Standards for last mile performance

Sat Apr 30 18:36:45 UTC 2016

For us (FTTH) we had/have enough aggressive foresight to do smaller
splits.. 1x16. Some are doing 1x2's or 1x4's at the corner somewhere into
1x16's or 1x8's, so at the point where you start to hit decent saturation
you can just shrink the upstream split and fuse onto a new upstream strand
/ optic. Once that gets overused, thankfully you can overlay NG-PON2.

As far as measuring, in our case just having your NMS of choice just
monitoring the OLT via SNMP.

For fixed wireless, monitoring and management are the easy part. The hard
stuff is detecting what amounts to "dark matter", or the things you're not
normally looking for. Channel utilization (completely variable from moment
to moment), over all AP capacity, CPU usage, retransmissions, keeping per
client modulation rates very high to limit tdma timeslot utilization, etc.

The fixed wireless side ends up requiring a LOT of experience, monitoring,
and guesswork.
On Apr 30, 2016 11:28 AM, "Jean-Francois Mezei" <jfmezei_nanog at vaxination.ca>
wrote:

>
> The CRTC hearing went well (thanks for all your help).
>
> One of the unanswered questions was how to set performance standards for
> the last mile to ensure people get advertised speeds (within reason).
>
> I had asked the question about contention ratio and it appears there is
> no proper way to set such a moving target as a regulatory standard.
>
> During the hearing, someone suggested that advertised speeds be
> achievable 80% of the time. (chairman then asked if "time" was 24 hours,
> or just the time you needed to use the internet (aka: peak).
>
> Out of curiosity, could such a thing be measured by the last mile operator
> ?
>
>
> There is the easy answer of synch. For DSL, non delivery of advertised
> speed is easy since that metric is on the modem statistics. However, for
> fixed wireless, would a customer too far from tower see a lower synch
> rate or would he just see poor performance due to lots of retransmits ?
>
>
> So some generic questions:
>
> What are the different ways used to determine if the last mile is
> congested and needs to be upgraded ?
>
> From the network operator's point of view, would it not be looking at 5
> minute throughput samples and trigger upgrades when it sees throughput
> reaching  x% of last mile segment capacity for more than X minutes per
> day ?
>
> What other means do network admins use to monitor when it is time to
> upgrade shared last mile segments such as coax or fixed wireless ?
>
> from a policy point of view, is it possible to set the same standard for
> different technologies or should each (dsl, coax, fixed wireless and
> satellite) have they own standards and methods of measurements before of
> intrinsic differences in how they work ?
>
>
> In the case of Rogers (canadian cableco), the CRTC record shows that
> they trigger node split process when capacity reaches 60%. This is
> because it takes them so long to do all the paperwork, committees etc
> that by the time the node split is done, that segment has grown to about
> 75% utilisation.
>
> Would that be a sound basis to set a policy ?
>
> For shared last mile, would different technologies have similar
> thresholds that trigger the need for upgrades or would coax start to
> degrade at 75% whereas fixed wireless or satellite  start to degrade at
> lower/higher number ?
>
> For FTTP, while likely not a big problem yet, would similar number apply
> when the ~2gbps download and ~1gbps upload start to get filled by the 32
> homes served ?
>