End all the 'up to' broadband speed bull. Release proper data – LGA ISPs should release data on broadband speeds at a household level so residents can easily compare speeds and switch providers, the representative body for 370 local councils has said. The Local Government Association (LGA) is calling for greater transparency of speeds direct to consumers' homes rather than just their postcodes …
> "Up to speeds" should be the expected peak time speeds for 67% of users
*ahem*bullshit*ahem*
"Up to" speeds should _soley_ refer to line sync speeds and should show typical speeds, not the maximum available speed (My car can go up to 140mph. I've never been that fast in it and in central London I'd be hard pressed to go 30mph)
ISPs should be forced to disclose multiplexing ratios in each area and overall. They can control what speeds are attainable up to the border of their network but everything beyond that point is not under their control.
What is a multiplexing ratio? How would consumers be able to interpret that? How would you even define it? It's fiendishly difficult to do this meaningfully.
Let's take it to mean sum on sync rates divided by sum of external network connectivity, counting things such as CDN as external connectivity.
Assume an operator has 10:1 "multiplexing" ratio in this meaning and they offer a 100Mbps service today. Let's say they decide to upgrade that to 200Mbps. Then that suddenly becomes 20:1. Does that mean the network has become worse for the customers?
On the other hand you can have 10:1 which a couple of years ago may have provided a completely uncongested service and today it's slow due as average consumption went up.
There are much more meaningful metrics such as off-peak vs. peak transfer speed. That is actually measured by Ofcom today.
Actually most ISPs today score very well on that.
More than that, it's really the best the providers can give because this is literally a case of "Your Mileage May Vary". Internet connections are like a chain: they're only as good as the weakest link, and many times the weak link isn't the ISP but somewhere else along the way. How can ISPs properly account for this?
How can ISPs properly account for this?
That is not their job. If I use a bus to get to the train station to take the Eurostar, do I blame British Rail for the slow bus? NO.
However, if the Eurostar somehow only achieves 33MPH all the way to Paris, and they blame it on London Transport, I would expect "questions to be asked".
"It's just taking advantage of the general population's stupidity."
Downvotes ahoy, but I really feel like I ought to say something about brexit here...
...thing is, MOST advertising takes advantage of people's stupidity and/or lack of attention. This is absolutely the best shampoo on the market (sample size of eleven bored housewives). Nine out of ten cats prefer it (I bet the alternative was a Tesco stripey tin cat food - and we must surely feel for the cat that preferred it). FullHD, just don't ask what the actual screen resolution is. And those loan adverts that finish with the Ts&Cs in rapid blurry text and spoken so fast you'd have to concentrate to catch even half of it.
Advertising is the art of taking something mundane and sexing it up into something desirable. Oh, look, now I feel like I ought to say something about apples...eeee, look at my big bramley.
"Up to" is okay, just not very useful. It tells you what you might get, possibly could get, but no indication of how likely you are to get it.
"From" isn't very useful either, because that's going to have to come with a cart load of caveats because sometimes things do just go slower and it's not always the fault of the ISP.
Some sort of average seems better but still isn't a guarantee of anything, nor an indication of what others may get.
The problem isn't so much what claims say but what they mean.
Maybe. But if you give "up to" for my postal code, which covers the edge of a large city and extends out into the countryside, only the urban part can get the "up to" speed even if they pay a premium for it. The rural folks don't have the infrastructure going to their homes to carry the "up to" amount no matter what they are willing or able to pay. Anything above street level is misleading, because the equipment can vary widely within an area the size of a postcode.
Interesting site - I'd been wondering when someone would come up with something like that. Interestingly their testing methodology is run with an HTTP Post file transfer, which would run over TCP, which has a warm up time, so may well under represent what you might possibly be able to get with a UDP connection (admittedly with the possibility of errors).
It would be nice if house move websites allowed for an actual test result to be included in the sales brochure, rather than linking to the BT checker which abuses the "up to" terminology
I don't trust this data at all.
The fundamental problem with the data is that it doesn't actually give you the service being used. So in my area for example, the speeds are very interesting, either people are getting sub 1Mbps or a variety of numbers which indicate people have subscribed to various FTTC packages. So is has neighbour with the 42.5Mbps service, actually paid for an up to 100Mbps service...
Obviously, I've caused problems because the dataset includes my mobile broadband connection which uses a roof aerial to improve performance - with the arrival of FTTC, the anomaly of this service isn't quite so striking...
"Internet connections are like a chain: they're only as good as the weakest link, and many times the weak link isn't the ISP"
In most cases, the weakest link is the connection to the property. In the UK the majority of broadband users only have the option of using the copper wire to their house, so the speed will be entirely dependant on whether FTTC has been rolled out to their area, otherwise they are on basic ADSL where distance to the exchange is the main factor.
The BT Wholesale speed checker can pretty much tell you exactly what's available if you put in the phone number corresponding to the broadband account in the "Further Diagnostics" page.
So maybe just getting BT Wholesale to supply a database of highest/average/lowest speeds by postcode will be good enough for most? Virgin and other cable suppliers could probably provide similar data for their connections.
"In most cases, the weakest link is the connection to the property."
Depends. If you're connecting around the world, there's plenty of potential for a weak link along the way. If your destination's pretty obscure, you raise the chance for a weak link. There's a lot of factors beyond your control, though I will admit if you're at the mercy of a DSL link in the middle of nowhere, you've got a pretty lousy hand to start with. I wouldn't know; I've had cable modem since about 1998 and wired my own house. With FTTH as the local alternative, the cableco's been steadily improving the service and are now starting to roll out Gigabit service (via DOCSIS 3.1) to counter. I don't trust the max speed, but the data cap that goes with it looks tempting.
What the LGA or the media don't really mention quite often is that a lot of the speed / performance issues are due to the poor state on phone wiring beyond the master socket in the home. The BT Wholesale speed checker site is generally a pretty accurate estimate as long as there aren't wiring issues in the home and anyone can use this site or use their ISP provided version of this info when signing up.
People will often complain about poor speeds without doing even the most basic of diagnostics on their own home setup and blame the ISP or Openreach when the responsibility for any wiring issues after the master socket is the householders problem.
As an example, I recently went to resolve an issue for someone where they had upgraded to FTTC (via BT as ISP) and the estimate for their line was 40mbps download but they were getting about 15mbps. I found this was entirely due to their extension wiring being spliced in before the master socket. As soon as the extension wiring was correctly wired into the removable panel on the front of the socket and a central VDSL filter fitted then the speed went to the full 40mbps.
I found this was entirely due to their extension wiring being _illegally_ spliced in before the master socket.
There, FTFY - and in such cases Openreach are fully justified in charging full whack in remediating the problem caused by interfering with the line on their side of the demarcation point (There are a still a couple of offences on the books which cover this kind of thing)
"Virgin and other cable suppliers could probably provide similar data for their connections."
They could but there isn't much point IMHO. From my experience and that of others I know who are Virgin customers, if you're in a Virgin area you pretty much get the advertised speed.
This comes down to the the type of cable used and the fact it is much newer and was always intended to provide more than just a basic audio telephony service unlike a lot of BT's copper, maybe not the 220Mbps it's now doing but still more fit for purpose than a single twisted pair phone line.
Occasionally with Virgin post install visits are required to fit attenuators or they have to tweak the power levels from their end but of the dozen people I know on Virgin in different areas, they all get the speed advertised.
Virgin's only real speed problem is contention in some heavily populated areas.
"From my experience and that of others I know who are Virgin customers, if you're in a Virgin area you pretty much get the advertised speed."
For a long time around here, people were getting the advertised sync rate but throughputs during peak evening period would have made a 33k6 modem look spritely.
Virgin have(had?) two speed problems:
Contention - this used to be awful, haven't been with them for a while now, but it was a serious problem. I wasn't in a large city either...
Upload speed - I used to saturate my upload trivially, IIRC it was <10% of my download rate. That's extraordinarily poor. And with the vast increase in 'sharing' and 'using someone else's computer' this is more and more important.
The download was great - their customer service when presented with logs indicating a failed coax in the street cabinet was appalling. The failure manifested between 9:30 and 10 at night until 6:30-7 in the morning (temperature related) and they consistently sent engineers out at midday - who said there wasn't a problem.
Eventually I got an actual network engineer and we looked at the houses with issues, and decided where the fault had to be. We were right - and it took 5 minutes to fix it.
local data for local people
Given that the LGA represent the bureaucrats that decided my festering rubbish only needed to be collected every two weeks, have shut down half the libraries, can't maintain roads, are slashing the social care and community health provision (to spend the money on other shit, like in my local council's case, a music festival) I'd suggest that the LGA stick to talking about things they know about.
It won't be a very long list.
I particularly like hair colouring products. "Covers up to 100% of your grey". Well.. obviously.. I mean even if you missed your head completely whilst applying it, the statement would still stand.
Most people who tried it, loved it and would buy it and recommend it to a friend and they'd recommend it to their friends it's that awesome!" And in the small print it says according to a survey of 11 people. There just isn't small print small enough to make that kind of claim and keep a straight face.
What I very cynically want to think happens there is they decide they want to say a particular percentage - so as soon as the numbers hit that percentage, they stop, lest it starts to go down if they carry on.
What probably really happens, though, is hardly anybody can be arsed to fill in the survey, so they're stuck with stupidly low numbers - and despite the fact that makes them meaningless, they use them anyway, hoping nobody will notice.
I wouldn't mind terribly if these claims can be controlled by the law in some way. Like testimonials can ONLY display typical (let's say modal to avoid wiggling) results. I will admit that "up to" claims will be hard to control since for some firms that's all they can promise (due to lack of control). But a law that demands as much truth in advertising and as little wiggle room as possible would be nice. But as they say, the devil is in the details.
"As has been mentioned previously, the BT Openreach estimator that's available to wholesale suppliers is pretty accurate."
The speed I'm able to get in the middle of the afternoon here might be a good deal more than what I'd get in the evening if a lot of people down the road start streaming stuff when they get home from work and my bits have to share the infrastructure with whole lot of others. It might also be better or worse than my neighbours; all our connections come from the same point on the buried cable. Mine comes underground, theirs are overhead from a cable running up a pole, some of them distributed direct from that pole and others from a second pole linked to the first. Clearly there are various options for water penetration, different wiring choices (Al vs Cu) etc.
The laws of physics are irrelevant.
Customers should only pay for the service they receive - not what the supplier might THINK they are receiving. And it's high time someone made that mandatory.
If only OFCOM actually worked for the consumer instead of big business.
Dave promised to scrap them....oh well, just another one of his broken promises.
Well, the "up to" is just garbage. I mean, I can sell an "up to 100Mb" to You, and deliver just 2Mb. It is "up to". Not good enough.
To me, the reasonable course is advertise a "no less than". Sure, go ahead an print the "up to", i don't care. How about "up to 100Mb, and no less than 10Mb"? It is fair, allows the ISP to sell what it wants to sell, and informs the consumer.
Here in Brazil, the law says the ISP have to deliver at least 80% of the sold speed, during no less than x% of the time (forgot how much, I think it's 60% of the time). It's another way to do this.
But I think the "no less than" is better. You can sell how little "no less than" You want, but cannot go bellow it.
If people want to complain, they should be required to submit a speed test at the time of the supposed problem as evidence that it was the ISP at fault.
Simples.
What you're recommending won't be very effective or useful:
1. ISPs do not offer real Quality of Service guarantees because punters don't want to pay for it. Your recommendation would dramatically raise rates.
2. Requiring network capacity to support the 60-80% usage mark for all users would require a huge amount of unused capacity sitting around, so there's a really idiotic amount of waste incurred that no ISP can afford to provide.
3. ISPs cannot and will not offer QoS guarantees for other ISPs, i.e. they can't ensure that some other ISP won't introduce packet loss that makes your connection suffer.
4. Service levels really don't just depend on local loop utilization. Given all the various L4 connections make between systems, it's certainly very difficult to ensure some sort of uniform performance guarantee with only one component of many in the end-to-end connectivity that exists with the Internet.
Network capacity planning is complex, but exponentially so when you have so many individual ISPs all interconnected to provide "Internet" services. Trying to regulate a minimum performance level with such random, aggregated connections is virtually impossible, since no single person understands all of the connectivity end-to-end.
What you're recommending won't be very effective or useful:
I am not recommending: this is how Internet works here in Brazil. I have 15Mb/s down and 2Mb/s up at home. It is extremely rare to see it goes down more than 30% of the sold speed.
1. ISPs do not offer real Quality of Service guarantees because punters don't want to pay for it. Your recommendation would dramatically raise rates.
Wrong. QoS is just a guarantee. I can sell QoS of 80% nominal speed. If it is on the contract is valid. What would be extremely costly would be to guarantee 100% of the speed, with a SLA of 99,99%. What we have if 80% of the nominal speed, during at least 60% of the time. A far cry from 100%% with 99,99%.
2. Requiring network capacity to support the 60-80% usage mark for all users would require a huge amount of unused capacity sitting around, so there's a really idiotic amount of waste incurred that no ISP can afford to provide.
But they can. They just don't want to. Time and again I see this in the real world. When the ISPs needed to provide only 20% of the nominal speed there was one that would do only this - and barely (its name was OI). There was another one that would give me ALWAYS 100% of the plan. Sure, the guarantee was only 20% - but I ALWAYS got 100%. Did a download of 160GiB (took me days), and the router's MRTG was a flat line - at 100%% speed.
3. ISPs cannot and will not offer QoS guarantees for other ISPs, i.e. they can't ensure that some other ISP won't introduce packet loss that makes your connection suffer.
They don't have to. They have to guarantee the band until its borders routers.
4. Service levels really don't just depend on local loop utilization. Given all the various L4 connections make between systems, it's certainly very difficult to ensure some sort of uniform performance guarantee with only one component of many in the end-to-end connectivity that exists with the Internet.
No one said that. The ISP only responds for its own network. Anything different would be insane.
Network capacity planning is complex, but exponentially so when you have so many individual ISPs all interconnected to provide "Internet" services. Trying to regulate a minimum performance level with such random, aggregated connections is virtually impossible, since no single person understands all of the connectivity end-to-end.
The regulation applies only inside the ISPs borders. No one is talking about anything beyond that.
https://www.dslchecker.bt.com/
shows exactly what you get at your home address.
me and my immediate neighbors are on a different cab to the rest of the street and benefit from slow vDSL from our fttc cab 1.2 km away vs slow adsl for the others on the road connected to the non fttc cab on our road ~200 meters from my door..
i understand that the product i subscribe to will provide a speed upto x mb/s which is Dependant on my line quality.
All the xDSL providers are in the same boat and will provide the same speeds as the limitation is the access line to the customers house from the local cab.
i can buy:
adsl up to ~ 18mbs
vDSL up to ~ 50mbs
vDSL up to ~ 70mbs
VM 50/100/300 mbs
whilst VM's coax technology ensures they can guarantee their access speeds to all, the others are different and variable.
the public just needs to know there are different tiers of access speed which is dependent on their individual line characteristics. nothing more nothing less.
Its like a choice of toll roads offering different max speeds with the car being the connection rate. A crappy car won't be able to reach the top speeds in any of the roads, whilst a decent car will on all of them.
the only fix is to buy a faster car. or in BB terms move to FTTP or move the cabs closer to peoples homes and continue to use copper as VM have done.
whilst VM's coax technology ensures they can guarantee their access speeds to all, the others are different and variable.
Actually, to a certain extent it's the other way round.
With xDSL connections, you have your own pipe back to the aggregation point - whatever speed that line supports is what you get, regardless of what the neighbours are doing. Yes that speed varies between lines, but it's more or less fixed for each line.
Cable internet (the coax sort) is a shared medium. A number of users will share one coax and it's possible for your speed to be affected by what others are doing. As an aside, that is one reason (of several) why they insist that you have to use their provided router or modem, becase unlike a faulty xDSL modem, a faulty cable modem can take out (or severely degrade) service for other users on the segment. But on the upside, as pointed out, the coax is designed to (and does) carry much higher frequency signals which allows for much higher speeds.
Switching to SSE fibre (they're a reseller of Daisy) has been a disaster for me.
Sure, I get the advertised ~38Mbps EARLY IN THE MORNING.
Evenings? Forget it, Daisy's backhaul is so hopelessly saturated I'm lucky if I get 2Mbps.
Complete waste of time - and it seems there's nothing SSE can do about it.
My own fault, I should have realised the deal was too cheap, but even if there had been postcode-level mapping available, I would still have been suckered. I'm currently arguing with them to escape from my 18-month contract, on the basis that what they're providing isn't worthy of the term "broadband" :-P
Now we're getting to the stuff that matters. Bandwidth to the ISP is one thing. What the ISP does with that is another. Traffic-shaping, for instance. When my old ISP fell into the clutches of TT they traffic-shaped Usenet out of existence for part of the day. And the first rate customer service had been wiped out by a previous owner - I still don't know whether it was run by a chatbot or humans that had failed the Turing test.
These are things which are under the control of ISPs and going to be experienced uniformly by all of the ISP's customers.
"At the moment, there is no one place that consumers can compare side by side estimates of the broadband speeds that providers could supply to their home"
Er, as effective speed is a result of the quality and length of the copper line to the premises, and nothing to do with the ISP(*), there's no point in 'consumers' being able to compare speeds from different providers - they will all be the same!
A shite line is a shite line... changing ISP won't help.
(* network-side traffic shaping aside)
Spot on - the same half-wits who do not understand what 'up to' means are the same people who think that switching between ISPs who use the exact same final connection will give them a better speed.
FWIW my parents two next door neighbours get 7Mb connections on ADSL, and my father gets 2.4Mbit despite bing on the same cable, and we have tweaked, adjusted, playe with the terminating equipment with the ISP.
Bad copper is just bad copper.
"Bad copper is just bad copper."
Yes, but if you have a decent ISP they'll ride Openreach until they fix it.
In my experience of supporting cow-orkers who need connectivity to telecommute, If you are with the Big6, you've got 2 chances of them doing that (Fat chance and no chance)
Very happy with my phone.coop connection - precisely BECAUSE they don't take any shit from Openreach.
.... is to wait until about three days after a very popular TV series has released an episode and then download it from the torrents early in the morning. Give it about 10 minutes to settle down and see how fast it's coming down the line. Note: This is for technical test purposes only. When it's finished downloading, or before, you must stop the torrent and delete it. You must not play it or store it on your computer.
I see many threads conflating two things:
- The speed of access to an arbitrary destination site.
- The connection speed to the provider.
The former is under nobody's control (or many bodies' control, which is the same thing), but it's what the customer PERCEIVES as his internet speed. The latter is very much under the provisioner's control, or at very least his responsibility.
Maintain a distinction between the two and a lot of these arguments will just vanish, because they don't make sense. It's the provisioning speed we care about in this context, and nothing else.
"Maintain a distinction between the two and a lot of these arguments will just vanish, because they don't make sense. It's the provisioning speed we care about in this context, and nothing else."
Nope, because for the customers it's the former they want improved upon, toot sweet. Those who know better are in the minority and don't dish out enough dole to make a difference.
re: Maintain a distinction between the two and a lot of these arguments will just vanish,
With ADSL I would disagree. In my area all the ISP's initially offered "up to 8Mbps" and then with ADSL2+ up'ed that to "up to 24Mbps", the fact that in my village the maximum speed was around 1Mbps using either service, was a problem because it inhibited usage.
With FTTC, I suggest that for the majority of the general public, provided they reliably get a 'few' Mbps (circa 10Mbps?) and so can watch streaming HD media and do some web browsing at the same time, then the actual line speed and the perceived speed can diverge.
Line speeds, whether theoretical or measured, are completely different to what is actually achievable in actuality. For example, whenever I get a notification that there is an update for my computer I log on for the update. When it is a large update there are a number of files to be downloaded. As the files download there is a display in Apper of the current rate of downloading. This can vary from a few hundred kB/sec to a couple of thousand kB/sec, because the files are not all on the same server!
I doubt that I have ever had a download measured at the rate indicated by line speed checkers.
I've never understood why companies are allowed to charge for a service that isn't as advertised. If you get 5Mb/s you should pay for 5Mb/s, not for a service "upto 25Mb/s". Gas, Electricity and water are metered and charged for as provided, why isn't broadband?
If companies could only charge a maximum of the actual achievable linespeed then there might be more incentive for them to improve linespeeds. As long as they can charge for a 8Mb/s service but only deliver 0.5Mb/s then why bother?
Because the alternative is metered Internet, where you get charged by the megabyte (this actually applies upstream, with upper-tier providers that either have to pay their way or agree to a peer exchange and maintain a balanced load). There was a time people were faced with this with their dialup or mobile data plans. Since so much of our lives uses mobile data, it tends to create more headaches than we want, which is why people tend to prefer flat rates, both for Internet and for telephone so that they can set a budget and not have to worry about going over.
No, it isn't metered usage. Friar proposed charging based on the actual capability of the line, not the actual use. This should be relatively easy to measure.
His analogy wasn't the best, as gas and electricity are metered based on actual use. But his proposal was essential to measure the capacity of your particular line, and bill for that. This could be done periodically, from either end (or both for validity checking). And it would have an upper bound (the currently advertised "up to" rate and its fee).
Being charged based on actual use is still metered, since it essentially means you're being measured and charged based on that measure. Remember back in the dialup and early mobile broadband days? Not only were they metered, but they were overcharged. Do you really want to go back to that?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
