Last Mile Design

Sat Feb 9 20:13:17 UTC 2019

GPON is 2.4 Gbps downstream and 1.2 Gbps upstream. Residential users are
download heavy and more than 1:2. However there is a big difference between
average, peak and micro burst. The conclusion is not simple.

We typically have 60+ users on each port. We sell 1000/1000 internet. And
yet we only get good ratings for the speed.

I find that many, that are sceptical about the shared bandwidth of GPON,
forget that a typical POP might only be fed by a 10 Gbps uplink. Usually
this has much lower bandwidth per user than the GPON link.

Regards

Baldur

lør. 9. feb. 2019 20.52 skrev Grant Taylor via NANOG <nanog at nanog.org>:

> On 2/9/19 12:12 PM, Miles Fidelman wrote:
> > With early PON designs, upstream bandwidth was horrible. Not
> > particularly useful if you're doing things like remote backup, or video
> > chatting, or running a server (business grade service). GPON does better
> > on upstream bandwidth, but it's still asymmetric.
>
> Intriguing.
>
> I would have not considered my municipal GPON to be asymmetric.  Well,
> not as such.  Routinely, when I do speed tests I get better upstream
> speeds than I do downstream speeds.  (More below.)
>
> > If you're marketing to business customers, or home office professionals,
> > of families with multiple users that consume upstream bandwidth, AE
> > gives you a lot of room for upside growth (assuming you provision the
> > right kinds of fiber).
>
> Are you referring to the dedicated bandwidth between the CPU and the AE
> equipment?  Or the fact that bandwidth feeding the GPON and all
> subscribers is aggregate?
>
> I have attributed the asymmetry in my speed tests to be that most people
> on my GPON are predominantly downloading, thus consuming aggregate
> download bandwidth.  Conversely, few are uploading more than requests,
> thus using relatively little of the aggregate upload bandwidth.
>
> Do I see asymmetry?  Yes.  Is it truly asymmetric?  I don't think so.  I
> think is just based on consumption of aggregate bandwidth.
>
> I have no idea if this is normal for GPON or not.  Hence one of the
> reasons that I'm finding this thread enlightening.
>
>
>
> --
> Grant. . . .
> unix || die
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20190209/bb2e0372/attachment.html>