It is not often I jump to NBN’s defence, however on this occasion I am going to. I should also point out that I do not speak for NBN in anyway.

It has been reported in ITNews and elsewhere that NBN is considering introducing a “Netflix Tax”, meaning an extra charge on streaming services. This is simply not true.

NBN is undergoing a pricing review. This is being held behind closed doors with their direct customers, the RSPs, and a number of other interested stakeholders, such as the ACCC, ACCAN etc. It is an iterative process with an initial list of ideas to stimulate discussion. Having received feedback (verbally or via written submissions), NBN will then further develop those ideas that seem good, produce a new list and get further feedback, before coming up with a final position paper. This is expected to take many months.

While I understand that everybody wants to know what is going on, it is behind closed doors because that allows everyone involved to speak freely, come up with a whole plethora of ideas, some of which may be very bad, but can be the launchpad for further much better ideas.

In my opinion, if there is one thing NBN needs right now it is some radical thinking, some radical ideas about how it can expand it’s market share, or provide higher quality, higher value services to Australians. To have these ideas enter the public debate too early, is detrimental to this process.

So the idea that has got everyone worked up about is this:

“Would your organisation support the development of a price response whereby charging of streaming video could be differentiated from the charging of other traffic/services”.

The key word here is “differentiated”. NBN is not looking to charge more for Netflix, indeed, after conversations with the author his intention was that the price of Netflix traffic be reduced.

The reason it is a lively subject is that NBN like every other bandwidth provider in the world is seeing the growth in streaming traffic overwhelm everything else on the network. NBN charges RSPs for bandwidth using the CVC construct and many RSPs have been complaining that streaming services are killing them.

NBN’s thinking is if they were to find a way to reduce the price of this traffic, then it would alleviate some of this RSP stress.

For the record I think it is a bad idea because as others have pointed out it fires up the whole Net Neutrality debate. I will be giving that feedback in our submission to the Pricing Review. Frankly I doubt this idea will make it into the second stage.

In our submission we will also be putting forward some of our own ideas about how we can increase the takeup of ultra high speed services (250Mb and above). I use the word “ultra” because NBN has already described the 100Mbps speed as “high speed”, which I find rather cute. Some of those ideas may well be considered by others as crazy and will not be taken up. That is the process.

Just as an aside one of the debates that has come up around this in the USA is whether RSPs should charge Netflix etc for sending traffic into their network. Accusations around “double dipping” (charging both the subscriber and the content provider for the same traffic) have abounded. 

However there are some companies, primarily Telstra and to a lesser extent Optus & TPG, who already do this. They do charge (either directly or indirectly) Netflix to send traffic to their subscribers via their IP Transit charges. All the other RSPs allow Netflix to send traffic onto their network for free via “peering”. However Telstra and the others refuse to peer. The big four (Telstra, Optus, TPG & Vocus) are forced to peer with each other by the ACCC, but nobody else.

Also for the record, we have found Netflix exceptionally good to deal with. They provided us with some caching (technically they are content servers) to install in Hobart which have enabled us to continue to grow our network in Tasmania despite the very high backhaul charges.

Damian Ivereigh
3rd July 2019

NBNCo (the company rolling out the nations broadband network) has announced a wholesale pricing review that. As a kick off to this review they also announced that they are going to increase the CVC inclusion on the 100Mbps plans from 2.5Mbps to 3Mbps.

Unfortunately there is perception that this as a performance boost to 100/40 services. It is, but it is nothing like the 20% boost you might think it is.

The way the bundles work is that all the CVC inclusion is put into a pool (the CVC) which is then shared among all of the RSP’s bundled services – including those on other, lower speed, tiers. Given that the number of 100Mbps services is, for most RSPs anyway, a small percentage (around 10%) of the total number of services, the increase per connection is much smaller when averaged over all of the connections.

Let’s put some example numbers into this. If your RSP has 1000 services in the pooled CVC you are connected to. Let us assume that your RSP has the typical spread of speeds, which is, 90% are on 50Mbps and below and 10% are on 100Mbps and above. The former gets 2Mbps of included CVC, the latter currently has 2.5Mbps of CVC, will soon be 3Mbps of CVC.

So currently the total CVC is 900 X 2 + 100 X 2.5 = 2050

After the change the total CVC is 900 X 2 + 100 X 3 = 2100

So the increase is just a little under 2.5%. Not something to get too excited about!

If you are with a good RSP, then they will already be buying extra CVC over and above the amount included in the bundles, called overage. This overage costs $8/Mbps, so the above example RSP will see a reduction in their costs of $400 – 40c per service. Again not something to get too excited about. If your RSP is not buying overage (which may be why you are seeing congestion), then you will see a 2.5% reduction in congestion (whatever that means!) as they increase the CVC size.

To be fair to NBN what they are trying to do is make the 100Mbps plans a little more attractive for RSPs to sell, they are not reducing the cost of the 100Mbps plans rather they are including more for the same price.

My concern is that NBN is currently in a game of chicken with the large RSPs who are essentially holding their users to ransom by constraining their CVC to just the included bundle amount and waiting for the bad press to accumulate blaming NBN for the problem. Since 2017, NBN has put customer experience (or “CX” as they call it) high up their priority list, we will see whether it is above getting revenue!

At Launtel we are already buying overage because we believe that having a high performing, ultra low congestion, network is our key point of difference.  This is the main reason we are more expensive than most of our competitors.

Damian Ivereigh

June 30, 2019

We are currently evaluating whether we continue offer 250/100 (250Mbps down and 100Mbps up) services in Canberra. Unfortunately due to NBN’s CVC pricing construct it is extremely expensive to offer these services for a relatively small number of people and we are very disappointed to say that it looks like we will just have to stop offering it.

So what’s CVC and what’s the problem with it?

NBN charges providers, such as ourselves, in two parts. The first is the AVC charge and is a single price per connection based on the peak speed of that connection. Higher speeds are higher priced as you would expect.

The second part is the CVC charge, which is the total bandwidth requested by the RSP to be used across all services in an area (called a CSA, which is pretty much just a POI – point of interconnect). In Canberra there are two CSAs: Civic and Queanbeyan. If an RSP does not purchase enough CVC then the result is congestion at peak times as there isn’t enough bandwidth to go around. NBN and the RSPs assume that a certain amount of CVC, typically just a few Mbps, is used on average per connection.

So for example if 2.5Mbps of CVC is allocated per connection, then if you have 1000 services, then your CVC should be 2500Mbps.

The problem with that calculation is that it doesn’t scale down. You have to have a minimum amount of CVC just to allow a service to work at all. Clearly you have have at least 100Mbps of CVC just to allow a single connection to peak at 100Mbps. However when you have two you probably need 200Mbps so they can both operate at the same time. However when you get more, you can start going back to the old formula. The received wisdom is that need about 3 times the highest speed of your services as a minimum on your CVC. 

NBN recognised this scaling issue and allows an RSP to get a minimum of 300Mbps of CVC effectively for free to get past this. This was clearly aimed at supporting 100Mbps services.

Great what about higher speed ones? Unfortunately NBN has done nothing to help. If RSPs want to sell 250Mbps and above, they need to buy more CVC, at $8/Mbps/month. The problem is that all that extra CVC is really just to support the higher speed services and it gets really expensive per service.

Let put some numbers on this. To support 250/100 services we really need to buy 750Mbps of CVC – that’s an extra 450Mbps, $3600 per month over what we need for the 100/40 services. We don’t actually buy that much, about 600Mbps and unfortunately we are seeing some occasional congestion when two people on 250Mbps service fire up at the same time, during peak times.

We don’t have many 250/100 services – around 15 – so that’s $240/month per connection – around $8/day. This is just the CVC! Add in the AVC charge and it’s about $11/day just for the NBN component. So yes we are making a big loss on 250/100 services!

The problem is that the higher the price, as you would expect, the less people will want to buy it, but the CVC cost is the same just spread over fewer people, making our cost per service higher. A vicious cycle.

To make matters worse our daily pricing works against us – people, quite reasonably, will just upgrade for the day just for some big downloads. The problem is we have to keep (i.e. buy) the bandwidth on hand regardless.  

Why don’t we have this problem in Tasmania?

Because we are well past the scale issue – we have about 10 times the number of connections in Tasmania that we have in Canberra, plus we setup our gigabit network in Tasmania mainly for businesses, whose CVC capacity we can use at night!

So where to from here?

We recognise that the 250/100 service is one of our unique points of difference and we are loathed to turn this off. However we have been running for almost a year and it is clear this problem is not going to fix itself.

We also recognise that some people get the 250/100 because of the higher upload speed. We could keep this part of it, by offering a 100/100 service (which is just a 250/100 service with the downstream side speed limited to 100Mbps). This would be cheaper than the current 250/100, but still at least $1.25/day more than the 100/40. Would people want to buy this?

We have considered providing a sort of “best efforts” speed where people can use the high speed if no-one else is using the CVC, however this could get quite complicated to explain and I really don’t want people to buy something thinking it’s one thing and then being disappointed.

I would appreciate any comments you have on the above – feel free to message resi-support@launtel.net.au where our staff will ensure that the emails are forwarded onto me.

Damian Ivereigh
17 Jun 2019

 

The 2019 Budget is almost upon us and unfortunately it appears that yet again our various governments (State, Federal etc) seem to prefer spending large amounts of (our) money on old-world, outdated infrastructure rather than the infrastructure we know we will all need in the future.

I note that another $115m is to be spent on the approach roads to Hobart which will join the $461m to build a new bridge across the Derwent River (at Bridgewater). That is almost $600m on roads and bridges just in Hobart! Now don’t get me wrong I like driving on roads and bridges, but having lived in Sydney for many years, Hobart’s traffic congestion still doesn’t come close to what’s going on over there (although I admit having a plan for the future would be prudent, but it should include the likely effects of technology).

Just for comparison, the cost to upgrade all of Tasmania’s fixed line areas, those currently using NBN’s Fibre-To-The-Node technology, would be about $200m. So take your pick Tasmania: a few minutes less commute time or ultra-fast broadband which would mean that you could probably work from home and avoid the commute in the first place, oh and you’d still have $400m to spare – maybe bolster the Fixed Wireless network with delivering fibre to the towers, as originally envisioned, instead of the limited (but cheaper) microwave backhaul links they currently use.

I get it that pollies, quite prudently, want to invest in technology they know will be useful and not soon get outdated. However I would argue that an investment in roads is riskier than fibre. Looking forward 40 years, are we sure we are going to need these roads, or at least such wide, multi-lane ones? By then there’s a strong likelihood that we will have self driving cars – ones that can drive at 110kph 1 metre behind the vehicle in front. They will be synchronised so that all the traffic moves in an amazing high speed dance through areas where we currently see congestion. We will need no traffic lights, cars will just call ahead to the other cars, then speed up and slow down as they approach intersections so they seamlessly join or cross the traffic flow. Will this happen? I don’t know, but it wouldn’t surprise me. It is certainly a cloud on the horizon for road investment.

Compare that with fibre. There is just no possibility we won’t need this and lots of it over the next 40 years. Every part of our lives is increasingly connected and our demand for data carriage seems to grow completely unabated (anyone see any hint that it is going slow down?). Forget 5G, it will be a great technology, but it will only ever be a “fill-in” where mobility is important, it will never be able to do the heavy lifting that fibre, with its almost unlimited bandwidth capabilities, will provide. Hands up anyone who wants to predict that the internet is just fad and in 10 years time we won’t need it! Investing in fibre is seriously a dead cert.

Tony Abbott said: “Do we really want to invest $50 billion of hard earned taxpayers money in what is essentially a video entertainment system?”. I wonder if this is the problem. A fundamental misunderstanding in what the internet really is. Let’s face facts, typically the most formative years of politicians, in previous business life, were before the internet really existed. For some, those who have been in politics from school, they’ve not been in areas where innovation has been needed . Perhaps, they just don’t get what the internet is about. For them the internet is just about entertainment. How do we change that?

Yes the internet can be used for entertainment and why not? When I last checked governments everywhere invest in stadiums, parks and supporting the arts. Not to mention that roads also get used by people in their leisure time. This is about making our lives more enjoyable, we don’t want to just live to work. However the internet is so much more than entertainment. It is now a key driver to our economy – sadly it’s becoming a bigger key driver in other economies. You try taking almost any business and disconnecting their internet – it isn’t long before staff are sent home. Xero published a very interesting report showing how the arrival of NBN has caused businesses to do more sales and employ more staff – oh and these business no longer have to be in the centre of the city, at the end of a long commute, in order to operate or even do sales.

So yes I get it that everyone is getting excited about 8k TV and many people are saying it won’t work here, but that’s not the only reason we should be rolling out fibre.

Underpinning our mindset with technology ensures our ideas are innovations at a global standard. But if ideas are constantly held back by the limits of technology then we’ll always be playing catch up to other parts of the world who are turning their economies around. In fact, as the pace of other gig-cities picks up our opportunity to be a leader in the new world economy falls further behind.

I get it that pollies love to turn up and cut a ribbon, great for roads, but there is no ribbon for a fibre project – it’s not all in one place – in fact that’s the point. Unlike a road that typically only connects two places (or crosses a river etc), what makes fibre great is that it’s a web of many points to many other points.

Indeed, roads themselves are being reimagined. China is likely to build the first truly smart highway and auto makers are investing heavily to ensure their model lineups stay relevant!

For anyone who clings to the idea that the current government are good economic managers, clearly has not understood the extraordinary damage they’ve done both to the financial viability of the NBN project and worse to our economy as a whole by switching to old copper technology. This damage is just beginning to show now, but there’s lots more to come as we continue losing our place in a connected global economy.

What do we have to do to get governments to plan for the future before the damage is too hard to repair? Does anyone care that Australia could in the not too distant future lose our place as a first world economy.

Damian Ivereigh
Launtel

We’re all rather used to politicians (of both sides, but mainly the current government) spruiking their achievements in rolling the NBN. However the latest one from Senator Mitch Fifield, the Minister for Communications and the Arts titled ‘3 in 4 premises able to connect to the NBN’, is so full of inaccuracies and half truths that I feel it needs a response.

Senator Mitch Fifield’s Media Release on the NBN

The media release talks about how they’ve achieved 3 in 4 premises being able to connect to the NBN and that they’ve done this more cheaply and quicker than Labor’s original FTTP rollout. He then goes on to quote some figures about why the Coalition’s MTM model is so much better than the original Labor plan.

The first thing to note is that regardless of the politics, the Labor party under Rudd actually went to the 2007 election promising a $4.7bn FTTN rollout in a partnership with private industry. However they soon came up against a recalcitrant Telstra (remember Sol Trujillo?) who essentially said:

“It’s our copper and you can’t have it unless you give us $20bn. We will then take that $20bn and roll out fibre to all the cities in competition with you”.

The country would have been left with a very expensive, poor quality network.

So the FTTP model wasn’t Labor’s initial option, but at the time the only viable one.

They managed to negotiate with Telstra (paid them a bunch of money and waved a big stick in the form of spectrum allocation) for them to get out of the access (delivery to the home) market. This then laid the foundation for the Coalition 6 years later to return to the FTTN model.

It is also worth pointing out that the previous Coalition government under Howard had created the problem with Telstra in the first place by, against a wealth of expert advice and experience in other countries, selling off the government’s controlling stake in Telstra without structurally separating it into retail and wholesale divisions. Indeed one of the early achievements of NBN was to force this structural separation.

Fifield criticises the Labor FTTN model:

“Under six years of Labor, 50,000 premises were connected. The Liberal National Government connects almost 100,000 premises every month.”

This may be technically accurate, but it is highly misleading. By measuring progress in terms of premises connected, they are completely ignoring the massive amount of planning and preparation that is required for a project of this size. IT systems, national network design, procuring construction contractors, designers etc – all that preparatory work needs to be done before the first premise is connected. This is not like building Snowy Hydro which is essentially a single project where each task is done once or just a few times.

Rolling out NBN is like operating a distributed factory – it is the same task over and over again in different places. Lessons are learned, economies are found, production ramps up. The feedback I got from engineers at NBN that they had reached their 3rd iteration of the design and were just hitting their straps when it all came to a halt.

“Labor’s NBN would have taken six to eight years longer to complete and cost an extra $30 billion. Every household would have been charged an extra $500 a year for broadband.”

Mike Quigley, the first CEO of NBN, fully expected the build to be completed for $54bn (about the same cost as the revised MTM cost) and to finish just one year later than the current MTM rollout – 2021. The cost savings and production ramp up, that Quigley expected to attain, have easily been achieved by Chorus in NZ, yet NBN and the government stick doggedly to their $4400/premise price for FTTP to justify their use of the MTM.

A recent article posted on The Monthly shares a very detailed reflection of Quigley’s views on how the Coalition broke the internet.

When the Coalition came to power the first thing Malcolm Turnbull (as Communications Minister) did was to launch an enquiry into the current FTTP rollout. They took the current rollout progress and simply extrapolated it to get their 8 years longer and $30bn more – clearly the enquiry was totally self serving and designed to justify their change to MTM. The old joke of a politician never announcing an enquiry unless they know the outcome seems very apt.

However the worst part of the whole thing is that Australia has spent all this money (same as the FTTP rollout) and we have been left with a dud – it may not be a complete dud now, but it will be in five years time. We got the expensive, poor quality broadband that the Labor government had originally considered and rejected. NBNCo is struggling to meet it’s revenue targets as people connect to the NBN at a lack lustre rate.

When rolling out infrastructure, which is slow and expensive, you don’t look at what the market needs now, but at what it will need in 10 or 20 years time. Anyone can see that our need for fast, reliable broadband is only increasing (40% per year according to the ABS). In the long term what the Coalition have done is set our economic future back at least 10 years. It will end up costing the country significantly more overall when we have to do the inevitable upgrade to full fibre (and no 5G isn’t going to cut it). This from the party that calls itself “the better economic managers”.

We will look back on this period as the greatest wasted opportunity. It’s an extraordinary shame that we failed to grasp it.

If you want to hear Mike Quigley talk about this much more eloquently than me, I highly recommend you watch his opening statement (the first 12 minutes) of his recent testimony to the Senate Committee. 

It is not clear what the long term fix for the NBN is, but the first stage has to be to recognise that we have a problem. The sooner we can get the politics out and the engineers in, the sooner we can make a start.

Damian Ivereigh
CTO Launtel
25/02/19

The company building Australia’s National Broadband Network, NBNCo, have said on many occasions, that they are confident of reaching their $52 ARPU (Average Revenue Per User) target at the end of the rollout in 2020. This is important because it underpins the value of the company as a whole and therefore any discussions of whether there should be a writedown in the debt that it owes to the Commonwealth.

They have made much of their various business offerings including Enterprise Ethernet and the business bundles as how they will raise this extra revenue. However they have really only hinted at where they actually expect most of this extra revenue to come from: CVC Overage.

A quick recap: CVC is the bandwidth charge that NBNCo levies on RSPs for the amount of bandwidth that they deliver to NBN for delivery to their customers. It was the cause of much argument in the latter half of 2017 as RSPs were clearly restricting their CVC purchases (in order to keep their retail prices down), allowing their customers to get very congested connections. Most people blamed NBNCo’s choice of FTTN, HFC etc as being the root cause of the poor performance, rather than the RSPs buying decisions around CVC.

NBNCo temporarily solved the problem by offering first “Focus on 50”, then the bundled plans that included enough CVC (at the time) for a reasonable discount to the RSPs.

Over the course of 2018 many RSPs took up Focus on 50 and then the bundles and the ACCC reported that congestion seemed to be a thing of the past. But is it?

The sting in the tail with taking on the bundles is that an RSP has to commit to not congesting their CVC – in other words NBN does not want them to sell a congested product over the NBN. They do this by explicitly stating that they must not exceed 95% utilisation for more than 7 hours in any 7 day period – there’s more in there, but that’s the gist of it.

The amount of CVC included in the bundles is 2Mbps for an nbn50 connection and 2.5Mbps for an nbn100 and above. However Australia’s bandwidth is currently increasing at about 40% per year and doesn’t look like slowing down anytime soon, indeed with the growth of high definition TVs it is only set to accelerate. So if 2Mbps was enough a year ago, it would need to be 2.8Mbps now just to keep pace with this growth.

So if you’re an RSP selling unlimited data plans what do you do? You can’t constrain your users (or even get more money from them by upgrading to higher data download plans) yet you also can’t let your CVC become congested. You just have to buy more CVC – more than the included bundled amount – to avoid congestion. This extra amount is called “CVC overage” and NBN currently sell this for $8/Mbps/month.

So some quick maths, if you are selling an nbn50 with its 2Mbps of included CVC, but actually have to allocate 3Mbps to avoid congestion, now instead of that connection costing you $45 per month, you are now looking at $53 per month. NBNCo’s revenue problem solved!

So will the RSPs actually stump up the extra money? If history is anything to go by, it is not looking good. Most RSPs are already complaining that they are making no money from NBN products so they have little incentive to play ball.

The good ones probably will go ahead and they will either have to stop selling unlimited plans or put their prices up. The others will simply restrict the bandwidth before it reaches NBN to keep it below the 95% threshold that NBN have set, in order to stay within the rules. So we will be back to 2017 with its congested networks all over again. At which time we can expect the finger pointing to restart.

NBNCo is clearly banking on RSPs buying this overage, only time will tell if they actually do.

For the record at Launtel we are already finding that the allocated CVC bandwidth with the bundles is not enough for a congestion free service and have started buying CVC overage. We believe that a high quality, low congestion, network is our primary goal, even if it means we have to charge more than the others.

Damian Ivereigh
Launtel
18 Feb 2019