Mobile broadband: not up to the job?
What we learned
Workshop Mobile broadband has enjoyed a meteoric rise to fame since it became widely available, with many mobile operators reporting triple digit growth year-on-year, and data usage surpassing mobile voice. Access to multimedia content and social networking sites from smartphones, and traffic to and from mobile broadband-based PCs, has contributed to this growth.
This rapid escalation of demand has apparently caught some operators by surprise, and the underlying network is increasingly struggling to cope under the extra demand.
Consumer and business users alike have not changed their expectations of consistent performance and anytime, anywhere access however; and so the decrease in quality they have experienced is now starting to filter through as widescale user dissatisfaction.
To be fair, there is always the possibility, in any comms service, of disruption. But the impact of disruption on data connectivity is arguably much greater than for voice. When a voice call drops, it is inconvenient and annoying, but you can always call back. Data connections, on the other hand, are a different kettle of fish. For example, a dropped connection can result in a download needing to be restarted, or even, in extreme cases, lost data.
When we wrote recently about mobile data quality, we suspected that things still weren’t all that great. The feedback from Reg readers, however, suggested that the problems were possibly worse than we had originally thought. A comment from one Reg reader based in the US, where, ironically, data service quality is often perceived to be much better than Europe, was in line with many of the responses:
You've only got to look at a map of coverage in the US to see the root of the problem - mobile networks want customers so they cover large population centres where there are lots of people, and the roads connecting large population centres.
Since they are looking to sign up customers, all they have to do is provide a mediocre service that works most of the time ... some signal strength is better than none but there's no real pressure to do more than provide some signal. This strategy works fine most of the time, but you can see it fall apart when large groups of people come together suddenly - sports events or disasters like Katrina where the mobile phone service in Baton Rouge collapsed overnight (and remained down for weeks) when 60,000 people with mobile phones moved into Baton Rouge from New Orleans overnight.
Mobile operators have no incentive to improve a service that generates a large cash flow for a mediocre service. Since mobile phones are rapidly replacing land lines we are - in effect - moving society’s ability to communicate in the event of a disaster back to that of the 50's.
There was also a sense that all operators were in this together, and there was no single culprit. As one reader wrote:
Mobile data is a race to the bottom. Cheapest provider wins. It's much like DSL but at least with that there are business quality providers if you're prepared to pay a bit more. With mobile there's no choice.
I like to do the 'spotify test'. If an (allegedly) 7.2Mbps connection can't reliably stream a 96kbps low quality spotify stream it fails. This isn't a hard test - I'm expecting them to be able to provide a mere 1/75th of their advertised bandwidth.
Where I am, there is only one provider that hasn’t consistently failed this test. Coincidentally, it is the most expensive. And it’s not exactly stunning either.
Maybe by the time LTE is rolled out 96kbps will be achievable
Comments like this are not so surprising, when we consider that the mobile data speeds advertised by mobile operators are rarely achieved in practice. For example, some recent, albeit small-scale, tests across mobile operators in the UK showed achieved data transfer speeds to be, on average, about a quarter of the rates touted as available. One director of a small UK-based business shared his thoughts about mobile broadband:
It's highly variable in my experience. Mobile broadband outperformed our home ADSL until we recently upgraded the ADSL. Even now it has 4 times the uplink speed. But then you can be sitting there with 6 bars of signal in Waterloo station and getting really dire speeds.
The downturn in quality is clearly taking its toll on the patience of users, and there is increasing pressure for mobile operators to address the problem. This is no quick-and-easy fix, however, as it will require significant investment. Somewhere along the line, there will be a need to recoup this, which translates to increased tariffs - always contentious for users. At this point, a premium tariff for businesses which are willing to pay more, to guarantee a given level of service will, maybe, become viable. This quote from one reader suggests that this whole area should be opened up for debate:
The real problem for mobile data quality is for the types of business roles that mean you could be at any location in the UK at any time. You simply cannot rely on the connection to be available (even for a telephone call never mind data). Given that, we have had to work around connectivity problems by transferring data whilst travelling or at a stronger connection location, but trying to hunt for those can be an issue on its own. Is 99% reasonable coverage possible at an acceptable cost to the providers? I don't have that answer, but if there could be a promissory that big clients will pay for the service they want to rely on, then it could be a good thing.
Our dependence on - and appetite for - mobile broadband is unlikely to dissipate over time. And unless there are viable alternatives, then the problem of mobile data quality will continue, and no doubt crop up time and again. Have you given up mobile broadband as a lost cause, or does your service deliver? Would you be prepared to pay more for better service, or do you think the mobile operators need to sort out their problems themselves, and not pass the bill onto users? ®
This is only a single anecdote but:
I run the IT for a small school in London. I have two very IT-knowledgeable people directly above me - the head and the bursar. Our ADSL provider recently took it upon themselves to chop our business connections, without warning, because "We were using more than an average ADSL home connection". Negotiations were completely fruitless, with the bursar starting out by literally saying the words "How much do we have to pay you to put it back on?" and ending with threatening to take the children out to the playground, and photograph them pointing at a load of laptops, netbooks, etc. showing a 404 page and sending it to the local press.
The ISP were complete idiots, no doubt, and lost a very reliable and well-paying customer for the sake of a few Gb (they also lost all our ordinary phone business because of their stupidity - that's only a dozen lines but it was one hell of a bill each month for the lot). They didn't understand the concept that 450 kids+staff use more bandwidth than a little grandma, and they knew we were a school from day 1 - they did the installation!
We asked for our options and were told that the *only* thing we could do was get another line installed to share the bandwidth across (which we would have to setup!) and that would take two-three weeks to install - in the middle of the exam period for students. In the meantime we would be completely without Internet access because we were limited to only 128Kbps which barely lets us collect email school-wide, let alone do anything practical. They refused to "re-activate" it, or increase our bandwidth allowance no matter what package we signed up to.
If we had to install another line, it might as well be with a more tolerant ISP who understands what "business line" means, so we severed the connection and instructed another ISP to initiate a dual-ADSL installation immediately. That left us with 2-3 weeks in which to cover ourselves for Internet access (when England were still in the World Cup, too, I might add!).
I posited an emergency measure - 3G dongles. I had one of my own that I always carried on me and I had in the past plugged it into our systems as a test device. I knew they worked well enough if we were careful, and I demonstrated it to the head.
Immediately, the bursar whipped down to Argos and bought a shed-load of T-Mobile 3G dongles with 2Gb allowance / month on each (and then they limit you, not charge you). We topped them all up with some cash, plugged them all into our Linux firewall/router, I wrote some scripts to automate things a little (e.g. switch between dongles that had already used up their monthly allowance etc.) and hey-presto - several 2-3Mb connections school-wide that I could load-balance and switch between in order to balance demand, speed, bandwidth usage, top-up-credit, etc. As far as anyone else on the network was concerned things just started working again, Internet-wise.
It cost us about one-tenth of what our up-to-8Mb ADSL package was costing us, it gave us similar speeds, much better upload, greater reliability, and seamless Internet access across the school without having to buy any extra hardware except the dongles (which we've since redeployed to staff for school trips, home VPN access, etc.). It worked perfectly for 2 weeks, then our up-to-24Mbps ADSL came online and we only kept a couple of them plugged into the router for emergencies.
Now consider that this was done by plugging in a handful of cheap 3G consumer dongles into the various USB ports available on a Linux PC that acts as a router/firewall, with some of them basically sandwiched in between two others because of the proximity of the ports. Consider that the room this was in is some way inside the school and had tons of copper running through it (including network cabinets, the phone system, servers, etc). Consider that it was in the middle of a school, filled with staff using mobile phones, in the middle of a London town, surrounded by houses and main roads and just a few hundred yards from the main shopping road.
The dongles didn't even sniff at it - they just got 3G connections each (not maximum theoretical, obviously, but each good enough to use it happily without noticing slow-downs), and worked flawlessly all day and night. We scripted overnight shutdowns for them in order to reduce bandwidth but each morning they came back online like a champ, and ran an entire school sometimes on just one or two when we were initially testing. It felt slow with only one, but nobody complained - they had "usable" Internet that was a damn-sight faster than our limited ADSL. Windows updates, anti-virus updates, web browsing, Java games, Facebook uploads, clipart browsing in Office as an entire class - you name it, it occurred during that time and worked perfectly. I was downloading a new copy of OpenOffice at the time, too.
We *did* ban any streaming of the world cup - we felt that was taking the piss slightly, so instead we streamed matches over the network from a DVB-T stick using VLC. Otherwise, 3G saved our arse. Our other alternatives were literally things like asking the neighbouring houses if we could tap into their wireless or run a cable into their house, or pointing a wifi antenna at a local OpenZone hotspot.
This was obviously only a static setup, but I think that's pretty much a worst-case scenario - a pack of 3G dongles all touching each other, all connecting at the same time, all from "unique" customers, all on the same base-station, all trying to get the maximum out of their connections simultaneously with real, varied traffic, within the borders of Greater London, near main roads, shops, houses, etc. It worked damn well, better than we ever expected, and I'd do it again in a trice if it was necessary.
Oh, the bursar complained because his office is in the next room and *his* own personal 3G dongle (which he used when uploading anything critical) occasionally went from 3G to HSDPA when we did this.
3G works. T-Mobile did a fantastic job, even if we were abusing their services a little. The mobile network worked better than I ever would have imagined. If anyone ever says that 3G is shit, or isn't capable of things like that, it can *only* be under-investment at the mobile carriers end. In our case, everything just worked. Best £20 I ever spent, buying my own 3G dongle.
Re : Well
Surely you must be mistaken. Linux can't cope with most hardware - everyone KNOWS that.
Just kidding of course. Nice example
The missing link
Whether mobile or fixed, the one paramter that's missing from all the ISP data is lowest speed.
Any crook can promise that I'll get speeds "up to" so many GBps. I'll continue looking for the one that tells me I'll enjoy speeds of "at least". Until then, ISP promises are just that - and woth about as much as politicians' promises.