Up to speed: Telstra still hard to beat

Up to speed: Telstra still hard to beat

Summary: The latest analysis of ZDNet’s Australian Broadband Speed Tests shows that Telstra is still faster, whatever the connection type.

TOPICS: Networking, Telcos

Telstra has a reputation for delivering fast wireless services through its Next G network, but they also seem to lead the way with fibre, DSL and cable.

Speed Test
(Credit: Phil Dobbie/ZDNet )

The cautionary note is that this is based on people taking the ZDNet Speed Test — so it's a cross section of the Australian population geeky enough to run the test, and it represents just one point in time. It doesn't have the mathematical rigour of constantly tracking speeds across a representative sample of users over an extended period of time. Still, with around 20,000 test results from July and August this year, it is based on a lot of data, and does provide a worthwhile snapshot. What's more, earlier Broadband Speed Test results have shown the same thing — Telstra does seem to be faster.

Back in March, Telstra's DSL users averaged a speed of a little under 6.5 megabits per second (Mbps). Now, the speed is up around 7.2 Mbps. That makes them 14 percent above the average speed for Optus, and 17 percent faster than iiNet and TPG (both around 6.1 Mbps). It's not a huge differential, but enough to ask the question of why? Why does Telstra consistently out-rate competitors who are using similar, and often the same, infrastructure.

The question is even more pertinent when it comes to fibre, where the NBN promises a level playing field. Yet, Telstra's average speed (24 Mbps) is 50 percent faster than iiNet's (16 Mbps). It can only be that Telstra is selling a higher ratio of higher plans, keeping lower spending users on their own network. iiNet is probably keener to move customers to fibre, particularly if they're currently an off-net user, reliant on Telstra's infrastructure.

Telstra also wins when it comes to cable speeds, although the differential is less pronounced; 18 Mbps for Telstra, 17.2 Mbps for Optus. Given the song and dance made about the national roll-out of DOCSIS3.0 upgrades to their hybrid-fibre coaxial (HFC) network, promising speeds of 100 Mbps, it's surprising the differential isn't greater, and the speeds faster.

In the wireless space, Telstra's 3G does seem to have a large lead on Optus — although, it should be noted that we measured only 36 speed tests from Optus 3G users. Even with Telstra, there were only 206 3G users, with an average speed of 4.3 Mbps. The fact that there were almost as many 4G users (154 tests) shows how quickly subscribers are moving to the new network, which averaged an impressive 18.6 Mbps. Of course, the impetus for a 4G test will be more "let's see what this network is capable of," rather than "why is this network so slow?", so results will be skewed favourably.

Overall, though, it's further evidence that Telstra's network investments are reaping rewards. Or the power of incumbency provides network advantages that others cannot match. It depends on your perspective. And average speeds continue to lurch forward, putting paid to the argument that we're happy with the speeds we've got. We want more, and we want it now.

Tests measured July to August 2012: Optus 3G 36; BigPond 3G 206; Primus DSL 393; Dodo DSL 450; TPG DSL 3043; iiNet DSL 2166; OptusNet DSL 988; Internode DSL 746; BigPond DSL 4889; Telstra 4G 154; Optus cable 1646; BigPond cable 2541; iiNet fibre 93; and BigPond fibre 164.

Topics: Networking, Telcos


Phil Dobbie has a wealth of radio and business experience. He started his career in commercial radio in the UK and, since coming to Australia in 1991, has held senior marketing and management roles with Telstra, OzEmail, the British Tourist Authority and other telecommunications, media, travel and advertising businesses.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • Stating the obvious

    We cannot go wrong, thanks to Sol Trujillo's ingrained trajectory.
    Vasso Massonic
    • Adios Kev, Hello Sol.

      Indeed Vasso, Sol really was a visionary. I also think he maintained his friendly relations with Rudd when he sent Kev an Adios message when Kev was knifed by Julia. Good to read that Sol is going great guns in the U.S. and is in negotiation to purchase one of Americas biggest com companies.
      • Adios Sol

        Yeah... Sol did such a terrific job with the Telstra monopoly that they had to embark on a 1 billion dollar image rebuild after he left. And yet the share price was always a dog....
        Ed Pryor
  • Syd, If only sanity prevailed then................

    We could have been enjoying what we continue writing about, ever since.


    Vasso Massonic
  • Measurement basis?


    All the major providers are capable of driving customer DSL tails flat out. So Occams Razor suggests there is an issue with testing methodology here.

    Potential sources of that include:

    a) that there is a difference in test distribution in the case of Telstra (more tests geographically closer to the data source for the testing) and/or

    b) there are non-DSL tail circuits included in the Telstra results (misclassification or not asking the question in the first place) so that the Telstra numbers are actually the average of DSL and some faster links (e.g. Telstra Velocity estates) and/or

    c) that the data sources concerned are (in network terms) closer to Telstra users than others (which may sheet home to an issue of peering constraint between the Telstra and non-Telstra backbone networks).

    d) The testing methodology may not be delivering enough data to end users to properly measure peak circuit performance. If the file being downloaded is too small, then a download from a closer (hence lower latency) source will 'win' despite there being no actual performance difference in the real world; Enough data has to be delivered to get past the initial TCP 'slow start' regime and into full speed data delivery over the path concerned. This gets especially important if the data source and the data sink happen to be in different states.

    Meantime I can routinely run any Internode tail circuit flat-out from any well connected file server in Australia, on any line I test it on - which suggests some or all of the above may be factors in the testing results reported on here.

    Can you point us to a link explaining the testing methodology used to get a speed figure, in the testing you are reporting about here? Especially including the geographic location of the data sources, the size of data file being transferred, and how the resulting speed is calculated?

    Simon Hackett
    • Good point Simon.

      At best this shows our average speeds are increasing due to upgraded national backbones, increased 3G coverage or a migration to more top hats on adsl rims.

      At worst this could show that telstra are shafting their competition in data transmission speed over wholesale dsl.
      • Speed is dependant on end to end infrastructure

        The access technology at the customer end is only part of the story. Data speed is dependant on the performance of the infrastructure end to end. The consistency of the results between carriers suggests that the difference is due to what is going on in each ISP's servers and backhaul capacity. The 'Telstra wholesale ADSL' excuse does not account for those using ULL, their own HFC cable as in Optus' case, or fibre be it NBN or their own.
        • Complete Bollocks!

          That's complete rubbish, my sync speed jumped up 2.5M when I switched from Telstra wholesale (iinet) to Telstra retail... same Rim, different speed... even the Telstra sales rep guaranteed me off the record it would be faster.

          Telstra wholesale 'nominally' limits sync of ADSL to 4.5M, so you can NEVER get full speed on a RIM, but telstra customers CAN off the same rim.

          I also notice Optus 4G is missing from the graph even though it was operating in this period.... nothing like another telstra advert. I'm starting to wonder if Telstra is paying zdnet as much as apple ATM - cash for comments.

          The whole thing is corrupt.