So is G.Fast a thing or not?



blog Those who were closely observing Malcolm Turnbull’s debate with Business Spectator publisher Alan Kohler yesterday (you can watch the recording here) will have noted Turnbull’s reference to an emerging standard known as G.Fast. Turnbull made the somewhat controversial claim, according to ZDNet (which has an excellent article on the subject here) that G.Fast could eventually spur 1Gbps speeds over FTTN networks. But is G.Fast a reality, or another broadband pipe dream? To our rescue comes Informa analyst Tony Brown, who has published an extensive analysis on the subject on the website of the Sydney Morning Herald. Brown writes (we recommend you click here for the full article) that there are limitations to the technology, but that ultimately, it could be a very valid part of the NBN rollout. One of his points:

“For NBN Co a G.Fast deployment in MDU’s would be an advantage because the company could then connect MDU subscribers to the NBN without going through the tortuous process of gaining permission to enter premises from building owners and landlords – let alone scheduling installation appointments with occupants.”

Given the continual problems which NBN Co is facing with multi-dwelling units (problems which, you may recall, were also suffered by Telstra and Optus with relation to their HFC cable rollouts), there is a growing body of commentary calling for the company to at least consider doing something like fibre to the basement and copper to individual apartments, and it’s hard to disagree with this logic. Turnbull’s right, to a certain extent, and so is Brown — if you can get fibre to the basement of apartment buildings, the copper inside those buildings is usually able to do a decent job of getting broadband to individual apartments, with the aid of standards such as G.Fast. And if needs replacing, there are existing mechanisms for that to happen too (ours did a while back, and the landlord just called Telstra to replace the copper). But we suspect this argument is not going to go down well with those who insist on fibre everywhere, all the time. It’s a tough world for any copper argument to get support, right now.

Image credit: Clix, royalty free


  1. Well, there is continuous movement on the SC0 front, with about 100/day being removed from the list. The total is only growing due to the increase number of premises passes.

    Secondly, there is only about 500k really high density brownfields premises anyway, in maybe 10-20k sites. FTTB might be of use there, but that also then causes problems with the phone connection, multiport capabilities and achieving real speeds rather than “up to” speeds.

    The 2 million or so low rise and town house MDU’s are far easier to rollout fibre to and why introduce the complexity of two systems when with a bit of time and money can provide a long term solution.

  2. The coalition could do stuff like this to improve the rollout of FttP. It’s dissappoint that both sides of politics are pushing an ideological approach instead of a pragmatic one.

    Somewhere between both policies is a workable future proof solution. FttN to 71% of the country is a waste of money (the investment is similar to Fibre and the time savings aren’t massive, so what’s the point), FttP to 93% of the country is time consuming and difficult to achieve.

  3. one thing i don’t get about the whole MDU thing is that it’s surely massively in the interest of the owners to facilitate getting NBNco fibre to all the apartments because it increases the value of the properties?

    • The problem isn’t with individual owners, it’s the difficulty of getting strata bodies to come to consensus on the issue. What might make logical sense is quite irrelevant when you have people who are simply opposed to change of any form.

      • Precisely – seems to be the norm that there is at least one idiot on a power trip who wants to make life difficult for everyone else just to make them grovel. The fact that they are sad, pathetic twats doesn’t change the fact that it takes just one AH like that to screw things up for everyone. In the case of NBN connection residents who want/need it should be able to bring suit against those who make connection impossible, particularly if it means they miss free connection on the initial rollout. It’s why the federal government should have legislation in place to circumvent strata veto in this circumstance – only those residents/apartment owners who specifically opt out should remain unconnected. They shouldn’t be allowed to f&#@ things up for everyone else.

      • Strata groups are funny: you always have one young hair dresser versus an old lady in her 80s who finished hairdressing before the young one was born and did the strata group for free for the last 15 years slugging it out over fees and the old lady always wins after everyone else comes to their senses 12 months later and realises: no, fees weren’t meant to be paid for a voluntary position! Good Karma that one, yehhh!! (Funny but!!!)

  4. I’m surprised that only ZDnet has the tech:
    “The International Telecommunications Union is moving toward making a DSL standard by March 2014. promises fibre-to-the-premises (FttP) speeds over short distances of copper through a combination of DSL technologies, including pair-bonding, vectoring to eliminate cross-talk on VDSL2, and “phantom mode”, which creates virtual pairs between copper pairs.”

    Anyone believe that MT is going to find many sites with available copper for bonding and vectoring within 100m or so?

    Whistling dixie methinks.

    • That is the thing isn’t it? Not to mention that there will need to be how many extra nodes? Double? Triple? (I honestly don’t know) This is an extra expense, plus powering them. Also, is there enough copper capable in premises to make this a reality? So far it just sounds as if Mal T is trying to null the advantages of Fibre now with something that might be able to be done 10 years from now (atleast in our nation).

      • He wants to install extra nodes and run extra copper to use a technology which isn’t even standard yet, just to do what we could do by running fibre instead?

        Yeah you’re right Renai, I can’t see that getting a lot of support.

        • He’ll undoubtedly get the support of the same minions who blindly support him now, I’d imagine.

  5. Just as a “in the interest of strict correctness” thing Renai:

    While G.Fast has been shown by AL to give upwards of 1Gbps (up to 1.3Gbps in fact), that was with 2 pairs of copper (bonding), phantomming AND vectoring.

    Without bonding (and therefore without phantomming as it requires at least 2 pairs as well) it has shown a maximum possible 500Mbps with vectoring and 330Mbps without, on 100m and approx. 160Mbps on 200m.

    1Gbps speeds would not be possible in the vast majority of MDUs because of single copper runs removing the possibility of bonding/phantomming. 500Mbps would be the best possible speed on lines 20 units) only make up around 500-750K actual premises or around 7-8%. The other 20-25% of MDUs it would likely end up costing almost as much as FTTP because of the number of G.Fast modules needed due to low density (and for the fact that G.Fast doesn’t even have commercial gear yet, so it will be HIDEOUSLY expensive to start with so early)

    I think Vectored VDSL in older and more difficult MDU’s is feasible. But I also wouldn’t want to be living in them if something went wrong with the internal wiring and their speeds went from 100Mbps to 5Mbps….who’s job is it to fix?…..

    • Not sure what happened in that comment….a sentence appears to have disappeared… the 4th paragraph was meant to read:

      1Gbps speeds would not be possible in the vast majority of MDUs because of single copper runs removing the possibility of bonding/phantomming. 500Mbps would be the best possible speed on less than 100m lines. But, as someone has already said, those 1/5 of MDUs that have the density to use G.Fast effectively (>20 units) only make up around 500-750K actual premises or around 7-8%. The other 20-25% of MDUs it would likely end up costing almost as much as FTTP because of the number of G.Fast modules needed due to low density (and for the fact that G.Fast doesn’t even have commercial gear yet, so it will be HIDEOUSLY expensive to start with so early)

    • In all the places I worked (and some I lived), there were always 2 pairs running to every unit from the MDU, but only 1 was connected. 2pr sheathed solid wire is pretty much the standard cable used to wire phone points in Australia, and has been for a long, long time.

        • @Dan

          I never meant to say they wouldnt HAVE 2 pairs- 90% of the copper footprint did to begin with (100% redundancy). But how many WORK? The numbers I’ve been given from 2 separate analysts are anywhere from as many as 60% and as few as 40% of all second pairs (in premises overall) are unfit to use or have been cut.

          Would some MDUs have second pairs that work? Definitely. But they would then need special treatment to bond them together on connection and other factors such as POTS would come into play. It’s not impossible, but adds ANOTHER layer of complexity- in a single MDU 50% may have bondable pairs, but that’s just luck of the draw, something I would wholeheartedly like to leave behind to ADSL and the noughties.

          • Thanks seven_tech for yet another great post, with actual facts that put things into perspective.

  6. Sortius posted an interesting tech article on this subject earlier, it suggested that G.Fast is only really economically viable on lines of less than 200m. How many nodes are we going to build before we decide to do it once properly, with fibre?

  7. “For NBN Co a G.Fast deployment in MDU’s would be an advantage because the company could then connect MDU subscribers to the NBN without going through the tortuous process of gaining permission to enter premises from building owners and landlords – let alone scheduling installation appointments with occupants.”

    That’s and oxymoron.

    Yes, the clear advantage is for MDUs, as you would expect to see cabinets every 100m down a street.
    The effective use however is in the basement of the MDU. The basement is still part of the property.
    Getting permission from building owners and body corporate would still be required before that happens.
    And uses the existing copper infrastructure, permission and scheduling installation appointments with occupants is required, otherwise tenants will find their communication cut off without warning.
    You can imagine how much fuss they would raise if they lost their internet or phone without their consent.

    • Bzzt! Wrong. (Or more politely, “I disagree with you wholeheartedly”)

      There would be virtually no requirement to obtain anything more than committee approval for the NBN contractor to install a node in a basement, and only then to facilitate the establishment of an electrical feed to power the thing. Most, if not all multilevel strata buildings have space(s) set aside for ‘electrical and communications equipment’ in the basement or car park and that area is formally assigned for that precise purpose in the registered community title. Smaller strata developments of around 8 units would simply have the node installed on the side of the building, or if no committee approval was forthcoming, on the existing council property in front of the building. The only notice the owners or tenant would get is that there would be an interruption to their phone/internet service on a particular day if it required the system be isolated for safety reasons. The existing copper lines would not be ‘cut’ from the old infrastructure until the new connection to the NBN was requested by the owner/tenant as the copper from the MDF/node to the unit would be the same in both the old and new systems.

      So yes, Mr. Brown is indeed correct insofar as claiming G.Fast would help NBN Co in MDUs. Mr. Turnbull may like to think G.Fast would help an FTTN NBN, but in fact it would only further exacerbate the issue of distance from the node, as G.Fast only operates up to 100m. 70% of the population within 100m of a node would mean 10-20x the number of nodes, and therefore 10-20x the cost of said nodes. Currently, at $4.5bn, that would then add another $40bn minimum to the build of an FTTN NBN.

      • There is one catch is all the standards use system that need access to the all the copper in the bundle to calculate out noise from crosstalk so all copper connection will need to be cut over new the new hardware(or it won’t work as well) causing a disruption(this was one of the outright lies in the LNP initial document). The advantage is that they won’t need to access each unit and pull fibre through the whole building for an installation.

  8. Two things here.

    The first is that Malcolm Turnbull is just throwing out another technology to make it seem like his plan can provide much more than it will actually be able to; yesterday it was “vectoring”, today it’s “G.Fast”. Note that he’s just throwing these things out as a “possibility” (we “could” do these things, they “might” deliver what we promise), and doesn’t have even a barebones plan for implementation.

    The second thing is Renai’s point with regards to MDUs, and it runs parallel to the larger NBN debate, and it’s that we can either choose a shorter term solution (FTTB and G.Fast ; FTTN) that looks nice now, or we can stick it out and wait for connections and it will be much better in the long run (getting those MDUs fibred up ; FTTP).
    Having said that, if G.Fast/FTTB delivers as promised, it’s a far better compromise for MDUs, than FTTN is a compromise for premises in general. However, reading the comments made by others, it sounds like there’s a lot of caveats.

    • “Malcolm Turnbull is just throwing out another technology to make it seem like his plan can provide much more than it will actually be able to; yesterday it was “vectoring”, today it’s “G.Fast”.”
      Oh for sure. If you follow his twitter you will see someone mention these techs to him with a link to an article of two, usually ones of his friendly analysts, and suddenly he is mentioning those techs in his next interview. Saw it with both vectoring and G.Fast. Follow his twitter, a lot of what he is saying is feed to him through it.

      • and Mini nodes in pits was mentioned by NBNCo in senate estimates and then he is going on like they are the best thing since sliced bread

    • The reason he does is because is whole strategy is based on “why go fibre now, when something better might come along”. His aim is to give the impression that something better than fibre may come along. This is likely to work with what they call ‘low information” voters. It also gives the impression that the shelf life of copper is limitless.

      • Yes it really is a contradictory statement by MT on multiple levels.. (as is the case with his usual suspect trusty followers here too too)…

        To harp on about “something better coming along” but …

        A) largely bypassing what is the current better and relying on copper.

        B) telling us xMbps (which fits his policy) is as fast as anyone will really ever need, but then jumping on the latest fad because it claims it will go way faster than… err what we ever need…

        On top of the $94B and TA’s 80 years claims, this is just more bottom of the barrel stuff, imo…

  9. I have no doubt FTTB will have to be considered by NBNCo because of strata issues, and is an enabler. However to quote The Castle, Turnbull is dreaming if he thinks G.Fast would provide a cost effective upgrade path for FTTN when compared to FTTP.

    In the most simple terms: the population density will not be high enough to put micro nodes every 200m compared to installing passive drops.

  10. I believe that on private property the owners of that property should install the cabling systems at their own expense and nodes installed on the property under the Telecommunications Act by NBN Co bypassing any need to negotiate with strata associations. There should have been legislation under the Telecommunications Act for MDU’s to have an annual vote for the installation of fibre, and a majority verdict with NO PROXIES allowed as sufficient to pass installation, NBN Co should have offered loans to the strata association for the installation of fibre.

  11. “But we suspect this argument is not going to go down well with those who insist on fibre everywhere, all the time.”

    But is is not fibre everywhere now, nor will it be (at least in the short to medium term).

    ” It’s a tough world for any copper argument to get support, right now.”

    It would be a much easier world if the copper was in very good condition.

  12. In a way MT’s approach is similar to have an old rusty car and saying it can be repaired for a fraction less than the cost of the new one.
    Then, reading magazines which talk about new technologies in the horizon and justifying that small saving by saying “here you go, I could make it even closer to a new car in the future” but forgetting that doing so would obliterate the saving and that waiting for the new technology, to be commercially available, would also mean that the car is getting older, and more likely to develop new problems.

  13. Let us remember the KISS principle, Keep It Simple Stupid. The more complexities and variables involved and interacting the greater the scope for worst case costly issues and disasters or even just expensive maintenance and reliability issues

  14. “And if needs replacing, there are existing mechanisms for that to happen too”

    I would assume that would include like for like only.

    Cable runs in most multi-story apartments blocks would require cat 6 cable to provide worthwhile speed improvement with F-Fast and this is not like for like.
    If you can manage non like for like cabling, why not fibre?

  15. So is Malcolm now talking about putting nodes every 150-200m as requires? Talk about policy on the run…

  16. Does anyone know what type of Cable is deployed in your typical MDU?

    I’ve heard that a large proportion of Australias copper is 0.4mm.

    From Alcatel Lucent, I found the following line:
    “This is why is seen as a technology that is intended to operate over loops that are less than 250 m on 0.5 mm cable. – from: here []

    I suspect that cable-types not just loop length is going to factor into speeds.

    That same article indicates 800 megabits over 100meters for a single pair. I suspect this is not bi-directional. And, since you can’t 100% rely on 2 pairs going to anyones property, you could only sell this is a single-line technology.

    This is probably a god-send for the MDU’s, but this is clearly not a FTTN technology, this is a FTTB technology. If Mal thinks it helps his FTTN he is wrong, we are OK with speeds FTTB (followed by vectored vdsl 2) gives in sub-100 meter loops.

    The problem isn’t FTTN in basements, its FTTN everywhere else that his solution is penny-pinching to the detriment of value.

    • I think you’ll find there is a wide variety of copper to residences from the MDU. Not saying it’s the majority of installs, but you’ll often find Cat5 – which makes bonding an option across it’s multiple pairs.

      For mine, NBNCo should forget copper and run fibre to each residence. I’m sure there will be challenges getting it into some, but that’s offset by not having to manage DSLAMs all around town on top of GPON, LTE and Satellite services! That, and fibre is going to have to be installed sooner or later – 5, 10, even 15 years time. Might as well install now.

      • Yeah, my point is no speed is guaranteed under any of these circumstances.

        If you’d love to be confused; try and figure out exactly what the whole thing is talking about on this page:

        In it; they talk about 300 megabits over 400 meters using a “quad pair”, and proceed to explain how you get 300 megabits over 100 meters using 2 pairs in their wonderful infographic (which almost looks like 4 pairs, but is labeled as 2).

        Strikes me like a bunch of salesmen trying to sell a big expensive network to people that don’t really understand anything. The fact that Malcolm Turnbull is going around advertising for them does not make me comfortable.

        • I take a bit of that back, its very confusing. (they say quad-pairs a few times, but I think they just mean Quad cables).

          What I want to know is, anyone in the world actually deploying / testing real-world vectoring?. Anyone at all?

  17. Having followed G.FAST (Fast Access to Subscriber Terminals) quite closely since it was proposed by BT and AT&T around three years ago and adopted by SG15 for standardisation I have been rather surprised that it has been absent from the “informed” debates on copper and fibre technologies for this long.
    The original idea of G.FAST was as to support FTTH as a FTTdp (Distribution Pillar) system, and for FTTN (loop shortened to the pillar) where 8 properties would be served from a DP, using the installed drop wire, founded on two key requirements: (1) that there be a reverse power feed arrangement form the home(s) served (ETSI have worked on this aspect), and (2) that it use frequencies above xDSL (17.644MHz for typical VDSL2 deployment and configurable down to 2MHz). The original thinking was that it would use VDSL2 or for L1, but in the event a TDD mechanism (90/10 to 50/50 asymmetry allowed) was proposed and adopted and it is this foundation that is working its way through the standardisation process, complete with all the integration considerations from associated bodies such as the BBF.
    Performance objectives have been based on cable modelling and provide for aggregated US+DS of 150M@250m, 200M@200m, 500M@100m and 500-1000M@<100m for FTTB. The PSD for the technology has just been published in Determined Recommendation ITU-T G.9700. It is useful to recognise that G.FAST reuses the learning’s from xDSL,, vectoring, and the same group of experts who drafted those standards, so that matters such as low power operation, latency objectives, retransmission etc are being accounted for at L1. Another architectural objective has been to support an ATA function at the DP, to allow for a complete self-install model to apply – the narrow band POTS and the data served reusing the standard ADSL filter/splitter techniques of today.
    The overall impact is that G.FAST would (1) isolate the dodgy internal wiring of the home from an FTTN build, where the loop is shortened and thus provide higher speed and (2) avoid the costly install of the fibre drop and ONT in the home for FTTH. For an MDU there have been proposals to increase the drop count as the potential benefits and issues associated with other alternatives have been considered. For the FTTN scenario it is conceivable that phantom vectoring and pair bonding to feed the DP G.FAST FTU-O would provide significant shared bandwidth and deliver overall performance improvement compared to the standard FTTN deployment, potentially delaying the need for a deeper fibre build.
    The concept of integrating the FTU-R (customer G.FAST modem) into a RGW that also has a VDSL2 modem and changeover relay would allow the concept of a customer installed device that can work as VDSL2 direct to a Node, or to the curb based FTO-O when installed and activated, with the changeover to G.FAST and FTTdp not requiring any change for the customer.
    Based on the development cycles I have observed for VDSL2, vectoring, and I would expect the first commercial systems to appear around 3 to 4 years after standardisation, so the usefulness is arguably toward the tail of the fast BB deployment plans given a standard 1 year of network integration testing before scaled deployment – let’s say 2014+5years = 2019.
    Timing aside, there have been views expressed by those close to the technology that suggest this be a permanent technology solution rather than an interim solution, given the bandwidth being supported and expected increases from pair bonding, pre compensation and cancellation developments – which is pretty understandable given the domination of either copper to the device (cat5/6) or WiFi to the device. I don’t see much interest in running fibre to the tablet, to the TV or the PC in a typical home networking environment.

    • @Peter Crosby

      I don’t think anyone doubts that G.FAST has been designed to target the problems FTTN has faced. I think what people are arguing is that after ALL that is considered and ALL examples and niche customisations are taken into account…it’s within 10-20% of FTTP anyway, so for a GOVERNMENT provider, why not just do it with FTTH? For commercial providers, sure, it’s a great boon for customers and CAPEX. But NBNCo. ISN’T a commercial provider and that’s what everyone keeps missing.

      As you say, G.FAST would likely not be able to be widely deployed for years anyway, so the point is likely moot.

      • This.

        The savings just don’t exist (at a node per 8 homes – that’s within 100 meters of every home in a suburban street!) you won’t be seeing a network that costs 29.5 billion dollars, nor one that is built significantly faster than a full FTTP/B network.

        That isn’t to say this technology doesn’t make perfect sense for basement installs.

        Finally, you don’t see people putting fibre into their tablet, but people put a copper cable into their TV right now [Foxtel], and while many don’t like the business model, many more don’t seem to mind.

        • At ~1 million nodes, a FTTN would easily double the cost of his network, and still not offer the speed/reliability of fibre.

      • As you say, G.FAST would likely not be able to be widely deployed for years anyway,

        Two things to this:

        1. If Malcolm want’s to use it for his FTTN, he’d need to plan ahead. It’s only any good for 100-150m.

        2. if he just wants to use it for MDU’s, he should stop talking like it will be used on his general roll out…

        He knows these things, why is he misleading the public so blatantly, surely it will blow up in his face in a spectacular fashion if the LNP win and folk’s figure out he was talking about “hypothetically”.

        • Let’s face it guys…

          We can sugar coat it all we like, but this is just another, the house needs to be knocked down and another built… but let’s just whack a new coat of paint on it instead…excuse, simply to try to justify anything “but” admitting, hmm they actually got it right!

  18. Malcolm Turnbull: “If we buy a lot of high-speed trains, we’ll be able to go from Sydney to Melbourne in only 2.5 hours!”

    Reporter the likes of which does not exist in Australia: “But you’ll need to invest in rail infrastructure too, right?”

    Malcolm Turnbull: “Why are you such being a zealot using stupid and quasi-religious arguments!!??!! The rail is fine.”

  19. These attempts to find ways of improving the performance of copper are understandable for an existing network. Getting the maximum out of your initial investment is logical, providing overall the network is still in good condition and well maintained. It makes not sense to apply this principle when the network belongs to a private provider and the goal is to build a network that will provide for the next few decades. This makes even less sense when the cost becomes greater than rolling out fibre and it results in greater technical complexity which will still play catch up with the possibilities of fibre.

  20. looks great. For specific purposes. Problem with this is that, like a lot of suggestions from Turnbull, they don’t generally work as well outside of the Lab.

    A concept that is good, pressed into being a “defacto” for the rest of the network. I’m sure Vectoring started out as a potential solution to a portion of the network – now it’s likely to be required across the entire network to allow the Policy speeds to be realised.

    It’s one thing to see someone race a ferrari on a tightly controlled test track, quite another to take it for a spin across the Nullarbor, both paved and gravel, and expect exactly the same outcome. Or worse, present it as though it most certainly will be the same.

    Australia isn’t a Lab; it’s hard to imagine an average MDU would match lab conditions, either. And the range is very tight for higher speeds. This is sensitive to line length and quality even more so that straight vectoring is.

    I can see how this actually might be a valid option for MDUs. But it’s likely to require quite a bit of copper to be re-run. If that can achieve a better outcome over Fibre, great. Awesome.

    But if it ends up the same cost, as dropping fibre runs, the question of longevity and upgrade viability comes to mind.

Comments are closed.