G.fast: a high speed cul-de-sac

The other day I received a marketing missive from Virgin Media about my broadband service - they’d tweaked their network and bumped me up to a 120Mbps service. They suggested I should run a speedtest.net to see for myself how fast this might be. So I did a couple, one with my laptop connected to the router over wifi, the other using an ethernet cable to the router, and here are the results:

Speedtests 2.png

125Mbps downstream is about 10 x the national average, 11.55Mbps upstream isn’t too shabby either. Of course, it would be better if the service was a symmetrical 125Mbps, but then the nature of DOCSISv3 over copper coax cable that Virgin Media uses prevents such a service being offered. 

But even so, according to a report from Ofcom,  I’m currently in the top 5% of Virgin’s customer base as far as speed is concerned. In fact, from the graph below you can see there’s been a massive upgrade in speeds offered to customers in the last 12 months, from 25% to 64% with 20/30Mbps and from 0% to 22% with 60Mbps. Based on my experience, Virgin have been proactively upgrading customers, migrating them to higher speeds.

Virgin customer base.png

However, these speed bumps, and the rather impressive 125Mbps I’m experiencing, only apply to customers on their copper coax-cable network, something they market as “fibre optic broadband”. In my opinion, it isn’t fibre optic broadband because it’s delivered on copper coaxial cable and not optical fibres, as the map of their network, and this Twitter conversation with Virgin shows:

The VirginMedia map is headed "Cable Modem Coverage". A cable modem is a modem used for coax-cable networks, not fibre optic networks. I think this is known as a smoking gun.

The VirginMedia map is headed "Cable Modem Coverage". A cable modem is a modem used for coax-cable networks, not fibre optic networks. I think this is known as a smoking gun.

The customer services rep that replied on Twitter may well be right - it’s a relatively short length of copper coax cable to connect to the home, in much the same way as BT uses relatively short lengths of copper wire to connect to the home instead of optical fibres for their so-called “fibre optic broadband” services, but this copper changes everything. Light in fibre is changed to electricity over copper, with different behaviours, which you can read more about on this blog.

Customers in the grey areas of the country in Virgin’s map are reached in the same way as all the other service providers in the UK - either reselling BT’s copper wires network or using their own copper wire network. There are one or two exceptions to this (Hyperoptic and Gigler) which I’ll talk about later. 

1000Mbps over copper - is optical fibre dead?

There have recently been press releases about an impressive 1000Mbps (1Gbps) being achieved over twisted pair copper wires. This is about 100 x the current national average for download speeds and on the face of it this satisfies the clamour for speeds normally only available with fibre optics to the home (FTTH). It’s not the fibre that people need, it’s the speed that until recently only fibre could achieve, and if a copper network can deliver the same speed as optical fibre, then there’s no need for the expense of creating a true optical fibre network. Is there?

A new signaling system called G.Fast achieves this spectacular speed by building on the current fastest broadband technology over copper wires, VDSL2, widely used in BT’s fibre to the cabinet (FTTC) “fibre optic” broadband service. Currently, optical fibres are terminated in anonymous green cabinets in streets. The optical signals are converted to electrical signals (VDSL2) within the street-side cabinets, and these are then delivered to houses over copper wires, the same copper wires that have been in place for years delivering voice calls.

I’ve discussed electrical conductive and resistive properties of copper wires in previous blog entries that explain why there’s always the “up to” speed “gotcha” from service providers’ marketing departments (something that would be unnecessary in a true fibre optic broadband world). See the following two screenshots from BT’s current website that neatly illustrate this point:

Two different screenshots showing the current confusion from BT about what BT Infinity actually provides. I think it’s disingenuous of BT to say that standard broadband is delivered completely over copper wires as this just isn’t the case - there's plenty of optical fibres going to the exchanges where older broadband is served from. The increase in quantity of optical fibre for Infinity is marginal.

Two different screenshots showing the current confusion from BT about what BT Infinity actually provides. I think it’s disingenuous of BT to say that standard broadband is delivered completely over copper wires as this just isn’t the case - there's plenty of optical fibres going to the exchanges where older broadband is served from. The increase in quantity of optical fibre for Infinity is marginal.

The result of fundamental laws of physics as applied to VDSL2 is represented by this graphic (thanks to Ian @ http://www.creativeimpetus.co.uk for helping with the visualisation), which shows the rate of speed drop off with length of copper wire from a VDSL2 equipped street-side cabinet, located in the middle of Slough.

VDSL final.png

The electrical signals (frequency) and thus the speed of broadband, runs out of steam very quickly, indeed the faster they start then the faster they run out of steam. So if you happen to live close to one of these street-side cabinets, you’ll get download speeds of around 80Mbps, but notice how quickly the speed reduces with distance, 1000m out and the speed has dropped by 66%. (the source data for this graphic came from from BT).

“So what?” you might well ask - after all Slough is obviously served by more than one street side cabinet, so the average distance most residents in Slough will be from one will be much less than 1Km, so the effect of the copper wire will be much less and as a result their actual speed will be somewhere in the 100Mbps to 30Mbps range, and that’s OK, right?

Well, not really.

Even if every streetside cabinet were fibre-enabled (which they aren’t) there would still be consumer confusion about what speeds they will actually get for their money. It’s like filling up with petrol but not knowing exactly how many litres you’ll get for your £1.42, a situation that would be untenable and intolerable. 

But also, and perhaps more to the point, the bandwidth delivered just isn’t enough. 

Experts have calculated that 24Mbps download is perfectly fine for simultaneously streaming multiple HD movies - and this may well be the case. The reason video is often used as a benchmark is because this is a bandwidth-intensive application and the higher the resolution of the video (Standard Definition to High Definition to Blu-Ray) the more bandwidth intensive it becomes. But there’s a but.

Download graph.png

Having performed the speedtests illustrated earlier, I upgraded the OS on two identical Apple MacBook Pros, a process that involved downloading a 4.4GB file - 4.4GB! The first machine to be upgraded was connected to the router via wifi, the download took 36 minutes. The second laptop was connected directly to the router by a 1000Mbps Ethernet cable and the download took 12 minutes. Why should it take 12 minutes? 1.2 minutes or even 1.2 seconds would be much more acceptable.

The killer app, the need for speed, isn’t video, although that’s an increasingly important part of the equation with services such as BT Sport being launched. The real need for what we regard today as unbelievably fast broadband (UFB?) is life. Just everyday ordinary life in the 21st century.

Is G.Fast the answer? Alas, no.

G.Fast achieves its speed by combining VDSL2 with other technologies such as pair-bonding (bonding two copper pairs together), vectoring (this techniques eliminates cross talk or interference between different VDSL2 lines) and “phantom mode” (creating virtual pairs between real copper pairs). Note that none of these tackle the root cause of signal loss: impedance.

This high data rate requires even higher frequencies than normal VDSL2 (so much so that they can interfere with DAB radio). Because of these ultra-high frequencies, the impeding effects of copper are even more effective, resulting in a much reduced distance that the signals can actually travel. This means that the switching equipment (DSLAMs) needs to be located even closer to the home.

Hence we have another FTT acronym - FTTdp, or Fibre To The drop point, or telegraph pole as they are most commonly known. The switching equipment (DSLAMs) will be installed on telegraph poles instead of within street-side cabinets.

Each telegraph pole mounted DLSAM will support from 1 to 16 ports (or end users). I was told by a dyed-in-the-wool BT engineer that not all telegraph poles will be able to support the extra loading generated by this new equipment and it’s possible that some telegraph poles will have to be replaced, thus increasing the cost significantly.

According to a 2012 joint ITU/IEEE presentation “G.Fast for FTTdp”,  the bandwidth achieved will be approximately:

  • 500Mbps @ 100m
  • 200Mbps @ 200m
  • 150Mbps @ 250m

So there’s still uncertainty and doubt about the speed that consumers will get.


Here’s a picture of the telegraph pole at the top of my drive. Notice the lack of power sockets to provide the required power to these new micro-DSLAMs. The proposal for electricity supply to these “DSLAMs-on-a-stick” is that they are powered by “reverse power from the customer’s residential gateway” - in other words the necessary new routers (Home Hubs) needed in homes will also feed power back from the home to the DSLAMs. We, the end users, will be powering BT’s network (assuming BT deploys this technology which their Bill Murphy has said publicly that they will).

Somehow, this just doesn’t seem right to me.

The technical ITU/IEEE presentation does say that G.Fast will be backwards compatible with existing routers (CPE) in the home, but of course as these don’t support the new technologies of pair-bonding, phantom mode and vectoring, there won’t be much of an improvement over existing speeds for those customers who don’t want to pay for an upgrade.

So problems abound with this technological cul-de-sac, but the biggest problem of all is the cost. Yes, maybe it does sweat that copper asset for longer, but even if the business case for deploying G.Fast makes sense as a stand alone project, the roll out of fibre to the home is inevitable. So as well as having the costs of rolling out FTTH, service providers will also have the cost of G.Fast, a service which at best delivers a worse service than fibre to the home.

G.Fast is being positioned as an “upgrade path”, because it extends the fibre nearly to the home (FTTnH?) but avoids the manual labour of actually reaching the customer home and the tedious task of obtaining residents’ permission - the wayleaves (although this is achieved in plenty of other countries).

But because G.Fast is as fast as it gets on copper, and the telegraph pole is as close as it gets, it’s the last possible hoorah for copper, the end of the road. MIllions of pounds will be spent deploying it, but it has no future. It’s a pointless deviation, a technological cul-de-sac.

Fibre to the home

So what’s so special about fibre to the home? First of all, different laws of physics apply which means that the end user can be assured that the bandwidth delivered will be as advertised - and the service provider benefits from this also as they no longer have tortuous complications when advertising their services.

Speed are dramatically higher. In the core of networks, 100Gbps is quite common these days, with Terabit per second speeds on the horizon. Once an optical fibre is laid, the only limitation to bandwidth is the equipment at each end. So 1000Mbps, the maximum possible for copper wires (over a very short distance), is a trivial speed to deliver on fibre optics over any distance and in BOTH directions. 1000Mbps to the home is only the start. What possible future could there be if we had 10,000Mbps to the home?

So what? The so what is 21st century life, as demonstrated by the download of the OSX update I mentioned earlier. It “only” took 12 minutes, but why 12 minutes? Why not 1.2 seconds? Or 0.12 seconds?

Unlimited computing power has transformed what we are able to do, even on a laptop. Now the internet is a fundamental component of our computers, we are hamstringing our use of it by having such feeble connection speeds.

In the same way that the pioneers of the electricity industry couldn’t have foreseen the plethora of devices connected to their new energy distribution networks, we can’t envisage what applications and uses a true optical fibre network will enable, what changes to our lives such a thing would make. This is often cited as a reason not to do it - in my view it’s the reason to do it.

The following graph (source of original data James Encke of Eurotelco Blog  http://eurotelcoblog.blogspot.co.uk) shows the growth of these devices that we now take for granted, from 1950 to 1970. Just for fun, why not add ones you now have to this graph? 

Electricity in the home graphic.png

So if true fibre optic broadband is so good, is anyone actually delivering it? I mean replacing copper wires with optical fibres that terminate on customers' premises? And if so, what speeds are being delivered and what prices are being charged? It must cost a fortune to get a symmetrical 1000Mbps service, right?

FTTH Analysis.png

The above analysis of true fibre optic broadband services from around the world reveals that there are five basic groupings:

  • A - Asymmetric
  • B - Low speed symmetric (25Mbps - 35 Mbps)
  • C- Medium speed symmetric (40Mbps-60Mbps)
  • D - High speed symmetric (90Mbps - 200Mbps)
  • E - Ultra high speed symmetric (300Mbps to 2000Mbps)

As can be seen the UK is woefully behind the curve, although there are one or two bright spots. BARN did it for themselves and Gigler are running on top of Cityfibre’s infrastructure in Bournemouth. Gigler is interesting because of there innovative business model - charging by data volume consumed, just like any other utility, rather than the rate at which it’s consumed. If you live in London, in an apartment block, have a chat with Hyperoptic, as that’s their market. Notice that So-Net in japan is already delivering more than 1000Mbps.

The prices quoted were correct at the time of this analysis (Jan to March 2012) and many included bundles of TV and voice calls. One of the recent announcements from BT is the launch of BT Sport - TV content delivered over their copper network. I suspect the viewing experience will be less than satisfactory.

So, to conclude the discussion, G.Fast is an amazing technical development and full credit must go to those that laboured away on it. The trouble is that it is a technological cul-de-sac that will simply delay the inevitable rollout of true fibre optic broadband to the home. FTTH is real, it’s now, delivers consistent effectively unlimited high speeds irrespective of distance and this country is lagging behind others, much to its detriment.

Finally, I’ve tried to come up with something that shows the difference between bandwidth as delivered over copper wires such as from BT, bandwidth as delivered over coax cable such as from Virgin Media, and true fibre optic broadband as delivered over fibre optic cable such as from Hyperoptic and Gigler. This is my attempt - drawn to the scale of one vertical pixel equating to 1Mbps.

There’s quite a difference, isn’t there?


Broadband compared.png