CBN FTTN test shows speeds of 105Mbps

98

fibretruck

news The National Broadband Network Company today revealed it had successfully conducted early trials of the Fibre to the Node (FTTN) rollout model mandated by the Coalition Federal Government, with the trials delivering “raw” download speeds of 105Mbps over a distance of 100m from a local test ‘node’.

Under Labor’s previous NBN policy, some 93 percent of Australian premises were to have received fibre directly to the premise and the remainder satellite or wireless. Fibre to the Premises, as a technical model, allows very high upload speeds. Current retail NBN plans, for instance, allow upload speeds of 40Mbps on 100Mbps plans, while the long-term migration to gigabit speeds will allow upload speeds of 400Mbps.

However, NBN Co’s Strategic Review published in December last year changed the paradigm, with the company recommending (and the Coalition formally supporting last week) a vision in which up to a third of Australian premises will be served by the existing HFC cable networks of Telstra and Optus, with Fibre to the Node and Fibre to the Basement used in other areas not already covered by Labor’s FTTP approach. Satellite and wireless is to be used to cover some rural and regional areas as under Labor’s previous plan. This new model is known as the “Multi-Technology Mix” (MTM), or “the Coalition’s Broadband Network” (CBN).

NBN confirmed late last year that it was planning to conduct FTTN trials, as well as accompanying trials of Fibre to the Basement technology in apartment blocks. And in February it confirmed the trials had begun with real customers.

The Fibre to the Node build pilot was to be undertaken in two locations: Umina near Woy Woy on the NSW Central Coast and Epping in Melbourne’s northern suburbs. NBN Co was to construct two small scale Copper Serving Area Modules, erecting kerbside node cabinets which will connect fibre to spare copper pairs in the Telstra pillar. NBN Co stated that neither site had been earmarked to receive FTTP within the next 12 months under the previous rollout plan. Once active, NBN Co was to invite RSPs to participate in a Fibre to the Node end user trial. This limited term trial will test the delivery of high speed broadband via FTTN to up to 100 premises at each location.

In a new statement today, NBN Co said the first in-premises test by NBN Co of the FTTN technology had delivered raw download speeds of 105Mbps and upload speeds of 45 Mbps.
The technology trial was conducted recently by NBN Co engineers in Umina, near Woy Woy, on the NSW Central Coast.

Engineers installed a VDSL modem in a shopfront and then connected the existing copper telephone line to a node cabinet in the street. The node cabinet, which has been erected alongside a Telstra pillar around 100 metres from the shopfront, converts electrical signals transmitted over copper wires into pulses of light that can travel over optical fibres.

The company issued the following rider to its speed claims, however: “NBN Co provides services to its wholesale customers, telephone and internet service providers, and does not provide services directly to end users. This trial involved testing speeds over physical network layers rather than end user speeds. End user experience including the speeds actually achieved over the NBN depends on some factors outside our control like equipment quality, software, broadband plans and how the end user’s service provider designs its network.”

NBN Co chief executive officer, Bill Morrow, said in the company’s statement: “This is an important milestone in the rollout of the National Broadband Network (NBN). It demonstrates that existing technologies such as the copper network are capable of playing a vital role in delivering high speed broadband to Australians.”

NBN Co chief operating officer, Greg Adcock said: “As outlined at our analysts’ update today, this speed test is a result of the work by our engineers and construction crews to install a small number of node cabinets in Umina on the NSW Central Coast and Epping in Melbourne’s northern suburbs to pilot the construction and delivery of FTTN.”

“Once active, the next step will see NBN Co invite retail service providers to take part in a limited FTTN end user trial in these locations. The aim will be to test the delivery and end user experience of high speed broadband via FTTN to a small number of premises at each location.”

According to NBN Co, the results of the technology trial bolster the findings of the Strategic Review conducted by the company and published in December 2013. The Strategic Review determined that the NBN could be delivered sooner and more cost-effectively by incorporating existing capable telecommunications networks with an upgrade path – such as copper and hybrid fibre coaxial cable (HFC) – into the rollout.

The company is now planning a widescale commercial rollout of technologies including FTTN next year, following the renegotiation of NBN Co’s Definitive Agreements with Telstra.

However, the so-called “Multi-Technology Mix” option which NBN Co is pursuing continues to be unpopular with Australians in general, with most preferring Labor’s technically superior FTTP model. For example, in mid-February Shadow Minister for Communications Jason Clare presented to Federal Parliament the signatures of 272,000 Australians who want the new Coalition Government to build Labor’s all-fibre version of the National Broadband Network instead of the technically inferior version which is currently being proposed.

NBN Co’s own Strategic Review shows that the full FTTP model, if reworked to be more efficient, could be delivered for peak government funding of $54 billion by the end of calendar year 2023. The Coalition’s preferred MTM model could be delivered for $39 billion by the end of calendar year 2020. Both would deliver a modest investment return on the Government’s capital injections. However, commentators have pointed out that over the long term in which the infrastructure is expected to operate — between 50 and 100 years — the time and cost differences between the two plans is negligible, leaving Labor’s FTTP proposal as the more technically superior.

opinion/analysis
Very good news that NBN Co has been able to achieve these very solid results in these tests. However, I will note that thus far we don’t have any data available about the quality and reliability of the copper run used in the tests. As has been noted many times previously, the quality of Telstra’s copper varies significantly between different areas and even between different streets.

I will also note that the speeds delivered by NBN Co here are not consistent with the experience in the UK, where BT’s Infinity service delivers speeds only up to 76Mbps. Given that the UK’s copper cable is widely considered to be in better condition than Australia’s, I am surprised at the results being shown here. I would also be curious to know whether the distance of 100m from the nearest node being used in these trials is considered a relatively normal distance in terms of a wider rollout of FTTN, and what distances comparable rollouts in other countries use from the node to the premises.

Image credit: NBN Co

98 COMMENTS

  1. They’re going to need a lot of nodes to ensure everyone is within 100M of one! Nobody said reasonable speed wasn’t possible (though this is at the current upper limit and no mention of upload speed), just that your mileage may vary.. Considerably!

    Also 1 result, in optimal conditions and probably on cherry picked copper. Call me cynical, but I’m hardly surprised they are gloating about their “success”.

    The whole point of FttP was to provide a consistent and future proof service. Which just isn’t possible without a massive investment in nodes and copper remediation.

    • “NBN Co said the first in-premises test by NBN Co of the FTTN technology had delivered raw download speeds of 105Mbps and upload speeds of 45 Mbps.”

      I am not sure if Renai added that after but they have talked about uploads. Whether that is enough going forward is highly questionable.

  2. I would just like to say that 100m is a really short copper run. I wouldn’t even call this a decent or realistic test as the distance between the customer and node is too short. I would prefer to see the test on line lengths around the 500m-750m mark and then judge what would be absolutely appalling speed results.

    • Wow 105/45mbps over 100m amazing!

      Ethernet does 1/1Gbps on Cat5e or 10/10Gbps on Cat6a over those distances so this is really pretty lame even by Copper comms standards!

      Why are we wasting 40 Billion on a temporary upgrade again? (which doesnt include the extra costs to access Telstra’s Copper CAN which will likely make MTM vastly more expensive)

    • “I would prefer to see the test on line lengths around the 500m-750m mark”

      No doubt they would be around ADSL with 25 Mbps down and 2 Mbps up. This said very few people would have lengths like that. It has been a few years but I used to look at Telstra’s cable records as part of my job, without access to them now by best guesstimate would be 90%+ of the city population would have able lengths length to nodes of 100-250 m.

      I would be interested hear more accurate figures if someone has current access or has done a proper statistical analysis of them.

      Extra note: Below comments claim Turnbull said a typical length would be 500m. To be honest I find it hard to believe this number from what I have seen, but I guess it was a few years ago. Unless the number is skewed by regional customer who will not be getting FTTN anyway.

    • Those speeds over 100 metres mean I will have to have two nodes on my place to get anywhere near the 100 Mbps mark. I’d say that would cost a little more to install run and maintain than 1 Gbps FTTP.

  3. Next time your in a public sitting with these people Renai, is there any chance you can ask these questions or anyone for that matter?

  4. I’m not overly confident…from the way the above is written, it seems like they tested one premise (or at least, have found one premise that matches their story, and have released its results). I’ll wait until there is a decent sample size and then find out what is actually going on.

        • Having more connections to the node won’t change the connection speed from that shop. Congestion from the node to the interconnect is possible. water ingress to the copper is possible to degrade the signal, but not people connecting to the node.

          100m is fairly optimal for VDSL. 500m is going to make things painful.

          • Yes, but the DSLAM in the node is going to be ‘busier’ holding onto multiple connections, which *might* induce reliability concerns…I probably wasn’t clear enough with what I meant… :)

          • “Having more connections to the node won’t change the connection speed from that shop. “

            That isn’t strictly accurate. Vectoring helps moderate crosstalk but it requires (or rather, it is more optimal) that bundles are under the control of the same MSAN; it cannot eliminate it completely.

            100 meters is within an optimal range, however, so the differences are going to be minor.

            But, this is all a bit redundant – 100 meters isn’t going to be a median length, if copper tails of 500+ meters are expected.

            Neither the Minister, Telstra or NBNco have stated what the maximum length will be, post acceptance of the mixed technology model.

            Until they do, we can only assume described numbers are best case scenario. The gulf in speeds could be considerable.

          • “Having more connections to the node won’t change the connection speed from that shop”

            Completely wrong. One of the biggest issues with VDSL is multiple connections. Since it operates up to 17 or 30MHz there is a lot of interference due to crosstalk. BT in the UK, with their small 10% takeup figures were showing slowdowns to less than half speed with only that 10% of lines active. If everyone shifts to FTTN, which they will have to, as I can’t see maintaining the copper to the exchange for years to come in the plan, we are looking at about 70% untilisation and huge speed lose due to crosstalk.
            This is the problem vectoring should address. If and when vectoring is added this may help with this. There is no clear indication if it is in the plan or has been costed. Who knows, it’s just had some lip service from Turnbull.

          • Actually having more than one modem on the line will reduce speed beyond what load on the node would cause. VDSL uses sophisticated math and it knowledge of all the signals in the copper bundle (which is why you can’t have true competition) to reduce the effects of cross talk and noise on the lines however the amount crosstalk(cause by having multiple modems running) it is still going to reduce speed. The only way to avoid this is use a transmission medium not susceptible to electromagnetic crosstalk, this doesn’t actually exist in any usable form but experimental physicist get to have fun* with it. With light over fiber optic we are a long way off having an issues with cross talk between fiber lines, but I suspect it won’t be an issue until the NPN(the National Panoramicband# Network) deployment in the year 2109.

            *depending on your definition of fun
            #Dear future dictionary people I want credit for coining panoramicband as the term for faster broadband internet.

      • Yeah, and what’s even better is that according the the DA maps the longest runs are at minimum 700-900m from the node. I say minimum as that is assuming the best possible run from the node to the end of the DA

  5. “.. with the trials delivering “raw” download speeds of 105Mbps over a distance of 100m from a local test ‘node’”

    100mbit. Not a bad effort. Trials elsewhere have shown that sub 200 meter runs result in 100mbit+; so this is reflecting the same outcomes.

    Presumably the trial is assuming that everyone will have a node parked in their front yard. Is Turnbull aware of that requirement?

    If NBNco is happy to fund deployment of a node on the kerb, then as I am just on 90 meters from the street, I’ll be all set.

    A few hundred thousand nodes should cover it.

    By the way – what was the upstream speed? Was that published?

    • edit: 40 mbit up. Cool.

      Yes, tried to update my post after hunting through the article again and was too slow.

      3 minute edit is sad making, Renai. Too little time. :(

      How many nodes are we going to need, to cover ~100 meters per-premise? That’s what the mixed-mode model will need to hit now, to deliver Turnbull’s desired outcome.

      Has anyone actually scaled up the last number (based on 500+ meters) to figure this out?

      • I believe the original amount was something like 70,000 nodes in total, each servicing up to 800mtrs, i can only guess if you divide the 800 by 8 you will need about 560,000 nodes.

        I can’t remember where the 70k figure comes from, i just remember hearing about it over the last several years.

        • The number will actually rapidly climb the closer to the premises you get. I don’t believe it’s linear as the closer you get the density of premises drop dramatically.

          70k was bandied about, but that was for 6-700 meters I believe. Once you’re down to 100 meters then physically speaking you can only fit so many dwellings in that space (and MDUs don’t count as they are FTTB already).

          As much as 100/40 is actually pretty bloody good from VDSL, it’s not reasonable to consider this a median speed, unless the nodes are deployed within the same distance of the trial as a general rule.

          • Ahh, makes sense. I was just thinking linearly. So as you drop the distance down, it exponentially rises?

          • It’s not complicated. How many houses around you would there be in a 500 meter radius? Shrink that radius to 100.

            Given we also tend to build housing in linear chains (roads do that) then you are potentially looking at orders of magnitude in scale. 70,000 could easily become hundreds of thousands.

            I don’t get the point of a trial that isn’t intending to match what the deployment model would be. I hope this was one of many pairs in a trial area and so is reflective of vectoring to combat cross-talk.

            The last thing we need is a trial of a single copper pair at 100 meters from a dedicated node.

          • Though those figures would be mostly for metro type areas, right?

            Living in a rural area in the south-west of Western Australia (Capel), the street i live on, in a 500mtr radius we probably have 20 houses and i live about 2kms out of the town centre.

            Drop that to 100mtrs and the number of houses becomes about 10.

          • Capel shows the dramas for both builds.

            Cleary larger than 1000 population, the central population is easy to follow under either plan (handful of nodes, or simply FttH up and down every street), but when you get a little out of town you end up servicing less and less people.

            So where do you draw the line? For FttN, Capel is laid out conveniently to cover 90% of the people with a handful of nodes – one near either end of Reynolds Rd, another 3 or 4 on the southern side, and probably one near Jenkins Rd and Capel Dr.

            Most properties are pretty easy to deal with. Its the others that are the problem, and its a problem repeated in every township in rural Australia – the edges of town. In every location, there are properties that neither build conveniently fits.

            And its going to suck being one of those properties in the same way its going to suck being the first place outside a new estate… So close, yet so far.

          • Ive got ~30 in a 100m radius in a area of what I consider normally Australian suburban sprawl.
            But hey I get fiber :), but hey I’m moving away from fiber not long after it is installed :( to an area that I suspect will eventually be known as Turnbulls HFC hell.

        • If you figure the area of a circle (pi*r^2), which the coverage basically is, 800m radius covers amount 2 million square meters. 400m radius covers about 1/4 that, and 100m covers about 30,000 square meters.

          One of the problems with FttN – it IS upgradable to faster versions, if you add more nodes. But the shorter the copper loop, the more nodes you need, on a logarithmic progression.

          Half the loop means 4 times the nodes. Quarter the loop means 16 times the nodes. One eigth the loop means 64 times the nodes.

          So even if the 70,000 nodes is intended for even a 400m loop, to get to a 100m loop would be 16 times that amount. Or 1,120,000 nodes… 1 node per 10 properties, each the size of a fridge.

          • I am new, but I wonder if anyone has noted or calculated the carbon footprint of all these powered nodes, compared with FTTP which I understand does not require any extra power.

          • I am new to this discussion, but I wonder if anyone has noted or calculated the carbon footprint of all these powered nodes, compared with FTTP which I understand does not require any extra power.

          • The problem with your maths is that a node isn’t a radio. The signal follows the cable and the cable approximately follows the road. Dropping onto google maps there are 35 house within a 200m walking distance from a point in the ‘burbs. There are the same number of houses within 100m “as the crow flies” from that same point. Then if you look closer you notice that all of the house are about 20m away from the road.

            That’s before you consider the possibility that the lines don’t necessarily follow direct line paths down roads to the closest part of the house and directly into the back of a modem……

          • Let me put this a different way. If the 70,000 nodes we supposedly need cover 800m loops, you are going to need a shitload more if the loops are reduced to 400m or 100m. I was trying to show, in the easiest way possible, how quickly that number increased.

            By the simplest math, using the radius of a circle, which is the easiest way to measure this, that means 4 times as many nodes if you simply halve the distance. For the purpose of this discussion, its the worst case scenario, and the easiest way to show the problem. Archimedes proved this stuff 2200 years ago.

            It doesnt matter if the 70k covers a 400m loop either, because that just means that 3/4’s of the population are going to be further away than this test, and because of the laws of physics will not be able to get this level of connection. The same laws demand that the further you are, the lower your speed.

            These tests also need to show what the speed will be at the maximum planned distance for it to serve any purpose. Until then, its a publicity stunt that does nothing but show that 3/4’s of those connected are going to get screwed in the same way ADSL2 did.

          • I don’t disagree. I was merely pointing out that the actual coverage is much lower than just drawing a circle on a map. A point 100m away from a node as the crow flies could in theory be 100m of copper away from the node. However it could be 500m or even kms away. You can think of it as a circle and it’s easy to understand it that way but drawing it as a circle is extremely generous.

          • Yeah, I’m just 3.2km from my exchange, but I’m on a 4.6km loop. Basically I’m on the other side of a river, and the cable goes a fair way north before crossing the river and coming back south towards me.

            If only it was a direct line, I’d have better than 3mbps internet!

          • I’m 800m from the exchange as the crow flies. I can see it from my front gate (its a little elevated on a small hill), there is ONE corner between me and the exchange. But the most direct line rollout stops 100m or so up the road, so for me the line ends up nearly 2.5kms.

            People across the road are near direct line, people behind me are as well, but due to either a dodgy line that was never sorted, or some quirk of build 80 years ago, I am left effectively 2.5kms from the exchange, and enjoying a 6 Mbps connection versus the 16 Mbps connection around me.

            And yes, I’ve pushed the issue repeatedly.

          • This is probably sounding like one of the four Yorkshiremen from Monty Python but I too am in a similar boat. 1.2km straight line, 1.6km by road and 3.8km by copper. I suspect that this is a pretty common problem.
            That’s for my business, at home I’m on a ADSL1 RIM approximately 700m away. With the NBN budget warnings that MT is giving, we’ll need to count ourselves lucky to get an ISAM upgrade in the existing cabinet. Forget getting a node closer than that.

            Easy to do now that all promises are off and we’re back to the “up to” drivel that is the current market.

          • Its a very common problem. The copper lines were laid out for phone services, not data services, and as the phone services didnt care about distance, holes in the build were filled in as needed. So you get situations where the copper lines were laid out in big spirals, because that was the easiest way at the time.

            Its only now you see the issues around that. I live in the CBD of Wollongong. Not a small township. But because the city was built in a specific way, it means I now suffer the consequences. And the fix is a ridiculously short bridging across one 10m gap, but nobody is ever going to do it, because all Telstra needs to guarantee is the phone service where the distance doesnt matter.

            Happily, it looks like we’ll get FttH, and when they roll it out the logical method is to bridge that gap. Even if we get FttN, the logical build is by bridging that gap, as the street I’m on is a convenient distance from the trunk lines and outskirts of the CBD areas natural boundaries.

            Its where that isnt an option that sucks, but the reality will be that if whole segments get pooched by FttN builds, something will need to be done.

  6. It’s all well and good doing it with one shopfront, but what happens when you have hundreds of pairs, all running this high frequency VDSL type signalling, side by side?

    I’m guessing the answer is a huge amount of crosstalk and interference.

    The only conclusion I can come to is that this is nothing more than a shallow PR stunt, a cheap attempt at fooling people into thinking that FTTN is somehow appropriate for the NBN.

    • A single connection, on a 100m length, on remediated copper.

      If we can only get 100Mbps in that artificial scenario then this is not good news at all.

      • That’s exactly my point, the scenario seems realistic at a glance, for the average reader.

        But as soon as you do any real kind of thinking about it with even with only some basic knowledge of Electrical Engineering things start to fall apart.

        And we all know vectoring is bollocks unless one provider controls everything at the node/exchange, so no, in this scenario it’s not appropriate.

        In my mind all this test does is prove how wrong FTTN is for us.

        • Actually, FTTN is reasonably well behaved if the copper tail is kept within an optimal range for VDSL2+ and Vectoring is in use.

          There’s a good set of infographics via Alcatel-Lucent: http://www2.alcatel-lucent.com/techzine/vdsl2-vectoring-in-a-multi-operator-environment-separating-fact-from-fiction/

          The problem is that that optimal range to achieve 100 mbit is less than 400 meters. In this example the length was 100 meters.

          If we consider lengths of 500-700 meters then performance drops considerably.

          • “If we consider lengths of 500-700 meters then performance drops considerably.”

            Does anyone have a link to an actual study of the distribution of length.

            Would NBN not simply try replace Telstra pillars with nodes? Whilst there are probably cases of pillar to premise cables being this length from memory the great majority were less than a few hundred meters. If someone still has access to these figures I would love to hear what they are.

          • Many pillars have been replaced via the ‘top hat’ project that was recently executed by Telstra. Most of those will now be cabinets with upgraded, or newly installed MSANs.

            Given the very short distances of the trial data being feed by Telstra, it sounds like lines have been re-mediated and test nodes upgraded.

            However, anyone on a not-Telstra ADSL service will by definition not be on copper that traverses a CMUX or MSAN other than at the exchange; ergo that service would create a need for a cabinet; that’ll be getting into the hundreds of thousands of connections.

            Many of us are simply on ye olde copper pairs of varying lengths, between a few hundred to a few thousands meters away from an exchange,

            70,000 nodes was the last guesstimate number to cover it. But that’s assuming somewhere around 7-800 meters which falls well outside of the VDSL2+Vectoring goldilocks zone.

  7. Notice how The Australian’s article on the matter fails to mention the distance involved. Guess that must have been an editorial oversight. I mean, are we really surprised by this? We will have to wait until the rollout proper before we discover the real truth.

  8. 105Mb with only one premises (no crosstalk) at 100m. Well yes, I wouldn’t expect any different. Apply the same testing to ADSL2+ and it would get 24Mb. What point is checking a technology works? Of course it does. Does it work with 70% of premises connected and at distances out to 500m? ie, does it work in real world conditions.

  9. Notice how this is just so nicely conveniently above GPON’s 100/40?

    Here’s a secret: if you turn up the power, you can do 105/45 over about a hundred metres of copper, even if it’s crappy, sure. Especially if you take advantage of frequencies above 17 MHz, easily.

    That’s right, if you’re given the freedom to shove as much power over it as you want, you can make it do more than 105/45 too. The one thing I find disturbing is that they didn’t actually get more. If they really wanted to make it a trial to show the best copper can do then 105/45 would be pretty underwhelming.

    So my guess is that they weren’t doing that but just put in a single VDSL2 line and turned up the power levels. Let’s see how well vectoring will mitigate crosstalk and how this result will fare when you have more than one connection For all we know, they may well have used a bonded connection here too, but probably not.

  10. isnt this just a single line, no other VDSL devices to kick in crosstalk etc? it certainly is peak speed not sustained or inclement weather speeds as well.

    and yes one might fully expect at 200m it would be significantly slower….

    its nice to see the result but i feel it is ultimately sound and fury signifying nothing, or being generous, signifying very little.

  11. These tests mean nothing. Connect a whole building with and get everyone using it at the same time. Then you’ll see actual (read: crap) results.

  12. If everyone gets a node within 100m, then this is great news!

    Shame the next step up from the 100/40 (g.fast) involves frequencies that overlap with FM radio though…

  13. basically Billion$ spent “upgrading” current ADSL2 copper for VDSL that only work for <1000meters compared to fiber's effective of hundreds of Kilometers o.O

    See how using VDSL compares to ADSL and Fiber:

    http://nbnmyths.files.wordpress.com/2010/09/dsl-speed-comparisons.jpg

    http://nbnmyths.files.wordpress.com/2011/06/fttn-speed-graph.gif

    http://www.abc.net.au/technology/images/general/blogs/nr/nbn/futurebycabidas.png

    thousands of power-hungry "Nodes" that can't compete on performance are the Coalition's answer to the NBN's "expense problems" ?

    morons, morons everywhere.

    • Those graphs are not very consistent. Consider VDSL2.

      At 100m, one gives ~220Mbps and the other ~70Mbps, then at 300m, one gives ~130Mbps and the other ~30 Mpbs, and at 500m one gives ~80Mbps and the other ~25Mbps.

      So between 100-500m there is a disagreement of a factor of ~3. Presumably this is largely depended on the copper and level of cross talk but considering one graph has VDSL1 better than VDSL2 between 250m to 1500m I am not sure how reliable this data is.

      • Vectoring, my friend.

        The ofcom image appears to be based on non-vectoring VDSL. The former do appear to match vectored services.

        It would have been more sensible to link to the article, however, as numbers can be made to say anything.

        This is the thing. For VDSL to work, you need vectoring to control the noise across each pair, and the goldilocks zone is within 500 meters.

        Outside that, it’s a wilderness of speeds.

  14. Can someone curate the information in these comments and send it to Labor so they can ask some hard questions? Ridiculous that nobody is doing this.

    • “Surely they are” asking, wouldn’t they be?

      Call me cynical, but perhaps it’s just a continuation of MSM/Pay TV providers and their extreme fear of obsolescence at the hands of the original FttP NBN, who simply aren’t reporting such questioning?

  15. Several questions about these results…

    1. Is the speeds quoted PHY layer, application layer or something in between?
    2. What VDSL2 profile was used?
    3. What gauge and what type of copper?

  16. 100mtrs? Why would you not just run fiber for that last bit?
    There is no way, the majority of the rollout will have premises within 100 meters. What a crock,

  17. You know what this looks like to me?

    This is the equivalent of having a postie stand about 3-4 steps away from a man w/ mail in his hand. Then gets timed as he walks and hands the letter to a person.

    And then he gets a pat on the back and everyone starts proclaiming! See guys? snail mail is still fast and effective! So much better than that email/texting thing that takes ages being transferred from your phone/computer to a satellite and the receiver!

  18. “….. over a distance of 100m from a local test ‘node’. ……. ”

    Best laugh I have had in years.

    Are the media really swallowing this BS.
    Do you honestly think everyone will be within 100m of a node.
    I can absolutely garantee you a node every 100m is just not practicable.
    You can pitch it 1000 different ways but its still BS.

    Fibre is the most common sense solution.
    longer runs, faster speeds, more bandwidth, the list goes on

    And Id also like an answer to Brendan’s question
    By the way – what was the upstream speed? Was that published?

    The article is garbage and should have been attacked by every single serious IT media person.
    Not a word from any single person in the media to attack such a ridiculous suggestion.

    (Renai, you have the knowledge, and understanding to attack such rubbish and you chose not to. Disappointed that you didnt. It could have been such a great article)

    • “(Renai, you have the knowledge, and understanding to attack such rubbish and you chose not to. Disappointed that you didnt. It could have been such a great article)”

      No matter what I write, there are some people for whom it’s never enough. Why don’t you write your own goddamn article on the subject. When you’ve done that, let me know.

    • There’s a good critique of NBN Co’s press release here at The Register:

      http://www.theregister.co.uk/2014/04/16/nbn_co_in_broadband_kit_we_tested_worked_stunner/

      You can read the announcement here if you want: it explains that the test achieved “delivered raw download speeds of 105 megabits per second (Mbps) and upload speeds of 45 Mbps.” We suggest you also read between the lines because the announcement is scandalously scanty because it omits discussion of topics including:

      Whether the test in any way resembled a real-world scenario
      Whether the test used a production network connection or something set up especially for the occasion
      Whether the test used the equipment NBN Co intends to use in production
      The state of the twisted pair between the node and the retail store where the test was conducted
      Length of the cable run, not just the linear distance between the node and store
      Setting used on the node and modem during the test
      What kind of data was downloaded

      We could go on but by now you probably get the point: a press release saying “we’ve proved FTTN probably works” and offering very few other details is an interesting way to prove the government’s new mixed media NBN is going to deliver the goods. Or even prove it can deliver the bits.

      • Beat me to it. :)

        Yes, The Register was, on the whole, not impressed. Not that I blame them, of course.

  19. Seriously, 100m from the node?

    How many people, or what percentage does Malcolm Turnbull think will be 100m from the node? My bet is 5%. Why not post answers in the 500-1000m range.

    One then has to also question, if you’re 100m from the node, does it seriously make any sense to not run the fiber cable directly to the house, FTTH?

    I mean, you’re basically having 95% FTTH / basically limitless speeds (ie – we were going to get 1000Mbps), but then you choke the entire connection to a pitiful amount (25Mbps), all because you replaced everything but 100m of copper cable? Seriously… what the ****

  20. First thing I noticed was that the cable in the photo is *black*, not the happy green or blue of the FTTP cables lol.

  21. Giving Telstra $92 billion hardly counts as a $42 billion rollout. If you can count it that way, then tax offsets from Labor’s made it a negative cost rollout.

  22. I am surprised they didn’t do the tests from 10 meters to make their business case fly, after all about unrealistic assumptions, then they can tell all of Australia “Faster, Cheaper Broadband” like numbnuts Mathias Cormann keeps saying in parliament.

    The realistic speed is likely to be the worst case so I would expect 100Mbps to be no more than 5Mbps.

  23. I’m sick of reading about these “tests” that cherry pick 1 unit or a premises that is basically sitting right next to the node and calling it a real world example. If this was a scientific research paper the scientific community would rip it apart.
    If you’re going to get serious about FTTN testing, can someone please do at least a few whole streets in various brown field landscapes and then show the various results for each premises.

  24. This is absolutely appalling!

    One single site running off a node is NOT in any stretch of the imagination “a valid test”.

    Show me the results when you have A FULL NODE of customers all accessing the service simultaneously (as fas as they can), and give us actual in deployment real world results.

    I can absolutely guarantee you will see performance drops.

    By Definition: on *DSL networks interference from nearby services leads to all services suffering degradation. (as opposed to FTTP networks where they do NOT, ever, interfere).

  25. Has anyone considered the power consumption and carbon footprint of running 70,000. nodes 24/7

    • NO because NBN wont be paying for it, you will be. This is the Coalition your talking about, carbon! what carbon?
      These idiots are going to spend a shit load of our money on something that will be replaced 5 years after completion. If your looking for logic and reasoning, look else where because this mod sold out to Murdoch.

    • There has been an estimate of around $20 million of electricity a year assuming 70,000 nodes. Not much at all in the scheme of things.

      Though from looking at the practicalities I can’t see there being so few nodes needed. I would like to see good news on speeds reached on realistic cable lengths of real-world cable gauge and condition with a realistic number of nodes and end-user connections running concurrently over a statistically significant range of sites, but there isn’t any. If there is a result from this particular trial that can be applied to the actual deployment, I would love to hear about it.

      • Is there a link to this $20million estimate? It seems a little on the low side, I mean less than $1/day for each node.

      • There has been an estimate of around $20 million of electricity a year assuming 70,000 nodes. Not much at all in the scheme of things.

        Are you sure about your numbers? $20 million for 70,000 nodes doesn’t pass even a basic analysis.

        Being generous with rounding, leap years and energy cost a back of the envelope calculation/sanity check is…

        $20,000,000 / 70,000 nodes ~= $285 per node per year

        $285 / 366 days ~= $0.77 per node per day

        Assuming $0.10 per kWh,

        $0.77 / $0.10 per kWh ~= 7.7 kWh per node per day

        7.7 kWh / 24 ~= 0.32 kW or 320 Watts

        Neglecting power supply inefficiency that means that all the following has to be powered on less than a modern Desktop PC power supply, ie 320Watts
        * AC/DC rectification (N+1 redundant?)
        * Cabinet cooling / fans (N+1 redundant?)
        * Battery charging
        * 192 VDSL subscriber lines
        * POTS replacement

        I think it is pretty clear that $20 million is a long way off as an estimate of the energy costs for a Node.

        320 Watts / 192 ~- 1.66 Watts per line (ignoring all other node/chassis power requirements).

        Even the VDSL2 modem an enduser will use to connect to the node will be up to 10 times that, and remember most estimates for node power requirements often overlook the impact of vectoring. Vectored lines will definitely be higher power than non-vectored lines, this is self-evident if you understand how vectoring works.

        Please remember that every stage of calculation has been generously rounded down or underestimated in your favour MartyvH

        • There’s a couple of things to note:

          I don’t believe FTTN nodes require hard drives and they are the only moving part in a PC (your comparison was with a PC) contributing significantly to the PC’s power usage: it takes a lot of energy to keep that disk spinning at 10k-15k rpm and also moving the heads back and forth in what looks like ‘instant’ movement – but of course isn’t.

          New micronodes are the latest generation of FTTN nodes – much smaller that the ‘bar fridge’ size nodes and they use more modern, power saving electronics which means they can be cooled by natural convention instead of fans – reducing the energy consumption even more.

          They can also be much more easily mounted – not requiring a large concrete slab.

        • You also need to consider that passive GPON ‘nodes’ use splitting (20 way) technology which split the light signal to up to 20 premises meaning that you need to use a much more powerful light source at the exchange end of each fiber that you need with FTTN fibers.

          This splitting also becomes (in the future) a constraint on the maximum speed you can achieve with FTTP because whatever bandwidth the fiber can handle you need to divide that by 20 to calculate the maximum for a single premise.

          FTTP via GPON also has the negative ‘feature’ that someone on the shared fiber could, with the right hardware, potentially see data that is being set to the other 19 premises on that shared fiber.

          • You also need to consider that passive GPON ‘nodes’ use splitting (20 way) technology which split the light signal to up to 20 premises meaning that you need to use a much more powerful light source at the exchange end of each fiber that you need with FTTN fibers.

            Sure, that is all accounted for in the optical budget when the network was designed.

            Once again, if you refer to the paper I linked you to before, Table I on page 5 of http://www.researchgate.net/publication/220144362_Power_consumption_in_telecommunication_networks_overview_and_reduction_strategies/file/32bfe510be5bceb295.pdf shows GPON to be 0.4-1.6W compared to 6-10W for FTTN.

            This splitting also becomes (in the future) a constraint on the maximum speed you can achieve with FTTP because whatever bandwidth the fiber can handle you need to divide that by 20 to calculate the maximum for a single premise.

            You are right, the fibre is optically split and shared amongst all downstream. This does reduce the range of the fibre and the speed.

            But given PON networks can currently do more than 10/2.5Gbps or 10/10Gbps (yes gigabits per second) per user over 15km, which is much further than ADSL was ever capable of it doesn’t seem much of a problem.

            But it doesn’t end there, there is also the option for them to go 40Gbps or more.

            But I suspect you may have already known that.

            FTTP via GPON also has the negative ‘feature’ that someone on the shared fiber could, with the right hardware, potentially see data that is being set to the other 19 premises on that shared fiber.

            Once again true, but the same can be said for wireless technologies can’t it, like wifi, 3G, 4G or whatever 5G will be. But that is why they use encryption to ensure privacy, which is more than can be said for many wifi networks.

            Oh, and do you really think you can’t tap the copper network “with the right hardware”? Just pop the top of the pit and away you go.

            Hell, I even know of some private investigators (PI) that have tapped someone’s home ADSL at the madison box.

          • Yes it’s true they anyone could tap the existing network but it requires someone to tamper with assets on the street which makes it a lot more noticeable than someone interpreting the data from 19 other subscribers on their same GPON that is delivered right into their home – much less visible tampering required and zero evidence left in public assets that snooping is taking place because it can happen right in the comfort of their home and no one would no.

  26. I live about 1.2km as the cable goes from my house to my supernode in canberra. Transact have mostly cat5 cable with some cat3 in their vdsl2 network. I get 43mb down and 5mb up.If they install vdsl2 at the node i will only be about 100m away and should get a lot faster speeds.I cant see the government having anywhere near the quality of Transacts network for the rest of Australia so I bet speeds will be a whole lot worse.

  27. I’m not sure if you have heard about this relatively new FTTN technology call ‘micro nodes’ but they could significantly change the FTTN/VDSL2 landscape for the better.
    They are much smaller nodes, much cheaper and passivley cooled. They are small enough to be installed without council approval in most places and installing smaller, cheaper nodes should allow more premises to be much closer to their node that if we installed the older technology ‘fridge’ sized nodes. Have a read, it seems like these micronodes are putting copper back into the conversation for a whole lot of telcos around the world. At a industry show last year apparently the ‘buzz’ was all about copper and not fiber because it means telcos can deliver amazing speeds at a fraction of the cost and fraction of the rollout time of a full fiber (FTTP) solution:
    http://next-generation-communications.tmcnet.com/topics/service-provider/articles/359995-vdsl2-micro-nodes-make-copper-important-element-ultra.htm

    • It’s still not a long term solution (even FTTdp will need to be upgraded in the not-so-distant future), but seems like a better idea than Malcolm’s FTTN brain fart.

  28. At a industry show last year apparently the ‘buzz’ was all about copper and not fiber because it means telcos can deliver amazing speeds at a fraction of the cost and fraction of the rollout time of a full fiber (FTTP) solution:

    You forgot to add “… for telcos that already own the copper.” Which is not the case in Australia.

Comments are closed.