Don’t overlook the expense and complexity of large splices. Strand may be
cheap but splice cases, splicing and handhole and manhole are not cheap. In a
dense environment you need PON. The few power users that need active can
still get it as long as each drop goes to a splitter in a cabinet.
From: Lewis Bergman
Sent: Saturday, February 13, 2016 6:13 AM
To: af@afmug.com
Subject: Re: [AFMUG] Active or GPon?
What about somebody like Zhone? Last time I evaluated them they had a "pizza
box" GPON you could get into pretty cheap yet they still had all the components
you could want from the OLT to ONT to a pretty inexpensive TR069 management SW
platform. Making good money in this business always seems to be about reducing
truck rolls. AE doesn't provide that much info end to end while GPON and TR069
seem to be able to drown you in whatever you want to see.
Like others have said, to me it is the cabinets spread over everywhere that
really turns me off. Negotiating, paying for, and maintaining all those spaces
just makes my head hurt. I don't know what the possibility to turn 110 homes
into something more are. If designed right you could always migrate it to GPON
to fold it into a unified management system. The numbers we looked at the ONT
cost savings started to catch up with active around 75 users I think.
On Sat, Feb 13, 2016 at 6:28 AM Chris Fabien <ch...@lakenetmi.com> wrote:
Josh,
I don't think anyone is disputing that gpon is the right solution for an isp
with 1000s or millions of users. But Andreas asked about 110.
That size of project is something I think a lot of WISP are likely to be
working on. Our fiber network is currently several projects of that size - 50
to 200 homes within a few miles of a powered cabinet in a remote area. Active
was the cheapest way for me to do that and supports 1gig to each home.
Power for a 20u cabinet ( 288 ports in our design) will be about $30/mo when
fully loaded. And just 2 strands back to our NOC instead of 9 with PON which is
very significant if you happen to be leasing those strands, which we are in one
case.
On Feb 13, 2016 4:48 AM, "Josh Reynolds" <j...@kyneticwifi.com> wrote:
Eric it doesn't matter. That's 1024 strands, 1024 SFPs, more power
usage, more cooling, in multiple bigass cabinets.
Does. Not. Scale.
You take that into a dense suburb and that's what you end up with.
This is precisely why every decent ISP of size is deploying GPON and
not "active" fiber. The costs to get up _and_ maintain active is
several magnitudes higher. Let's say you were comcast and you were
rolling this out to your 22 million users on active. That's 22 million
SFPs, 22 million ports, an asston of strands, huge cabinets, large
batteries that have to get changed out every few years, HVAC, etc.
Even on a relatively common GPON deployment (32 way), you're talking
about a 32x reduction in port count, sfps, strands to pops, etc. from
22million ports to 687k. That's nothing to sneeze at.
On Sat, Feb 13, 2016 at 3:24 AM, Eric Kuhnke <eric.kuh...@gmail.com> wrote:
> That's assuming all 1024 active ports are in one central location and not
> distributed around, like 96 ports in one place, accomplished with a pair
of
> 48-port 1u switches (fed on a 10Gbps ring) accompanied by a beefy UPS, in
a
> weatherproof ventilated 16U cabinet.
>
> Multiply by location of several network nodes each with anywhere from 1
to 6
> 1U switches.
>
> On Feb 12, 2016 7:47 PM, "Josh Reynolds" <j...@kyneticwifi.com> wrote:
>>
>> If you're doing a super small project, no more than a hundred or two
>> hundred customers in an area, then it can make sense. There comes to
>> be a point where the port cost of active does NOT scale.
>>
>> 1024 subs on GPON with a modest 32 way split is done with 32 GPON
>> SFPs, 32 ports, 32 way split per GPON SFP. 2 line cards in a 2U
>> chassis.
>>
>> On active, that's 1024 active ports and SFPs. That's insane.
>>
>> On Fri, Feb 12, 2016 at 9:44 PM, Chris Fabien <ch...@lakenetmi.com>
wrote:
>> > I am also a proponent of active. Especially for small projects like
>> > this.
>> > Very low cost of entry.
>> >
>> > We looked at gpon including Alphion and ended up with still needing all
>> > the
>> > strands home run to the cabinet to fully load up each PON or we ended
up
>> > with a bunch of money wasted on PONs that would never be fully utilized
>> > if
>> > we did splitting closer to the customer.
>> >
>> > On Feb 12, 2016 10:30 PM, "Andreas Wiatowski"
<andr...@silowireless.com>
>> > wrote:
>> >>
>> >> So, I understand the benefits of GPon ... What brand would you
>> >> consider?
>> >> ... I have been looking at Alphion. Huawei seems like a good option...
>> >> But
>> >> much more expensive.
>> >>
>> >>
>> >>
>> >> Cheers,
>> >>
>> >> ______________________________
>> >>
>> >> Andreas Wiatowski | CEO
>> >>
>> >> Silo Wireless Inc.
>> >>
>> >> Email andr...@silowireless.com
>> >>
>> >> 19 Sage Court
>> >>
>> >> Brantford, Ontario N3R 7T4 (CANADA)
>> >>
>> >> Tel +1.519.449.5656 Extension-600|Fax +1.519.449.5536 |Toll Free
>> >> +1.866.727.4138
>> >>
>> >> -------- Original message --------
>> >> From: Josh Reynolds <j...@kyneticwifi.com>
>> >> Date: 2016-02-12 10:21 PM (GMT-05:00)
>> >> To: af@afmug.com
>> >> Subject: Re: [AFMUG] Active or GPon?
>> >>
>> >> You realize the transport core to the gpon OLT chassis is still active
>> >> fiber in many designs, right? I also am unsure if you are aware of the
>> >> upgrade process to NG-PON2 - you can run it on the same fiber strand
as
>> >> your
>> >> existing PON split. Add the new card into the chassis and move the
>> >> split
>> >> over to the new SFP. Upgrade the customers at your leisure.
>> >>
>> >> On Feb 12, 2016 9:13 PM, "Eric Kuhnke" <eric.kuh...@gmail.com> wrote:
>> >>>
>> >>> Key part there is, is going to be... is it available or shipping
now?
>> >>> If somebody wants to start a build now, the choice is between GPON or
>> >>> active.
>> >>>
>> >>> Having an active fiber path, even with just one strand (for BiDi
>> >>> optics)
>> >>> gives you a nearly infinite lifespan of the installed light path and
>> >>> cable
>> >>> plant, if things are maintained correctly. With a dedicated light
path
>> >>> from
>> >>> each powered network node to the customer you could upgrade to
>> >>> active-E 10,
>> >>> then 40, then 100Gbps someday. Yes we will see customers with 10GbE
>> >>> optics
>> >>> in the next ten years. And maybe in 20 or 30 years from now it'll be
>> >>> cheap
>> >>> and easy to connect each customer with an SFP-sized coherent QPSK
>> >>> 100GbE
>> >>> optic at each end.
>> >>>
>> >>> On Fri, Feb 12, 2016 at 7:08 PM, Josh Reynolds <j...@kyneticwifi.com>
>> >>> wrote:
>> >>>>
>> >>>> 10-40Gbps on NG-PON2 is going to be the real deal, and betting
>> >>>> against
>> >>>> it vs active ethernet at scale for residential service is just...
>> >>>> dumb, to be honest (IMO).
>> >>>>
>> >>>> The size of your backbone ends up being monstrous with active, as
>> >>>> well
>> >>>> as having to keep the cabinets powered, UPS+batteries, enclosurers
>> >>>> maintained, etc. PON is simply so much cheaper are scale, and in
>> >>>> residential every dollar counts.
>> >>>>
>> >>>> On Fri, Feb 12, 2016 at 8:56 PM, Eric Kuhnke <eric.kuh...@gmail.com>
>> >>>> wrote:
>> >>>> > I did forget to mention that I'm firmly on the side of activeE
>> >>>> > being
>> >>>> > the
>> >>>> > best choice, for one big reason... You can use all kinds of
>> >>>> > SFP-based
>> >>>> > equipment (24/48-port 1U switches) or chassis based switches and
>> >>>> > routers
>> >>>> > with 24/48-port blades from a huge variety of manufacturers.
>> >>>> >
>> >>>> > There's a lot of 48-port SFP stuff out there on the
>> >>>> > grey/refurb/used
>> >>>> > market
>> >>>> > that came out of datacenters, and no longer meets the bandwidth
>> >>>> > needs
>> >>>> > for
>> >>>> > people who are doing 10GbE (or 2x10GbE) to each bare metal
>> >>>> > hypervisor.
>> >>>> > But
>> >>>> > that same equipment is perfect for activeE.
>> >>>> >
>> >>>> > Same idea as a Cisco 3750G-48 is no longer enough bandwidth for
>> >>>> > 1000BaseT to
>> >>>> > the server in colo environments, but is perfect for MDU use.
>> >>>> >
>> >>>> >
>> >>>> > GPON/EPON/whateverPON is all a mess of manufacturer proprietary
>> >>>> > CPEs
>> >>>> > and
>> >>>> > non-interoperable stuff. Whereas with activeE and a real ethernet
>> >>>> > port
>> >>>> > for
>> >>>> > each customer you can use $30 media converters as your demarc.
>> >>>> >
>> >>>> >
>> >>>> > On Fri, Feb 12, 2016 at 4:53 PM, Andreas Wiatowski
>> >>>> > <andr...@silowireless.com> wrote:
>> >>>> >>
>> >>>> >> Hi all,
>> >>>> >>
>> >>>> >> Looking to do my first ftth for about 110 homes.
>> >>>> >> If I do active, what switch platform would you use for sfp in
>> >>>> >> cabinet and
>> >>>> >> in home router/cabinet.
>> >>>> >>
>> >>>> >> If GPon, what vendor would you choose that is cost
>> >>>> >> effective/reliable
>> >>>> >>
>> >>>> >> I understand the full limitations of GPon.. But I feel it is an
>> >>>> >> attractive
>> >>>> >> proposition compared to active... And the few systems I have seen
>> >>>> >> have a
>> >>>> >> road map to faster olt access.
>> >>>> >>
>> >>>> >> Cheers,
>> >>>> >>
>> >>>> >> ______________________________
>> >>>> >>
>> >>>> >> Andreas Wiatowski | CEO
>> >>>> >>
>> >>>> >> Silo Wireless Inc.
>> >>>> >>
>> >>>> >> Email andr...@silowireless.com
>> >>>> >>
>> >>>> >> 19 Sage Court
>> >>>> >>
>> >>>> >> Brantford, Ontario N3R 7T4 (CANADA)
>> >>>> >>
>> >>>> >> Tel +1.519.449.5656 Extension-600|Fax +1.519.449.5536 |Toll Free
>> >>>> >> +1.866.727.4138
>> >>>> >
>> >>>> >
>> >>>
>> >>>
>> >
