Testing Telstra Ultimate Wireless Broadband Over Four States

For the past week, I've been testing Telstra's Ultimate Broadband across a large chunk of the Eastern states. The conclusion? It's delivered some impressive results, but it hasn't yet threatened to get close to its claimed typical download speed of 20Mbps.

During that time, I've run speed tests to see how the newly-released device performs in Sydney, Melbourne, Canberra, Brisbane, Townsville and Dubbo. (Yes, that's technically three states and a territory.) All of those locations fall within the 50% of the country which Telstra says can access the high-speed service, so I shouldn't constantly find myself falling back on the basic Next G network, which offers impressive speeds but doesn't claim the same potential performance.

Telstra claims that the typical download speed on the Ultimate Network is between 1.1Mbps and 20Mbps (as distinct from 8Mbps on enhanced Next G services, or 3Mbps). Across my tests, the performance never got better than 15.54Mbps, and it more often found itself below the 8Mbps level. The average result across all tests was a download speed of 7Mbps, an upload speed of 1.4Mbps and a ping time of 109 milliseconds.

That wasn't necessarily to do with location either: while I got my best figure at Sydney Airport, the Sydney CBD didn't offer a better result than Dubbo, as I noted in my write-up of my rural visit last week. Indeed, my worst-case score was on George St in Sydney, the very hub of a CBD area as Telstra describes it, delivering 2.71Mbps download speeds. That's not unworkable by any means, but it's nowhere near the typical speed promised.

This isn't unusual for wireless broadband, of course. Performance and speeds vary depending on your location, how many other people are also using the service (often a factor of the time of day), and the actual online activities you're trying to conduct. Nonetheless, the gap between what's being claimed and the actual performance is a little higher than I was expecting, especially in business areas. Here's a table of some of the more notable results (click for a larger version):

Note that being in a moving vehicle didn't necessarily mess up the results, but that performance could vary widely even in fairly close locations (such as Sydney's two airport terminals).

I'll happily admit that it's hard to do a perfect test for wireless broadband. The enhanced DC-HSPA+ service isn't being used by anyone much at the moment, so these numbers undoubtedly represent best-case results. That said, if I can't achieved 20Mbps for downloads in this scenario, I can't imagine anyone will in the future, when it's a widely available consumer option. (Business users will be able to buy the device from Telstra shops from October, and a consumer release is expected later this year.)

It's also easy to become obsessed with speed without considering other factors. The most important element in a wireless broadband service is consistency — not suffering constant dropouts if you're working in a single location — and Telstra has a little bit of work to do on this front. On the prepaid Telstra service (which is what I regularly use), dropouts are marked by a 'No signal' message, which at least tells you there may be a problem.

On several occasions with the Ultimate client, the connection stops working but the software still claims it is in place. That's not an earth-shattering problem — disconnecting and reconnecting usually fixes it — but it shouldn't be down to the user to have to work that out.

My testing isn't quite done yet. Later this week, I'm going to take a train up the east coast from Sydney to Brisbane, stopping overnight several times along the way and testing how the network performs (a project I've informally dubbed "Off The Rails"). While some of my previous train tests have suggested Next G and trains don't get on, some of the best results in the current roundup came in moving vehicles, so anything's possible.

Lifehacker Australia editor Angus Kidman would become a completely dysfunctional individual without wireless broadband of some sort. His Road Worrier column, looking at technology and organising tips for travellers, appears each week on Lifehacker.


    Seriously you need to use an FTP server with Telstra

    Like a test file from

    Not speedtest.net (slow and invalided data)

      I'm not a fan of FTP-based testing -- too removed from what the average user will do most of the time. Would also need separate test suites for uploads and ping, which complicates the process. No testing suite is perfect, but using the same one all the time should mean the same advantages/disadvantages for all the locations.


        I agree with you but 'mirror aarnet' also has http access. For instance I just changed ISP's so the file I use is;

        its 100megs but of course I don't grab the whole thing. I don't get ping but it gives me a good ballpark figure......


        The problem with www.speedtest.net is that it displays a peak transfer rate which is neither a realistic user case scenario nor a scenario that happens IRL.

        IMO Adam Nelson's suggestion isn't as bad as you make out. Its a test that's more realistic scenario than www.speedtest.net since it will still test TCP data transfer performance over a period of time in a manner similar to HTTP.

        Angus. But not acceptable method of testing your not getting true speed

        Reason for saying this if you came back to adelaide. I reckon I could max the connection out for 3 mins tests at 3 different cell sites

          "True speed" is a tricky concept. While I'd agree that the methodology used by Speedtest.net is liable to perhaps overstate the figures, in a context where I'm still not getting near the claimed speeds that's arguably less of an issue.

    I'm very interested in the ultimate speeds. I'm currently using the elite modem and my best speedtest.net result was inside the new Melbourne Convention centre. 13 Mb/s down and 3 Mb/s. Theoretically the ultimate should pull 26 Mb/s down right? I would be interested to see you do single channel HSPA+ tests at the same time.

    Appreciate your detailed work.


    I'm just wondering if you know where I can find a list of the towns/locations where this DC-HSPA+ has been enabled?

    Thanks in advance,

    Angus, can you try to make a Skype or other VoIP call in your travels and post your findings?

    If wireless is to be the primary (or only) broadband service available to someone, they need to make it pay its way by eliminating STD, IDD and mobile phone call charges. If VoIP calls cannot be sustained due to frequent signal dropouts, it gives the lie to anyone suggesting it is adequate for regional Australia.

    For those wondering, I was able to dig up this link after many hours of searching on the Telstra site:

Join the discussion!

Trending Stories Right Now