[AusNOG] NBN

Nathan Sullivan nathan at nightsys.net
Tue Nov 22 22:50:31 EST 2016


A few follow-up thoughts:

- I didn't include the 52,641 satellite subscribers when spreading the CVC
across all activations. This would dilute all of those figures above
further, dropping them marginally (probably only cents not dollars per AVC
per mth).

- All of this is based on once RSPs have some minimum critical mass of
provisioned CVC per PoI, the additional CVC requirement as new customers
are activated would be linear to the # of new activations. This assumes NBN
Co's modelling of future year CVC revenue would grow close to linearly
(after an initial ramp up) with activation volumes. The above revised AVC
figures would grow the same way. Obviously without knowing what NBN Co's
internal models look like there's no way to judge the accuracy of that.

Regards, Nathan.

On Tue, Nov 22, 2016 at 9:40 PM, Nathan Sullivan <nathan at nightsys.net>
wrote:

> Thanks Mark, OK this has prompted me to do the math, heh...
>
> Anyone who wishes, feel free to peer review these numbers, if you find any
> assumptions are "off"/could be done differently etc.
>
> First off, Premises Activated stands at 1,378,273 as of 30th September
> 2016 (aligns with between the weekly figures
> <http://www.nbnco.com.au/content/dam/nbnco2/documents/nbn-rollout-metrics/nbn-rollout-metrics-22102016.pdf> from
> 29-Sep-2016 and 06-Oct-2016). Premises Activated at the start of the
> quarter was 1,098,634 (30-Jun-2016). Need to factor in that this metric is
> not static throughout the quarter, so will need to amortise everything down
> to monthly to work with next.
>
> Revenue of $181 million total. FTTN + Wireless + Satellite + FTTP revenue
> adds up to $115 million (page 3 of the results presentation). CVC/NNI
> revenue of $54 million. Pity we don't have the split for this, but let's
> continue.
>
> Now I'm going to aggregate the fixed line + fixed wireless speed breakdown
> %s to ease the next calculations. Using figures on 29th September 2016 (4k~
> subs lower than actual), wireless activations sat at 136,935, fixed line
> sat at 1,185,353. Page 5 of the results presentation gives us:
>
> fixed line 12/1: 31% = 367459 subs
> fixed line 25/5: 51% = 604530 subs
> fixed line 25/10: 1% = 11853 subs
> fixed line 50/20: 4% = 47414 subs
> fixed line 100/40: 13% = 154095 subs
>
> fixed wireless 12/1: 17% = 23278 subs
> fixed wireless 25/5: 80% = 109548 subs
> fixed wireless 50/20: 3% = 4108 subs
>
> And aggregated:
>
> total: 1322288 subs
>
> 12/1: 367459 + 23278 = 390737 subs / 1322288 = 29.55%
> 25/5: 604530 + 109548 = 714078 subs / 1322288 = 54%
> 25/10: 11853 subs / 1322288 = 0.8%
> 50/20: 47414 + 4108 = 51522 subs / 1322288 = 3.89%
> 100/40: 154095 subs / 1322288 = 11.65%
>
> Based on current AVC pricing, working on September 2016 activation figures:
>
> *12/1: $24* * 390737 = $9,377,688/mth
> *25/5: $27* * 714078 = $19,280,106/mth
> *25/10: $20* * 11853 = $355,590/mth
> *50/20: $34* * 51522 = $1,751,748/mth
> *100/40: $38* * 154095 = $5,855,610/mth
> Total: $36,620,742/mth
>
> Being the above metrics exclude satellite services, let's work in reverse
> to verify the monthly revenue figure is accurate enough. FTTN + Wireless +
> FTTP revenue for the quarter is $112 million. 3 x $36.620m = $109.86m.
> Close to accurate, the only concerning point is that the subs at June 30
> was substantially less than September 30 (approx 20% less), which means the
> revenue figures could be closer to $28m for July 2016... so theres a good
> couple of million unexplained in revenue, which may skew things here
> (possibly the "other charges" stuff on the NBN Price List, support
> callouts, NTD modifications, no fault found callouts, etc etc. Let's assume
> that is explainable...)
>
> So let's make a calculated assumption that NNI charges are about X % of
> CVC/NNI. Based on NNIs being connected by 14~ service providers
> <http://www.nbnco.com.au/sell-nbn-services/wholesale-providers.html> (a
> good portion of RSPs will be using one of these wholesalers and not paying
> for their own NNIs), 121 PoIs, without being in the "know" of how many Temp
> PoIs remain active vs currently active permanent PoIs this get's
> interesting... $400 per 10Gbit (assume most are using LR instead of ER,
> assume most are using 10G also for simplicity). Let's assume that 50% of
> all wholesalers are at all PoIs (to account for the fact metro PoIs will
> have almost all of the wholesalers, and the most remote PoIs will barely
> have Telstra + one other). ((14 * 121) * 0.5) * $400 = $338,800/mth. For
> the quarter, let's say $1mil. Note there are probably also a heap of NNI
> setup fees hiding in here also (new PoIs connected by various wholesalers),
> let's allow for $1mil extra there for the quarter also. Leaves $52mil for
> CVC. Note this breakdown was probably the most rough of all figures here
> due to lack of public data on the NNI side of things.
>
> Let's assume that Jul 2016 CVC is $15m, Aug 2016 CVC is $17m, Sep 2016 CVC
> is $20m, round figures... gets us to $52m.
>
> For September 2016, let's break down the subscriber figures above and take
> a proportionate % of the CVC and add to the AVC revenue, to get an
> aggregate we can divide by sub count to get the revised AVC per port
> (without CVC):
>
> 12/1: 29.55% * $20m = ($5,910,000 CVC + $9,377,688 AVC) / 390737 subs = *$39.12/mth
> revised.*
> 25/5: 54% * $20m = ($10,800,000 CVC + $19,280,106 AVC) / 714078 subs = *$42.12/mth
> revised*
> 25/10: 0.8% * $20m = ($200,000 CVC + $355,590 AVC) / 11853 subs = *$46.87/mth
> revised*
> 50/20: 3.89% * $20m = ($778,000 CVC + $1,751,748 AVC) / 51522 subs = *$49.10/mth
> revised*
> 100/40: 11.65% * $20m =  ($2,330,000 CVC + $5,855,610 AVC) / 154095 subs = *$53.12/mth
> revised*
>
> The biggest flaw to this is the proportional consumption of CVC by each
> port speed is probably far different to the % of services activated by
> speed. NBN Co probably have more internal metrics of actual traffic
> throughput averages by port speed activated, which would be a far better
> way to split up the current "CVC revenue pool" amongst the various AVC
> speeds.
>
> Noting all dollar values above are ex GST.
>
> PS. I should have thrown all this in Google Sheets, meh.
>
> Regards, Nathan.
>
> On Tue, Nov 22, 2016 at 8:05 PM, Mark Smith <markzzzsmith at gmail.com>
> wrote:
>
>> On 22 November 2016 at 19:30, Nathan Sullivan <nathan at nightsys.net>
>> wrote:
>> > Just to touch on the CVC topic again... has anyone done the math on how
>> much
>> > the AVC would need to be increased to match the CVC charge revenue base?
>> >
>> > As we know, NBN Co has some revenue targets to meet all of their
>> financial
>> > viability requirements, and there are a few cost bases that they recoup
>> this
>> > through. The 2 large ones being CVC and AVC charges, with other smaller
>> ones
>> > such as NNI (from the perspective of overall revenue, not itemised
>> cost).
>> >
>> > Based on the fact NBN Co has provided some figures in the past of
>> percentage
>> > of services on each tier (I don't have the most recent media articles
>> that
>> > were quoting those handy), that would allow calculating the AVC revenue
>> > component to an extent, as services activated is a public metric on the
>> > weekly reports.
>> >
>> > Excluding the "other revenue" portion of things, this would then leave
>> CVC
>> > for the most part, plus a little for NNI's etc.
>> >
>>
>> Quite a lot of what you're after is in here. ARPU (per month not for
>> the quarter), revenue broken down in to FTTP, Satellite, CVC etc.,
>> speed tiers and percentages etc.
>>
>> http://www.nbnco.com.au/content/dam/nbnco2/documents/Q1%
>> 202017%20Results%20Presentation.pdf
>>
>>
>> > Being that NBN Co has stated that as traffic volumes increase, they will
>> > drop the per Mbit CVC price (within a defined range for now, publicly a
>> > figure of "$10"/mbit has been thrown around), that would effectively
>> argue
>> > that they just need to meet that % of overall CVC revenue somehow. If
>> this
>> > were done in the form of increased AVC, and in return CVC goes away and
>> all
>> > services can consume up to their AVC rate limit, or the maximum
>> capacity of
>> > aggregated NNI's on a per provider basis, would this be a more suitable
>> > solution?
>> >
>> > I suspect the AVC prices would have to go up a semi-decent amount for
>> this
>> > to stack up, but if the only scaling cost to RSPs is then adding NNIs
>> (which
>> > are closer in price to IX ports), would this avoid a lot of concerns and
>> > confusion? I'd be interested in seeing what those numbers actually
>> looked
>> > like, and if someone does publish them what other RSPs would think of
>> > them... I haven't had enough spare time to do the math myself yet :)
>> >
>> > Open to feedback/suggestions/criticisms.
>> >
>> > (Pricing document incase anyone wants to do the math:
>> > http://www.nbnco.com.au/content/dam/nbnco2/documents/sfaa-
>> wba2-product-catalogue-price-list_20161028.pdf
>> > )
>> >
>> > Regards, Nathan.
>> >
>> > On Tue, Nov 22, 2016 at 5:20 PM, Mark Smith <markzzzsmith at gmail.com>
>> wrote:
>> >>
>> >> On 22 Nov. 2016 11:33 am, "Alex Samad" <alex at samad.com.au> wrote:
>> >> >
>> >> > Hi
>> >> >
>> >> > there was some talk about our network being able to handle amazon
>> video
>> >> > coming to oz.
>> >> >
>> >> > I'm thinking this might have more of an impact
>> >> >
>> >> >
>> >> > http://www.theverge.com/2016/11/21/13703152/netflix-4k-pc-wi
>> ndows-support
>> >> >
>> >> > All those 4k tv's being sold..
>> >> >
>> >>
>> >> How many are actually being sold and how many are actually big enough
>> for
>> >> how close people sit to them for those people to benefit?
>> >>
>> >> To sit a normal distance of at least 6 feet away, you need a 70"+
>> screen
>> >> to actually benefit. Smaller 4K TVs look great in the shop because
>> you're
>> >> standing much closer to the screen.
>> >>
>> >> http://referencehometheater.com/2013/commentary/4k-calculator/
>> >>
>> >> > from
>> >> > https://en.wikipedia.org/wiki/4K_resolution
>> >> > Streaming video[edit]
>> >> >
>> >> > YouTube, since 2010,[67] and Vimeo allow a maximum upload resolution
>> of
>> >> > 4096 × 3072 pixels (12.6 megapixels, aspect ratio 4:3)
>> >> >
>> >> > High Efficiency Video Coding should allow the streaming of content
>> with
>> >> > a 4K resolution with a bandwidth of between 20 and 30 Mbps.[72]
>> >> >
>> >> >
>> >> > Not sure what the current average DSL connection is . But all your
>> going
>> >> > to need is 2 kids (teenagers) and ....
>> >> >
>> >> >
>> >> > if only it hadn't become politicised
>> >> >
>> >>
>> >> No even wrong.
>> >>
>> >> CVC cost is the problem not FTTN. Changing to a literal 100% FTTP
>> network
>> >> would not solve that problem and would make it worse - NBNco have to
>> pay for
>> >> thosr loans somehow or other.
>> >>
>> >> > A
>> >> >
>> >> >
>> >> > _______________________________________________
>> >> > AusNOG mailing list
>> >> > AusNOG at lists.ausnog.net
>> >> > http://lists.ausnog.net/mailman/listinfo/ausnog
>> >> >
>> >>
>> >>
>> >> _______________________________________________
>> >> AusNOG mailing list
>> >> AusNOG at lists.ausnog.net
>> >> http://lists.ausnog.net/mailman/listinfo/ausnog
>> >>
>> >
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20161122/e0e05f65/attachment.html>


More information about the AusNOG mailing list