I think this is an exciting development for apartment dewellers in places like the Netherland and San Francisco, but the DSL problem in America is in rural areas where even 6MBps is possible.
My parents have been capped at 1.5 mbps (in practice closer to 768 mbps) since around 2003, and they're about 1/3 to 1/2 a mile away from the closest box. Over the years I've made multiple requests that resulted in improved line quality (change in PoP wiring, removed some home outlets with corroded contacts, remove bridge taps on the lines) but Ameritech-SBC-AT&T have never offered a better speed. They abandoned DSL in favor of U-Verse which, for rural locations, is of absolutely no use since they're never going to run fiber out to those semi-remote and remote DSLAMs. I wish something like G.fast would be useful in the US but the locations for which the distance isn't an issue are generally locations which have already been better served for a decade or longer by another technology than copper (cable, FTTH, FTTN, even Wireless in some cases).
If your parents are 1/3 to 1/2 of a mile away from the closest box with those speeds, then it's likely the 'box' is a crossbox (wiring junction) instead of a DSLAM (networking card). Where the DSLAM is that is feeding that crossbox... well that could be much further.
If the loop length is .34-.5 miles (which is quite good) with no impedance mismatch or bridge taps or other various copper faults, then you would/should be able to sustain 10+ Mbps on an ADSL2+ connection.
My guess is that the distance is further than it seems, and the line may not have been reconditioned from phone to internet purposes.
Anyway I bring it up because distance is the biggest determinant of bandwidth when it comes to copper, and if the distance is great then there is little that fixing copper faults can do to improve internet speeds at your parents' place.
It should be a DSLAM (by now) as it was a regular junction that was replaced and added to with one of the larger white DSLAM boxes. I could be mistaken though, so that's a fair thing to point out.
If there is a DSLAM right next to the crossbox, then these are the two possibilities:
[1] (Common and very likely) The DSLAM they installed there is a fiber node and feeds only VDSL over < 50 ft wire to the crossbox. ADSL also feeds into the crossbox, but not from the new DSLAM. It still travels from the CO (central office) and may still be 1-2 miles of wire away.
[2] (Uncommon and less likely) The DSLAM has VDSL and ADSL cards in it. In this situation the DSLAM can support high speed VDSL to very close residents and ADSL to further residents. The CO is no longer the source of ADSL signal, and this is typically a nice reduction in loop length for people.
In your shoes, I'd get in touch with the ISP and see if you're CO fed or not. If you're CO fed, there's little that can be done to improve the speeds =(. You could ask if it's possible to do pair bonding to improve your speeds. Just beware, switching to pair bonding means you'd need to lease/purchase a new modem that can do bonding and the ISP would have to roll a truck to wire up the 2nd pair.
I had a friend who worked for an ISP in Montreal, and he showed me the map of coverage of the island, colour-coded by which exchange served that area.
There were a lot of rectangles and squares, but there were also some weird shapes, strange cutouts, or long stretches. I think there were even some islands of coverage A surrounded by coverage B.
You can be fifteen feet from the CO, but that doesn't mean that's the one they ran your lines from.
That's one DSL problem. In urban areas the lack of speed is also a problem; I am constantly being invited to sign up with AT&T but we stay with cable for the sake of network speed (work VPN, among other things). It'd be nice to have proper competition. a 50 meter distance to the repeater takes in a lot of cities and even some suburban areas.
I live in the heart of San Francisco. Blocks From haight Ashbury even. Fasted DSL speed ATT will offer to my house is 768Kbps. This is utterly appalling. I had faster home internet service than this in Canada in 1996. Comcast is the only other option at $85/month all-in for "50"MB burst service which is really 20-30MB sustained service. Either way 50MB is 5% of the speed of Gigabit and twice the price of Fiber in a google city.
When people speak of grand global projects to better connect grossly under-served communities around the world to the internet, I suggest, you mean like San Francisco and New York?
I grew up in poor, rural Appalachia. You can typically get 20 to 30 mbps cable there for $50 to $75 depending on the provider. While that's not cheap, it's also not bad if you're ditching your $80 or $120 / month TV plan and moving to Netflix / Hulu / soon HBO etc.
I'm not sure where these rural areas in America are, that are poorer and or with less population density than Appalachia (North Dakota? Wyoming?), and that have nothing over a few mbps, but it certainly blows my mind that they still exist.
If the issue is specifically with DSL, then I find it hard to justify that it should continue to exist in rural areas if it lags so far behind cable.
So nice to read about all this exciting network technology that won't matter at all because CenturyLink and Verizon et al are still going to artificially throttle the services I would use them for.
Thank you to the engineers for still trying, though.
Friend of mine is still on EarthLink DSL, as Verizon wouldn't rollout to her town, and I think they're really the market for this.
I can't be sure about this, but I've read somewhere that FiOS has stopped new deployments. In a lot of these places, CenturyLink isn't available and everyone hates Comcast, so the few independent ISPs around may benefit along with their customers.
I moved a couple of years ago. At my old place, I had the choice between cable and DSL. Verizon was the DSL provider, and the best they could give me was 1.5Mbps. I went with cable.
At my new place, I have the choice between cable and fiber. Verizon is the fiber provider, and they'll happily sell me up to 500Mbps if I want to pay for it, and they deliver. That's symmetrical, too, as of a couple of months ago.
I'm not a fan of big ISPs, but I don't think it's quite as bad as you say either.
He's referring to http://blog.level3.com/open-internet/verizons-accidental-mea.... Verizon will happily take your money to give you 500 Mbps to anything on their own network, but for anything on the internet at large, be prepared to wait during peak traffic hours.
I've had occasional problems with YouTube (I don't use Netflix) so it's true that this is a real problem, but for 99% of what I do, I'm able to max out my 300Mbps with no problem as long as the other end can do it too.
I don't mean to cheerlead for Verizon here, but the fact is that higher speeds aren't made pointless, even if they're not quite as good as they could be.
It's peak usage traffic that's affected. During the day, when nobody is using the network, it should be pretty zippy. In the evening, their connections to the internet will be congested, and I wouldn't be surprised if you had trouble with 1 Mbps streams. It's ludicrous that they charge for 300 Mbps service while actually providing measurably worse service than 20 Mbps Sonic DSL, and they have gotten away with it.
One of the big hurdles that FTTP faces is getting the fibre into the customer's house. Perfectly doable, but expensive to achieve. G.Fast enables the upgrade to ~gigabit speeds without needing to drill through brick walls and/or ensure that the customer is at home.
I am happy to be home at anytime if it means I get Fibre. As OP mentioned these technologies are only really suitable in specific cases such as in apartment buildings. In a suburban environment no one will ever see gigabit speeds.
You might well be a reliable customer and at home for the engineer visit, but not all of the general population is quite so helpful!
The head of BT's R&D group gave a presentation in Cambridge yesterday. Slide 15 of his deck shows real world speeds ranging from 700Mbps-1000Mbps for vectored G.Fast obtained in a live suburban environment: http://cwbackoffice.co.uk/Presentation/PrestigeLecture_21.10...
I think those values are optimistic. http://en.wikipedia.org/wiki/G.fast#Performance says that the speeds reported are actually shared between uplink and downlink. Furthermore this was in laboratory conditions. Most people have copper lines that are at least 20 years old.
One way to get around customers not being home is to make Fibre rollout opt-out. Then only wire Fibre to the outside wall of every customers house. At a time that suits them they can organise a tech to come out and complete the connection.
Those values are measured in real-world conditions, not the lab - they've deployed to 3 homes in a housing estate near Ipswich from one distribution point - though I believe that you are correct in saying that they are aggregate throughput.
You can't dig up peoples' gardens without permission! That suggestion also requires 2 separate truck-rolls, and still relies on someone being home for the appointment to complete the connection. Of course it can be done - is being done - but relying on the customer increases the cost.
I'll believe that when I see it (not the technology - that actually has a higher chance of happening than DSL providers offering 1Gbps for a reasonable price).
One of the issues here is that many of these DSL breakthroughs are under ideal conditions. One of the things slowing down DSL is the poor quality of the wiring in many older homes (read: most homes).
I was surprised to learn that mobile networks use a lot of xDSL links to their base stations. They're faster to get provisioned than a traditional data link. So, anything which increases bandwidth in DSL will benefit mobile networks.
not on a 200m loop - that's only 600 feet or so, with where most cell sites are, you might as well just trench the fiber in the last little bit. I know that 'yellow' carrier just extended fiber to most of their network footprint, with microwave serving the rest.
This is of great benefit for in building wiring - I'd love to see a table for speed/distance, that'd give you a better idea of real world practicality. OSP is honestly in pretty decent shape in most of the US, its a lack of careful conditioning and poor records that causes most DSL issues.
Singapore is an extremely small, very rich country with hyper population density. I don't see how that's a great example. They should have some of the fastest Internet speeds on the planet.
How about in Russia, Canada, US, China, Brazil, Australia, India, Argentina, Kazakhstan, Algeria, Congo, Saudi Arabia, Mexico, Indonesia, Sudan, Libya, Iran, Mongolia, Peru, Chad, Niger, Angola?
Those are the largest countries in the world. Presents a drastically different challenge wouldn't you say?
In Australia at least it's not actually a problem. While we have an incredibly low population density, it turns out that almost all (92 to 95% or something) of the population actually live in fairly highly populated cities and towns. The vast majority of the country is mostly uninhabited apart from a thin strip around parts of the coast and then a few inland cities.
This is why the previous Australian Government set out to build a wholesale-only fibre network that the private carriers could competitively deliver services over, to reach 93% of the population (the remaining 7% is a combination of satellite and fixed wireless). This was to fix a problem that had been created by a past Government when the old Government-owned monopoly carrier had been privatised into a single huge anti-competitive company who doesn't do any upgrades unless somebody else does - such as when another company put in an HFC network, the monopoly (Telstra) overbuilt almost the exact same areas with a competing network, not touching anywhere else...
Unfortunatley the new Government is dismantling the new plan, and replacing it with a token FTTN effort to "save money" because of a fallacious "budget emergency", even though financial modeling shows that the FTTP plan actually had a better chance of making a profit (which was to be re-invested in expanding the fibre footprint and upgrading the network). Strangely enough, it's the party who created the Telstra problem last time they were in power...
> Singapore is an extremely small, very rich country with hyper population density. I don't see how that's a great example. They should have some of the fastest Internet speeds on the planet.
Why shouldn't at least New York and Bay Area be comparable? Rich and hyper-dense.
New York and the Bay Area aren't all that dense by major city standards.
But the real reason for the difference isn't really density. It is government telecom policy. The countries that deployed fiber have, for the most part, either heavily subsidized it via direct subsidies or tax breaks. In Sweden's case, for example, forced state run utilities to build the backbone for the telecoms.
The US expects the telecom investment to be 100% privately funded. That means it has to be profitable enough to attract capital investments.
Unlike say Sweeden, were everyone has to use the single telecom provider's network, in the US there is no such mandate. Verizon's FIOS network has to compete with the cable companies network (which is vastly cheaper). Fios is only getting around 25% of people who have potential access to sign up. The Singapore network probably gets 90% or more.
If you really want fiber in the entire USA. Give Verizon, ATT, Century Link, and the other incumbent local loop carriers huge tax incentives to do it. Otherwise it'll happen slowly when people are willing to pay a premium for higher speed.
The main issues with this is the fact that the speeds reduce quite spectacularly as you get further from the node - while at 50m you might get ~400 to 800Mbps (depending on the quality of the lines), by the time you're several hundred meters away you're back to the tens of megabits...
It's got to do with how the low quality telephone cables attenuate high frequencies. They're typically thin wire (0.4 or 0.64mm gauge) designed for a voice signal with a bandwidth in the kHz ranges, and then these DSL technologies use tens to hundreds of MHz, which works to some extent over shorter distances but the higher frequency you go, the more quickly attenuate over longer distances.
Then you have electromagnetic interference between different subscriber's lines, which can make a huge dent in performance (literally as high as 40-50%). Vectoring is a technology that tries to model the noise and send signals to cancel it out to reduce this. It's fairly effective, but not completely.
What you get is the requirement to basically have a node, fed with fibre and electricity (unless they're back-powered from houses, which was suggested for G.fast) that only services a dozen or so houses. With the cost of the capital works required, it's debatable whether it's not worth spending the ~$800-$1200 per house (this is actuals from the Australian NBN rollout) to just complete the fibre all the way, which could then in the future be upgraded to 10Gbit without any capital works (just swapping the terminating hardware in the exchange and at the customer premises - no requirement to touch any of the cabling).
The significant question here is whether providers will offer this both ways - i.e. symmetrical uplink/downlink. If the asymmetry is going to persist, then game over for DSL, welcome fibre.
And here in the UK the Gov have labelled superfast broadband as being greater than 2mbps. The speed it takes for rollout of FTTC means we will see this in 2020 when we are all using 6G mobile networks at 10Gbps anyway.
55 comments
[ 6.9 ms ] story [ 123 ms ] threadIf the loop length is .34-.5 miles (which is quite good) with no impedance mismatch or bridge taps or other various copper faults, then you would/should be able to sustain 10+ Mbps on an ADSL2+ connection.
My guess is that the distance is further than it seems, and the line may not have been reconditioned from phone to internet purposes.
Anyway I bring it up because distance is the biggest determinant of bandwidth when it comes to copper, and if the distance is great then there is little that fixing copper faults can do to improve internet speeds at your parents' place.
[1] (Common and very likely) The DSLAM they installed there is a fiber node and feeds only VDSL over < 50 ft wire to the crossbox. ADSL also feeds into the crossbox, but not from the new DSLAM. It still travels from the CO (central office) and may still be 1-2 miles of wire away.
[2] (Uncommon and less likely) The DSLAM has VDSL and ADSL cards in it. In this situation the DSLAM can support high speed VDSL to very close residents and ADSL to further residents. The CO is no longer the source of ADSL signal, and this is typically a nice reduction in loop length for people.
In your shoes, I'd get in touch with the ISP and see if you're CO fed or not. If you're CO fed, there's little that can be done to improve the speeds =(. You could ask if it's possible to do pair bonding to improve your speeds. Just beware, switching to pair bonding means you'd need to lease/purchase a new modem that can do bonding and the ISP would have to roll a truck to wire up the 2nd pair.
Hope that helps!
There were a lot of rectangles and squares, but there were also some weird shapes, strange cutouts, or long stretches. I think there were even some islands of coverage A surrounded by coverage B.
You can be fifteen feet from the CO, but that doesn't mean that's the one they ran your lines from.
When people speak of grand global projects to better connect grossly under-served communities around the world to the internet, I suggest, you mean like San Francisco and New York?
I'm not sure where these rural areas in America are, that are poorer and or with less population density than Appalachia (North Dakota? Wyoming?), and that have nothing over a few mbps, but it certainly blows my mind that they still exist.
If the issue is specifically with DSL, then I find it hard to justify that it should continue to exist in rural areas if it lags so far behind cable.
Thank you to the engineers for still trying, though.
I can't be sure about this, but I've read somewhere that FiOS has stopped new deployments. In a lot of these places, CenturyLink isn't available and everyone hates Comcast, so the few independent ISPs around may benefit along with their customers.
At my new place, I have the choice between cable and fiber. Verizon is the fiber provider, and they'll happily sell me up to 500Mbps if I want to pay for it, and they deliver. That's symmetrical, too, as of a couple of months ago.
I'm not a fan of big ISPs, but I don't think it's quite as bad as you say either.
I don't mean to cheerlead for Verizon here, but the fact is that higher speeds aren't made pointless, even if they're not quite as good as they could be.
You can rag on Verizon for being anti net neutrality, but they are the only significant deployment of fiber to the home in the USA.
The head of BT's R&D group gave a presentation in Cambridge yesterday. Slide 15 of his deck shows real world speeds ranging from 700Mbps-1000Mbps for vectored G.Fast obtained in a live suburban environment: http://cwbackoffice.co.uk/Presentation/PrestigeLecture_21.10...
One way to get around customers not being home is to make Fibre rollout opt-out. Then only wire Fibre to the outside wall of every customers house. At a time that suits them they can organise a tech to come out and complete the connection.
You can't dig up peoples' gardens without permission! That suggestion also requires 2 separate truck-rolls, and still relies on someone being home for the appointment to complete the connection. Of course it can be done - is being done - but relying on the customer increases the cost.
It would be interesting to see a graph of distance/speed comparing it with regular DSL.
http://www.ericsson.com/ericsson/corpinfo/publications/revie...
How about in Russia, Canada, US, China, Brazil, Australia, India, Argentina, Kazakhstan, Algeria, Congo, Saudi Arabia, Mexico, Indonesia, Sudan, Libya, Iran, Mongolia, Peru, Chad, Niger, Angola?
Those are the largest countries in the world. Presents a drastically different challenge wouldn't you say?
This is why the previous Australian Government set out to build a wholesale-only fibre network that the private carriers could competitively deliver services over, to reach 93% of the population (the remaining 7% is a combination of satellite and fixed wireless). This was to fix a problem that had been created by a past Government when the old Government-owned monopoly carrier had been privatised into a single huge anti-competitive company who doesn't do any upgrades unless somebody else does - such as when another company put in an HFC network, the monopoly (Telstra) overbuilt almost the exact same areas with a competing network, not touching anywhere else...
Unfortunatley the new Government is dismantling the new plan, and replacing it with a token FTTN effort to "save money" because of a fallacious "budget emergency", even though financial modeling shows that the FTTP plan actually had a better chance of making a profit (which was to be re-invested in expanding the fibre footprint and upgrading the network). Strangely enough, it's the party who created the Telstra problem last time they were in power...
Why shouldn't at least New York and Bay Area be comparable? Rich and hyper-dense.
--submitted from an 4 megabit connection in SF.
But the real reason for the difference isn't really density. It is government telecom policy. The countries that deployed fiber have, for the most part, either heavily subsidized it via direct subsidies or tax breaks. In Sweden's case, for example, forced state run utilities to build the backbone for the telecoms.
The US expects the telecom investment to be 100% privately funded. That means it has to be profitable enough to attract capital investments.
Unlike say Sweeden, were everyone has to use the single telecom provider's network, in the US there is no such mandate. Verizon's FIOS network has to compete with the cable companies network (which is vastly cheaper). Fios is only getting around 25% of people who have potential access to sign up. The Singapore network probably gets 90% or more.
If you really want fiber in the entire USA. Give Verizon, ATT, Century Link, and the other incumbent local loop carriers huge tax incentives to do it. Otherwise it'll happen slowly when people are willing to pay a premium for higher speed.
It's got to do with how the low quality telephone cables attenuate high frequencies. They're typically thin wire (0.4 or 0.64mm gauge) designed for a voice signal with a bandwidth in the kHz ranges, and then these DSL technologies use tens to hundreds of MHz, which works to some extent over shorter distances but the higher frequency you go, the more quickly attenuate over longer distances.
Then you have electromagnetic interference between different subscriber's lines, which can make a huge dent in performance (literally as high as 40-50%). Vectoring is a technology that tries to model the noise and send signals to cancel it out to reduce this. It's fairly effective, but not completely.
What you get is the requirement to basically have a node, fed with fibre and electricity (unless they're back-powered from houses, which was suggested for G.fast) that only services a dozen or so houses. With the cost of the capital works required, it's debatable whether it's not worth spending the ~$800-$1200 per house (this is actuals from the Australian NBN rollout) to just complete the fibre all the way, which could then in the future be upgraded to 10Gbit without any capital works (just swapping the terminating hardware in the exchange and at the customer premises - no requirement to touch any of the cabling).
Until you realize that there are 5m houses.