Muni fiber: L1 or L2?

Owen DeLong owen at delong.com
Sat Feb 2 06:22:43 UTC 2013


On Feb 1, 2013, at 21:22 , Jean-Francois Mezei <jfmezei_nanog at vaxination.ca> wrote:

> On 13-02-01 22:52, Owen DeLong wrote:
> 
>> Since the discussion here is about muni fiber capabilities and ideal greenfield
>> plant designs, existing fiber is irrelevant to the discussion at hand.
> 
> Not so irrelevant.  If the municipality wishes to attract as many
> competitive ISPs as possible, it wants to build a "standard" last mile
> that ISPs can easily interface to. One which is compatible with other
> FTTH systems.

Yes and no. As I said, I think it's more important to build a system that can
accommodate as many different potential technologies as possible rather
than to follow the conventional wisdom of the day developed by single-provider
monopoly environments.

> Currently, the standard is GPON (even though there are many variations
> to the theme).

Meh... Not in South Korea... The standard there is Gig-E to the home.

> Sone may say that having L1 service with each ISP having their OLT with
> splitters at the CO is an advantage. It also means that each ISP has to
> have its own ONTs in homes and they can all choose different configs for
> OLTs and the light in the fibre. Greater flexibility to differentiate
> between ISPs. (one may choose RFoG for TV with DOCSIS for data while the
> other is an all data link with  IPTV.)

Exactly.

> But for an end user, switching ISPs would mean switching the CPE
> equipment too since the ONT installed by ISP-1 may not be compatible
> with OLT used by ISP-2.

So? I don't see that as a problem.

> Requiring an ISP to have its own OLT at the CO with its own splitter
> also raises startup costs and reduces the chances of having competitive
> ISP environment.

Hence my suggestion that in environments where it may make sense to do
so, the muni could offer an optional enhanced L2 service. In this case, the
muni would supply OLTs, ONTs, and hand off the L3 work to the provider(s).

> Providing L2 service means that ISPs connect to a municipal OLT, so they
> do not have to purchase OLTs and bother with splitters. At that point,
> it si simpler and cheaper to deploy splitters in neighbouhoods. It also
> reduces number of splices.

Which I advocate as an OPTIONAL additional service.

> When you do 1:1, you may have a big cable with lots of strands leaving
> the CO, but you'll have a JWI in neighbouhood where you cross connect
> the strands from CO to the strand that uses the pre-fab cable to the
> backyards of homes served.

I'm not sure what your abbreviation "JWI" means.

> So in all the calculations made on dB loss, the number of splices was
> not factored in. You're not going to get a continuous cable from the CO
> to the telephone pole behind a home.  If you put the splitter at the CO
> you get the losses from the splitter, and then losses from a splice at
> the neighbouhood where trunk from CO connects to cables that runs
> through backyards.

Sure, but you get those same losses regardless of which side of the splitter
they are on.

> 
> When you put the splitter in the neighbouhood, it performs both the
> splitting and the connection of the cable from CO to the backyards. So
> you eliminate one splice.

According to http://www.thefoa.org/tech/lossbudg.htm
this is about 0.3db, so reduce the served radius by ~1km.
I think I already allowed for that in proposing an 8km serving radius
for 10km optics.

Given that 48 Gig -> 2 10G switches are getting cheaper and cheaper
(even in the managed variety) to the point where being able to deploy
them would be about 1/10th the cost per port of an OLT, I'm not sure that
GPON is necessarily the clear winner in a carrier neutral scenario.

Put splitters in the neighborhood and don't build for home-runs, then
you eliminate the ability to introduce new technologies. IMHO, that's
a really bad bet at this point.

Owen





More information about the NANOG mailing list