On Sat, Feb 02, 2019 at 04:50:45PM -0800, Russell Senior wrote:
> There is fiber in Portland now. I have gigabit fiber to my house (since
> 2015). Most of the footprint of Portland has it available right now. How
> complex was that to install? Not so complex that they didn't manage to do
> it. 

THEY did it, WE did not.  Expertise has its rewards.

Years ago, two partners and I started an integrated circuit
company, I-Cube Design Systems, which designed and sold
"non-blocking crossbar routing" chips. 

Think of a grid of wires, with switches between every wire
and every other wire.  Our "secret sauce" pushed signals 
through the grid more quickly, switched configurations
more rapidly, and did all this with a quarter of the power
and a tenth of the price of our nearest competitor.

Our routing chips (the largest was 640 "wires", 408,960
switch points) were designed for logic simulation, then
television studio equipment.  Sales skyrocketed when the
public internet appeared. 

For "N" endpoints globally, there are "N squared" potential
connections.  For those who understand logarithms, that 
implies that the cost of the network goes up as the square
of the number of endpoints, and that if the monthly growth
of the internet is "X" percent new endpoints per month, the
number of switches grow as "2X" percent per month. 

In reality, the switches are arranged in "trees", juggling
routes through the switches to maximize the re-use of long
distance fiber, combining low speed signals into high speed
signals.  That reduces cost but increases latency, and has
inherent "net neutrality" consequences; most of the points
are at the edge of the connection graph, not the center. 
The center is owned by the highest bidders.

So our growth was a more complex function, smaller baseline
but still roughly 2X per month.

Because we outperformed and underpriced our competitors, we
put them all out of business, including "Aptix", a company
founded by Bay Area venture capitalist Vinod Khoshla.  
Handy Life Planning Tip; do NOT get on the wrong side of
Vinod Koshla.

However, before the elephant rampaged, we were earning 75%
profit on sales, and growing at 37% per quarter.  I was
guiding dozens of engineers.  Our customers, like Cisco,
wanted extremely rapid execution from design concept to
production parts, and we delivered. 

However, our principal investor, Draper Associates, wanted
us to diversify ... into ethernet chipsets.  Those were
(and are) low margin, with eight established competitors. 
We expected there would be three after the shakeout, and
the three would not include ethernet newbies like us.  

The diversion of income and engineering resources slowed
our response to our main customers.  They hired their own
engineers and designed their own chips.  We were out of
business in a year.

------

And what the hell does this have to do with Russell's
discussion of citywide fiber?  The internet does not just
connect your endpoint to the endpoint down the street, or 
in the city, or in your county;  it can connect to any
other IP address in the world, potentially 4 billion
other addresses, and 16 quintillion switch points.   Move
a node in the graph, and you've moved 4 billion virtual 
switchpoints.  Double your bandwidth, and you double the
virtual bandwidth of 4 billion switchpoints.  You aren't
just buying a fiber and a port on one switch, you are
buying a piece of a planetary-scale grid.

It is WAY more complicated and nuanced than that, but you
can imagine how and why the fiber from your house to the
neighborhood router is just the tip of the iceberg.  When
you pay your monthly internet fee, you aren't just paying
for your fiber, you are paying to increase the capacity of
the ENTIRE INTERNET.  And that isn't just pouring magic
pixie dust on all the existing switches; it is removing and
obsoleting slower equipment, and replacing it with new and
very expensive faster equipment.  Office parks with dozens
of aircraft-hangar-sized buildings full of thousands of
racks of equipment and megameters of optical fiber.

"Tubes - A Journey to the Center of the Internet" by
Andrew Blum is a decent general-audience book about this,
available in dead tree libraries near you.

-----

The internet isn't like a telephone.  My infancy was in
a house with a "party line";  one pair of wires serving
multiple homes, with distinctive ring telling us who the
call was for.  Long distance calls outside of the
Portland area cost many dollars per hour.  No, we did
not power the phones with whale oil.

Decades later, modern electronics and fiber optic networks
implemented for other reasons (SPRINT = Southern Pacific
Railroad Internal Network Telecommunications) became the
new backbone.   We can move a hell of a lot of bits through
a fiber with modern electronics, though we are nearing
physical limits for fiber and endpoints. 

But we still consume voice connections locally; very few of
our calls are out-of-state or international.  We "know"
that "voice long distance is expensive" (it isn't, a two-
way voice channel peaks at 0.000128 gigabit/sec), but we
don't understand that a data channel has similar economic
constraints.  Besides light traffic local like PLUG,
almost all of my traffic beyond my house network is global. 
That works very differently than my childhood training.

Expectations formed in the age of landline telephones are
very misleading.  Politics based on those expectations
verge on insanity.

-----

BTW, if you want to slash your telecom costs, consider Ooma
VOIP.  I spend something like $200 per year for two local
numbers and free US-wide long distance, with best-available
spam call blocking.  The Ooma box connects to an open port
on the Comcast router, and to the POTS lines in the house.
The voice data is encrypted - until it connects to the 
public switched telephone network.  

I dream of the neighborhood wifi advocates building on this
idea, providing "alternate cell network" service to their
neighbors, especially phone data service.  Many people
spend far more on their cell phone data plan than they do
for their vastly more capable fiber internet.  

If we can somehow invert this, we can spend far less money
on those far wealthier fatcats.  After all, the reason that
Verizon sold their pacific northwest telephone and internet
franchise to incompetent Frontier is that the profits are
higher and easier to sustain with cell data service.  The
cell moguls throttle bandwidth to the voice channel, which
makes voice cell phones almost impossible to use for the
hearing-impaired, like me.  That increases data channel
use, and unregulated profits.

We CAN fix this much larger problem.

Populist advocacy is great, but not when it is misinformed
and misapplied.  That puts control directly into the hands
of corporate spinmeisters.  Truth walks, lies have wings.

Keith

-- 
Keith Lofstrom          [email protected]
_______________________________________________
PLUG mailing list
[email protected]
http://lists.pdxlinux.org/mailman/listinfo/plug

Reply via email to