last mile, regulatory incentives, etc (was: att fiber, et al)

Mailman List Mirror (Read Only)
1

This sharing can be done at a layer-3 or as you say at the time slot level or lambda level. It's no different than what is happening with the copper already. It's not like they have to give it away for free. They just have to offer it to other carriers at cost. This will hopefully provide more of a competitive market. But I don't see Verizon giving into it, nor Comcast or any other provider that has fiber. Verizon campaigned hard to have fiber removed from the equal access legalize so like most of these other large companies, they don't want to share their new toy with the other children.

-John

2

If it's done on a box owned by the incumbent then sharing has evolved into
giving away free service to competitors. It's different when copper pairs
into a house could be latched onto anyone's switch. Once you start
requiring a carrier to give away capacity in it's network that's
different. Also, diversity/redundancy becomes dodgy at this point. Not
that the billions of dollars they are making didn't come into the
discussion, but it seems like its more complicated to share fiber access
than it was to share copper pairs.

3

Oh, it's *much* worse than that, John.

The *right*, long term solution to all of these problems is for
municipalities to do the fiber build, properly engineered, and even
subbed out to a contractor to build and possibly operate...

offering *only* layer 1 service at wholesale. Any comer can light up
each city's pop, and offer retail service over the FTTH fiber to that
customer at whatever rate they like, and the city itself doesn't offer
layer 2 or 3 service at all.

High-speed optical data *is* the next natural monopoly, after power
and water/sewer delivery, and it's time to just get over it and do it
right.

As you might imagine, this environment -- one where the LEC doesn't own
the physical plant -- scares the ever-lovin' daylights out of Verizon
(among others), so much so that they *have gotten it made illegal* in
several states, and they're lobbying to expand that footprint.

See, among other sites: http://www.muninetworks.org/

As you might imagine, I am a fairly strong proponent of muni layer 1 --
or even layer 2, where the municipality supplies (matching) ONTs, and
services have to fit over GigE -- fiber delivery of high-speed data
service.

I believe Google agrees with me. :slight_smile:

Cheers,
-- jra

Cheers,
-- jra

4

And the country of Australia agrees with you (at least more than half of
it). The current nature of FTTH makes it hard to do a layer 1-only service
and the NBN has opted to provide layer 2 hand-off at POI's (points of
interconnect) around Australia.

jy

5

I believe Google agrees with me. :slight_smile:

Are they? Last I saw they were building out a layer 3 network -- no wholesale access -- did this change?

It sorta fit with their goals in that it meant they could build a faster/simpler network for less money and make a big/bold 1 Gbps to every home (not really true) statement, but it doesn't end up serving a very practical model for most of the world who believe the separation needs to happen at layer 2.

Layer 3 is interesting, but is everyone happy with saying goodbye to the ISP entirely and accepting regional monopolies on that space?

6

It is already a monopoly. Most places are served by one of the utilities: power, telephony or cable. He that controls the outside plant controls your fate.

Jared Mauch

7

Jared Mauch wrote:

It is already a monopoly. Most places are served by one of
the utilities: power, telephony or cable. He that controls
the outside plant controls your fate.

The difference is in how the services can be unbundled.

Power is additive (if in phase) that network topology is
irrelevant.

For telephony, unbundling for DSL at L1 is just fine.

So is optical fiber if single star topology is used.

WDM PON can still be unbundled at L1.

However, with time slotted PON, unbundling must be
at L2, which is as expensive as L3, which means
there effectively is no unbundling.

Or, CLEC may rent a raw fiber at L1 and operate its
own PON. However, as CLEC has less customer density
to share the fiber than ILEC, CLEC's fiber cost per
customer is higher than that of ILEC, which is why
PON promotes local monopoly.

          Masataka Ohta

8

I strongly disagree. If this were true, there would be no market for
MPLS service: folks would simply buy Internet service and run VPNs.

If you take my packets off at the first hop and deliver them to a 3rd
party provider, I can buy service from that 3rd party with as many IP
addresses as I want, I can buy service with BGP routing, I can buy
non-Internet services and I can buy bandwidth-hungry services that
aren't cost effective when they take a trip through the Internet
backbone.

Even if the cost for the unbundled L2 circuit was *identical* to the
cost of the bundled Internet circuit it would enable a huge range of
niche products that aren't practical now.

Regards,
Bill Herrin

9

It doesn't promote local monopoly if you don't allow the L1 company to provide L2+ services.

If the L1 company is required to be independent of and treat all L2+ services companies equally, then, the ILEC, CLEC, et. all have the same cost per customer.

Owen

10

Hi Owen,

Just for grins, I wonder: what is the minimal set of _structural_
requirements that could end the kind of abuses we see from the ILECs
without relying on good behavior? The problem with regulatory
compulsion is that it restrains the march of technological progress
too. Minimum is good.

Here's what I'm thinking:

1. Any company which provides more than "5%" of the OSI Layer 1
services in a given locality is prohibited from providing any Layer 7
services except those strictly incidental to the operation of the L1
service (e.g. billing or customer service web sites, internal
corporate network).

2. Such a communications infrastructure company may vend L1-L6
services only in units suitable for connecting single customers. For
example, they're not allowed to lease "a multi-customer coaxial cable
in the King street neighborhood." The service unit is "a dedicated
coaxial cable from 44 King street to the head end" or "A dedicated
cable channel from 44 King Street to the head end" or "25mbps/25mbps
from 44 King strreet to the head end" or "25 mbps / 25 mbps from 44
King Street to 888 King Street".

3. Such a communications infrastructure company is compelled to
provide reasonable and non-discriminatory access too all who would
interconnect. Charge whatever you want but no quantity or special
discounts and if you bill any service provider at the head end of the
connection then you bill them all the same. No settlement free peering
for this guy while that guy pays.

Regards,
Bill Herrin

11

No, you're right; that was me being flippant. ("He thinks flippant is the
name of a dolphin...")

Cheers,
-- jra

12

William Herrin wrote:

However, with time slotted PON, unbundling must be
at L2, which is as expensive as L3, which means
there effectively is no unbundling.

I strongly disagree. If this were true, there would be no market for
MPLS service: folks would simply buy Internet service and run VPNs.

You agree with me. MPLS at L2 sucks because it is as expensive
as, but less capable than, IP at L3.

If you take my packets off at the first hop and deliver them to a 3rd
party provider,

If you are saying delivery as IP, your local provider is an
ISP with monopoly.

Even if the cost for the unbundled L2 circuit was *identical* to the
cost of the bundled Internet circuit it would enable a huge range of
niche products that aren't practical now.

See the reality of your example of MPLS.

            Masataka Ohta

13

How many munis serve the rural like they do the urban?

In the vast majority of cases the munis end up doing what ILECs only wish they could do -- serve the most profitable customers.

Frank

14

From my own experience in my $DAYJOB, separating capital decisions at the L1

and L2 layers would end up adding cost. As mentioned elsewhere, GPON and
similar shared medium approaches do not lend themselves well to structural
separation. The most practical approach is dark fiber runs from the
customer to as few centralized places as possible. The CLEC would co-locate
their equipment at those centralized places. The CLEC is then free to use
ActiveE, GPON, whatever-the-next-gen-of-PON.

Structural separation works best when the cost to build to a customer are
roughly the same. Wherever there's significant disparaties, those will be
exploited and people will overbuild to the highest-margin/lowest cost
customers to avoid the averaged cost of L1 network.

Frank

15

Who cares?

It's time to stop letting rural deployments stand in the way of municipal deployments.

It's a natural part of living outside of a population center that it costs more to bring utility services to you. I'm not entirely opposed (though somewhat) to subsidizing that to some extent, but, I'm tired of municipal deployments being blocked by this sense of equal entitlement to rural.

The rural builds cost more, take longer, and yield lower revenues. It makes no sense to let that stand in the way of building out municipalities. Nothing prevents rural residents who have the means and really want their buildout prioritized from building a collective to get it done.

Subsidizing rural build-out is one thing. Failing to build out municipalities because of some sense of rural entitlement? That's just stupid.

Owen

16

That is why I believe that the L1 buildout should be done by or under contract to the local authority (whether that be a municipality, county, special district, or whatever) and then leased to L2+ service providers on an equal cost per subscriber basis.

Now it doesn't matter which subscribers cost more or less to build out, they all cost the same to serve. Yes, the more expensive subscribers are being subsidized by the less expensive ones. Overall, I don't really have a problem with this as I don't think that the discrepancies within a given authority area will be that large. I do think that we should require each authority to build out to all end sites within their jurisdiction not served by a smaller authority.

For example, Contra Cost County, California would be required to build out El Sobrante (unincorporated area of the county), but, not Pinole, Rodeo, Crockett, Hercules, etc. (since they would be required to be built out by their cities).

Yes, it's likely that the L2+ providers would have a higher cost per customer to serve El Sobrante than to serve the cities. However, since that increased cost would apply equally to all L2+ providers, it would easily be passed on to those subscribers and they would, therefore end up paying roughly the true cost of their choice to live in an unincorporated lower-density area.

Yes, higher-density authorities would have a better chance of attracting greater competition and diversity in L2+ providers. However, nothing would prevent or exclude smaller authorities from working out colocation deals with nearby larger (or even groups of smaller) authorities and bringing the termination points of multiple authorities together in the same location. Likewise, nothing would prevent authorities from building inexpensive backhaul facilities to adjacent larger centers.

If you cleanly separate the L1 infrastructure from the L2+ services providers, you really do have opportunities to do better for the subscriber base overall.

Yes, the L1 buildout will cost slightly more than an optimal monopoly build-out by a service provider. However, that small increase in cost yields huge benefits on the other side in terms of reduced barriers to competition, increased diversity, and more price pressure on the L2+ services side of things.

Owen

17

Well, for my part, /most of the poiny/ of muni is The Public Good; if /actual/ bond financed muni fiber is skipping the Hard Parts, it deserves to lose.

Time to assemble some stats, I guess.
-- jra

18

In a message written on Sun, Mar 25, 2012 at 11:47:58AM -0400, Jay Ashworth wrote:

Well, for my part, /most of the poiny/ of muni is The Public Good; if /actual/ bond financed muni fiber is skipping the Hard Parts, it deserves to lose.

I agree.

If a commercial company goes in to serve folks with fiber they
expect a relatively short ROI, 3-5 years typically. This is why
rural customers aren't "profitable"; they can't get money from a
bank or wall-street for a longer time so they are trying to spread
out the build costs over too short of a recoupment period.

Fiber has a 20-50 year life. Munis could finance fiber with a 20
year bond at a much lower interest rate than any corporation. By
spreading out the costs over 20 years these customers become
profitable, often quite so.

While in the CBD you might find more than one fiber provider passing
a building, for 99.999% of residential users there will only ever
be ONE fiber provider to the home. It's hard enough to make the
first fiber cost effective, there's no way to go into an already
served area incurring all the costs for < 50% of the customers up
front.

In many small towns muni-fiber in a single star topology to a central
switching station where multiple providers can co-locate would bring
competitive services at a very attractive cost for both the end user
and the services (IP, telephony, video) provider. It's also a topology
and technology that easily has 20-50 years of life.

19

most of the expense of laying fibre is associated with ducting + wayleave.
Once you have that in place, blowing new fibre is relatively inexpensive.
So rather than amortising the cost according to the lifetime of the fibre,
it makes much more sense to amortise over the lifetime of the ducting.

Nick

20

In a message written on Sun, Mar 25, 2012 at 05:29:04PM +0100, Nick Hilliard wrote:

most of the expense of laying fibre is associated with ducting + wayleave.
Once you have that in place, blowing new fibre is relatively inexpensive.
So rather than amortising the cost according to the lifetime of the fibre,
it makes much more sense to amortise over the lifetime of the ducting.

Maybe.

In rural deployments it's much more likely the fiber is aerial,
it's far cheaper to attach to existing poles with few cables on
them than it is to bury the fiber.

Even in urban areas where buried duct is the norm, being able to
use old ducts varies a lot with the geography and how active the
area is to other development. I've seen plenty of ducts where it
had been cut and repaired several times before use that running a
new cable through it was impossible and it simply had to be replaced.
In other locations 20 years later a new cable goes through like
butter.

But I think it's all a bit of a tangent; when talking about
_residential_ fiber it's prudent to run 2-6 strands to every home
day one, and then, well, there's basically never a point in running
more. The chance of blowing more fiber down the duct later is near
zero. It's also why I'm not a fan of *PON schemes, eliminate the
splitter and run a single star topology. 20 years from now Petabit
optics will look different than today's GigE in some way, but I'll
bet money they are tuned to run on single mode fiber. They may not
like the splitters and the like though. By doing a star back to a
wiring center you enable all technologies. GPON today, direct GigE
or 10GE where necessary, and all future technologies.