The overseas broadband bottleneck

It was a relatively small part of my piece in the Age yesterday but this has garnered lots of chatter:

I’m sitting here in the US at the moment on a 100Mbps maximum speed . But if I look at a website in Melbourne, my speed drops to 2Mbps. That is pretty much the maximum you will get from Australia to much of the internet, regardless of the theoretical maximum of your provider. This is because a key bottleneck is our submarine fibre link rather than our backbone network, or even the last mile.

It is worthwhile unpicking this. Broadband speeds and especially the user experience of those speeds is a function of a ton of stuff but is driven by an O-Ring quality. You will recall that the space shuttle Challenger blew up because a relatively minor component, an O-Ring, froze and broke. For broadband, your speed is limited by the weakest link.

At the moment, the weakest link for most people is their distance from the exchange unless you don’t even have ADSL in which case it is something else. Fibre to the Node and to the Home would alleviate that weakest link.

However, when I was in Australia, I was blessed with Australia’s fastest residential internet connection provided by Big Pond Cable Extreme. I received measured speeds of over 40Mbps but sometimes more. However, when I accessed overseas sites, it dropped to 2Mbps. By the way, for YouTube or a video download, that really wasn’t a constraint at all. I had to wait a little longer. Now there are issues of shared connections and network configuration and so I suspect you might be able to squeeze some more out of this with an NBN in Australia but the point is that the weakest link will soon become our overseas connections.

Now one thing to note is that this can be solved by moving the data. Data centres often play the role of hosting content locally. So that may come with the NBN because there would actually be a point to that. But if you are looking to have high definition video communications or things that might rely on it (like some fanciful remote surgery!) the interconnectivity issue will be there.

The question is: why? I have been contacted by some industry experts who claim to me that the capacity is, in fact, there and it is a puzzle as to why our overseas connectivity speeds are low. As I noted yesterday, it works both ways. Australia is isolated from the rest of the world. Thankfully, the only thing for which that would really matter for me is ABC News video and they block me from looking at it for ‘copyright reasons.’ (Hey, Mark Scott, I am still paying taxes and voting Mark Scott!)

I’ve asked these questions for years and have never received a satisfactory answer. Someone out there knows the truth but not even the Government has been willing to investigate even though at present it looks like crippling the NBN. I have long suspected that international carrier arrangements were to blame. They are still possibly high cost and local carriers possibly cripple overseas speeds to save on those costs.

One reason why that might be the case is that this phenomenon is not confined to Australia. Over the fold, you can see my results for a variety of countries including between the gold class network I am sitting on here in Cambridge (MA) and Japan and South Korea. You judge for yourself whether this is an issue worthy of official clarification.

Finally, I note, as always, that this is an issue because the Government refuses to consider the NBN as anything but broadband and is not structuring competition policy nor its own public policy to ensure the NBN becomes more.

Local transfer

To London

To Silicon Valley

To Japan

To South Korea

To Melbourne

10 thoughts on “The overseas broadband bottleneck”

  1. “Now one thing to note is that this can be solved by moving the data.”

    Currently there is no incentive for a consumer to lessen the load on the international links if their international speeds are the same as local speeds.

    (If I have a choice to download data from either locally in Aus, or from the US, both cost me exactly the same. if my local ADSL link is the bottleneck speed plays no factor)

    One can only assume that this means the ISP has enough international capacity, because otherwise they would make local traffic cheaper than international traffic to the consumer.


  2. How much would it cost for NBN Co to lay its own cable from Australia to the US?  Would this solve the problem?


  3. Joshua, I’m not sure what you’re implying when you say that “this is an issue because the Government refuses to consider the NBN as anything but broadband and is not structuring competition policy nor its own public policy to ensure the NBN becomes more”.  If broadband provides the best speed what else should it be?  Who owns the overseas cables(?) these days.  I thought it was still the OTC, but I’m obviously a bit out of date.


  4. Andrew: You are obviously unfamiliar with the practice several Australian ISPs have of “unmetered” data — traffic that does not count towards your data cap because (AFAIK) it is hosted directly on that ISP’s servers and is, for that or for other reasons, low-cost.
    From other ISPs’ advertisemens I also gather they host copies of popular web content on local servers to deliver improved speeds (effectively a price decrease).
    While I don’t know anything about the current restrictions on our international connection, I do know that Pacnet and Pacific Fibre are going to lay a new cable from Sydney to the US, with a planned capacity of 5.13 Tbps by 2013. (,pacnet-joins-447m-sydney-us-undersea-cable.aspx) That should improve conditions somewhat.
    As an aside, the benefits of the NBN are really only limited by this to the extent that the services are international; remote surgery where both parties are in Australia won’t care a whit about our undersea cables. Neither will e-learning, e-health, smart grids, or the rest of the domestic services.
    The NBN really isn’t so we can get faster access to foreign internet sites.


  5. The PIPE Networks cable from Sydney to Guam, PPC-1 went live last October. It interconnects with other trans-Pacific cables at Guam and has 2.56 Tb/sec capacity. It can offer channels of 40 Gb/sec. This is a major addition to the Southern Cross cable (Sydney-California, 2×480 Gb/sec with planned upgrade to 2×620 Gb/sec) and the Australia-Japan cable (currently 240 Gb/sec with potential for upgrade to 1 Tb/sec).
    PPC-1 is significant in that no established telco has significant stakes in it. (PIPE was recently taken over by local ISP TPG.) Southern Cross is owned by Telecom NZ, Singtel Optus and Verizon. The Australia-Japan cable is owned by Telstra subsidiary Reach.
    New Zealand firm Pacific Fibre is currently raising funding for a 5.12 Tb/sec cable via NZ to the US (upgradeable to 12 Tb/sec with 100 Gb/sec channels).
    On local mirroring of data, Google already has a mirror site in Sydney so Google queries stay onshore (although not nor YouTube). In due course others may follow.


  6. Its a bit more complicated – youtube is an overseas site but its content delivery network may well have the data you want in Australia so it won’t make a trans-pacific journey.  Akamai is the best known content delivery network if you want to know more.   I know little about networking but your tests results might suggest latency limiting bandwidth. If this is the case in theory you can modify networking parameters to get larger bandwidth despite high latency.


  7. Martin: All true, but that only works for content that the ISP choose to make unmetered. The alternative is a neutral one where they charge less for local content (which could be layered into ISP cache, same ISP, same country…) rather than on some arbitrary defined unmetered data.


    I think the idea of the NBN is to have super-fast speed within Australia to make all sorts of content and business’ possible. e.g. Medical and educational.
    Transferring lots of data from overseas live is likely to turn us simply into content consumers rather than content creators.


Comments are closed.

%d bloggers like this: