[GRLUG] FiOS -- NOT

[Adam Tauno Williams]

On Sun, 2010-03-28 at 16:00 -0400, John-Thomas Richards wrote:
> On Sun, Mar 28, 2010 at 02:58:56PM -0400, Bob Kline wrote:
> > On Sun, Mar 28, 2010 at 1:38 PM, Adam Tauno Williams
> > <awilliam at whitemice.org
> [snip]
> > > It would be very interesting to see real-world benchmarking of such
> > > services;  but I've not seen any [you'd need to connect each service
> > > to the same place - which would be expensive - otherwise your tests
> > > wouldn't be comparative].  On the other hand I'm pretty confident
> > > that paying for "50MBps" would land squarely in the "waste of money"
> > > category as I doubt you will get any substantial improvements of
> > > real-world performance once you past 10MBps.  You rather quickly run
> > > into constraints [very possibly administrative] on the remote.
> > Why 10 Mbps?
> > It's chicken and egg.

Or it isn't.  There is diminishing return to increased throughput once
you reach a certain point.  Ten years ago if you had asked me: "What
would you like to use the Internet for that you currently can't?"  I
would have a had a list.  Ask me that question today, and.... Uhm... I
can't think of anything.

>   In 1980 I had a 300 bps modem.  A "Cat," with
> > the handset cradle.  Today I have an actual 16 Mbps - under many
> > conditions anyway.  More bandwidth has always found uses, and nothing
> > I know says some kind of optimum speed has been passed already.

Or does it?  That some [not nearly as large as promoted, but not
insignificant] number of users have migrated to netbooks and the like
indicates that there is certainly a diminishing of the Moore's law
effect.  I have a Core i7 laptop I use for most of my work - a latest
and fastest machine.  But I also have a P4 desktop [which I'm using
right now]... that works just find for 85% of what I have to do
[Evolution, Firefox, & Monodevelop].  Would a faster computer add much
of anything to that experience?  No.  Would a faster Internet
connection? No.  Is there really anything else I want to do that I can't
now?  No.  Once you saturated the need...  that saturating the need took
some time [a decade?] certainly does not imply that the need is
insatiable.

> I have Comcast's basic internet service.  I regularly see download
> speeds around 10Mbps.  It really is fast enough.  Both my wife and I can
> watch hulu.com (at the same time) without a problem.  I can download a
> ~680MB iso in a few minutes time. 

Ditto.

> indicates the optimum speed has been passed.  Streaming hi-def?  Some
> day, but that's not how people are using the web.  Ten years ago many
> were still transitioning from dial-up to broadband.  Dial-up was
> insufficient based on today's usage.  Most people use the internet to
> surf the web (including the social networking sites), send email, chat,
> and, occasionally, download a show/watch tv.  Ten Mbps seems adequate.

Ditto.

> You could argue that it's simply chicken and egg.  I wonder how Korea
> and Japan uses its 100Mbps connections.  Based on the exchange students
> from Korea and Taiwan we've hosted, my guess is it's very similar to how
> we use our 16Mbps connections, which means the speed is overkill.

My observation as well.  Of course, over there everything is already HD
and has been for awhile.  But it is also IPv6 which significantly
increases the efficiency of streaming and large file transfer.  IPv6
certainly lets them get more actual throughput from their pipe
throughput than IPv4 would permit us.

>   It's
> like a car.  Sure, the speedometer indicates it can go 120mph—and maybe
> it can.  But how often do you need your car to go that fast?

Yep.