Saturday, 7 November 2015

10Mb/s USO?

The Guardian reports that the government want to impose a universal service obligation to ensure people can get super fast broadband and a minimum of 10Mb/s Internet access even in remote/rural areas.

The comment “Just as our forebears effectively brought gas, electricity and water to all, we’re going to bring fast broadband to every home and business that wants it." in interesting, as many places do not have gas, and remote locations do not necessarily have electricity without paying a fortune.

Thankfully for Internet access, unlike gas, electricity, and water, it can be carrier over radio waves, so any of the various satellite services that can deliver 10Mb/s would meet this objective. They are not even stupidly expensive. They work anywhere you can see the sky. So, well done David Cameron, "achievement unlocked", 10Mb/s Internet access to even the most remote areas of the UK is now available.

The problem is that satellite links are "a bit crap", but to be fair, they are probably fine for the objective here - allowing access to crucial services. The two issues with satellite are total bandwidth which can create levels of congestion, and latency due to laws of physics.

So, if satellite is "not good enough" for this objective, we have to consider why, exactly. Why not clearly state the objective, and perhaps an objective test for what type of service would meet it? I suspect any reasonable test of such services would say satellite is fine for access to services, even some basic streaming.

But if we are to make a set of measures for whether a service is good enough, and somehow we want to deliberately exclude satellite links, because somehow we hate that industry (?), rather than simply saying "not satellite", why not state a latency and packet loss requirement along side that 10Mb/s.

That would allow innovation in solutions, whether copper, fibre, wifi, or even balloons!

But there is more... We can install 1Gb/s anywhere you like if you have deep enough pockets. Running a fibre to the most rural areas is "possible", just damn expensive. So again "achievement unlocked", 10Mb/s non satellite really is available everywhere now!

That means we need to actually pin down a price point for services (install and ongoing).

So what you need please, Mr Cameron, from your consultation and meetings, is:

  • Minimum Internet downlink access speed (e.g. 10Mb/s)
  • Minimum Internet uplink access speed (e.g. 2Mb/s)
  • Maximum packet latency (e.g. 100ms)
  • Maximum packet loss for a link that is not full (e.g. 0.5% over a day)
  • Maximum level of congestion permitted (how much of the day can it fail to meet above specs)
  • Maximum install price
  • Maximum monthly rental
What will be interesting is testing if any of the existing Internet services actually fail to meet the same criteria :-)


  1. Curious to see the full proposal and Openreach reaction.
    1) It seems the cost burden would be on the ISP. Who would this USO applies to? USO were easy to implement when there was a national (or area) monopoly (BT, Post Office...). Should this only applies to Openreach?
    2) Massive amount of rural location are still on ADSL <8Mbits and too far from exchange/cabinet to reach anything close to 10Mb/s.
    Once again really short term views of the government, change/conflict with the BDUK last 5% target and would be, once again below realistic requirements by the time it's implemented.

    1. At the very least of course they'd need to eliminate the remaining 20CN services, which top out at 8M, unless they resort to bonding!

      The "ADSL from the cabinet" might get us closer to this 10+M future one day (or with some hybrid for rural areas, where there isn't anything as grand as a cabinet serving anywhere - not enough lines to need one).

      Remember we already have a system where some lines cost more than others: 20CN costing more than 21CN, popular unbundled exchanges costing less than BT-only ones; I expect this USO would just extend that if introduced, so we might see something like "Long Line ADSL" tariffs with a higher monthly fee to recover the cost of ADSL repeaters or whatever Openreach might use to reach the required bandwidth.

      The real question, surely, is how much will it cost for a given speed? As Adrian points out, we could all have 1 Gbps fibre services if we wanted/needed them, within months - we just aren't prepared to pay what that costs, with current technology and the current BT network. Maybe 4 Mbps would be a more realistic target, which even 20CN can deliver on short enough or good enough lines, and Openreach could add repeaters or whatever as needed to achieve that?

  2. Again, you expect people writing legislation to actually know about what their trying to legislate!

  3. If it's anything like the USA's regulatory body, the FCC (Federal Communications Commission), they want fiber everywhere but want to only pay for 10/1 (ADSL2+).

  4. A fun metric to play with - how about throughput divided by latency?

    You could then say 1Mbit/s/1ms is your USO latency, and someone who wants to do 300ms satellite just has to provide 300M instead of 10M.