the right of AYCE on your =
=E2=80=9Ccontinuum=E2=80=9D to be simply variations of usage-based =
billing.
>=20
> Sure, to a consumer who stays within their usage tier, their tier =
looks like AYCE (until it doesn=E2=80=99t), but it certainly isn=E2=80=99t=
actually.
>=20
> I agree.
> =20
>=20
>>=20
>>=20
>> How much uncapped LTE spectrum is needed before we can hit that 2Mbps =
per customer referred to recently?
>=20
> I would assume quite a bit. There are 7 billion potential subscribers, =
so that=E2=80=99s 14 billion Mbps or 14 Petabits per second world wide.
>=20
> Heh. Gary said it better =E2=80=94 it's about user density. All 7 =
billion aren't on one set of sectors.
>=20
> The architecture for "repeaters", as Gary pointed out, is suboptimal, =
which is why we rely so heavily on Wifi, and why the WISP world is up in =
arms over LTE-U. Or so it seems to me.
>=20
> And NYC is just now getting wifi in the tunnels?
>=20
> I apologize if this has grown off-topic.
Meh, most useful threads wander significantly.
Owen
