Slashdot Log In
BBC iPlayer Bandwidth Explosion Bodes Ill For ISPs
Posted by
kdawson
on Friday February 22, @10:22AM
from the economics-of-broke dept.
from the economics-of-broke dept.
penfold69 writes "Dave Tomlinson is one of the network gurus at PlusNET PLC, a Tier-2 ISP in the UK. He recently put up a blog post about the ramifications of the BBC iPlayer for the ISP industry in the UK. The post makes some very interesting reading regarding the bandwidth usage triggered by the iPlayer, and raises timely questions about the Net Neutrality debate. The Register also picked up on this story with a good review of who is going to have to pay for all this legal video streaming."
Related Stories
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
Copyright or Tech? (Score:5, Insightful)
all the best,
drew
Re:Copyright or Tech? (Score:4, Funny)
Re:Copyright or Tech? (Score:4, Insightful)
What I don't get is where this cost of x pence per Gb comes from. If an ISP has the wires and the routers all running, why does it cost extra to be sending more data? I see that you might ramp up electricity costs slightly in the systems that route this data when it's processing lots of packets, but I have trouble seeing this being the source of the cost.
Once the infrastructure is in place, then where is the big cost? That's what I'm not getting.
Re:Copyright or Tech? (Score:5, Insightful)
Bingo! You hit the nail on the head.
Bytes don't cost money. The capacity to transfer them does. That T-1 costs the same amount of money sitting idle as it does running at 100% 24/7.
To be fair, the ISPs wind up paying higher costs because they have to purchase more capacity when their users adopt higher bandwidth applications -- but this idea that bytes have a direct cost that can be calculated is absurd. A byte of data is not the same thing as a kilowatt hour or liter of gasoline.
In any case, I don't see how they think they can get away with not investing in network upgrades. Is innovation on the internet going to stop because ISPs would rather rest on their laurels cashing checks instead of investing in infrastructure upgrades for the next killer app?
The standard response to "increase bandwidth" is "P2P apps consume all available bandwidth, increasing bandwidth won't solve anything", but that response overlooks the fact that you aren't automatically obligated to increase the bandwidth provided to end users. Improve your core network while keeping your customers in the same bandwidth tier they currently have and you'll solve the problem of p2p bogging things down.
It would be a lot more fair to provide a 3.0mbit connection that actually delivered what it promised then it is to provide a 10.0mbit connection that achieves that speed at the expense of your neighbor.
Re:Copyright or Tech? (Score:5, Informative)
That's half the problem
The other half is stuck in the last mile. Cable is a bad way to upload a lot of data. Sure there's a lot of bandwidth, but cable has very poor uploading characteristics. Just a few people in the highest paid tier of service using all the upstream can easily deny the rest of the people of the node access to the Internet.
It's not just the ISP, but the last mile technology used. Cable and DSL came about with the assumption that most people download way more than they upload. Unfortunately, Bittorrent doesn't do this (if you want a good ratio, you have to upload as much as, or more than you download). A few people paying for 10M/1M service in a cable node can easily take down the entire node.
You may notice that the companies having issues with this tend to be cable companies. Shaw (BitTorrent throttling) and Rogers (encrypted traffic throttling) in Canada (two largest cable companies), Time-Warner Cable (iTunes throttling, byte metering), Comcast (RST packet spoofing for P2P), amongst others. Cable just can't handle the upstream component of P2P.
Re:Copyright or Tech? (Score:5, Interesting)
There's nothing inherently wrong with over subscription -- it would be pretty stupid to pay your Tier 1 provider to provide a dedicated 3.0mbits for Grandma who only wants to check her e-mail -- the problem starts when they try to cheap out and use a bad oversubscription ratio.
To be fair, a few years ago nobody could have seen the rise of p2p (though foresight should have predicted the rise of streaming video), so that probably changed the ratios they should be using. I lose all sympathy for them though when they whine about how much money upgrades cost.... most of these outfits (here in the states anyway) are literally swimming in profit. It's not as though they are running their businesses in the red and can't afford to invest in upgrades.
Beyond that, I really don't understand this push to "shape" p2p traffic. Wouldn't it be much more fair to just give your customers the highest amount of bandwidth that you can provide them with and allow them to use it as they see fit? What's the damn point of raising the speed again and again if you can't actually provide it to your end users?
Re:Copyright or Tech? (Score:5, Informative)
Ian Wild, a PlusNet employee, left the following comment on TFA:
Not entirely sure what the implications are for caching solutions, but it sure is interesting.
Re:Copyright or Tech? (Score:5, Informative)
Those of you in the US will not be familiar with the UK internet backbone arrangement.
The overriding majority of cost for a UK ISP is the 'backhaul' from the consumer in their house, to the ISP network.
Peering with other ISPs, Backbones and content provider is *very very* cheap, as they practically all peer into Telehouse via LINX.
As Cable rollout is severely limited in scope in the UK, the majority of internet traffic is routed via BT from the consumer to the ISP network. BT have a fixed base price plus a per-GB charge for this facility.
Thus, it costs the ISP to transfer data to the consumer. Caching only helps to reduce the traffic at the ISP peering points (which have negligible cost). It doesn't help reduce the cost to transfer that information to the consumer.
The other alternative to BT is to use LLU (Local Loop Unbundling) providers. These ISP's have installed their own DSLAMs in the various BT exchanges, and rent 'backhaul' off of BT at more favourable rates than paying BT for the entire ATM circuit back to the ISP.
However, the LLU providers are still charged a per-GB fee for the rental of the backhaul.
This means that every bit of traffic passing from an ISP network to the consumer costs a set amount. This is where contention is used heavily (and by BT not by the ISP, actually).
Multicasting won't help, as each multicast stream still needs to be transferred over this backhaul to the consumer, with BT charging for each GB.
Yes, it's retarded, but yes this is how the UK internet industry works.
B.
Re:Copyright or Tech? (Score:5, Informative)
That's why major ISPs peer with major content providers instead of trying to use their main edge connections to pull down all of that traffic. Here in the states I know that Roadrunner at least (possibly others, though I don't have direct experience with them) is working on building out their own nationwide IP network and relying less and less on their Tier 1 provider (Level 3).
I think peering arrangements like this will prove to be more fruitful in the long run then trying to cache the data locally. It's a hellva lot easier to peer with Youtube/Netflix/the BBC/what-have-you then it is to try and mirror terabytes of content on your own network and keep it up to date.
Sometimes supply drives demand (Score:4, Interesting)
Speaking as an American, where all our telecoms basically conspire to screw the consumer and offer substandard bandwidth, I long for the day when the demand for bandwidth surpasses the ability of their crappy networks to handle it, sparking an all out bandwidth arms race amongst providers desperate to cater to the needs to demanding consumers. I dream of the slug-like cable and phone companies being driven under by agile local providers...It will get to the point where small networks will be able to compete, because the advantages of a giant infrastructure are of limited use in a local environment.
So pardon me if I don't give a crap if the little ISPs are feeling the pinch. If they'd used a little foresight, they'd have plenty of free bandwidth.
It should be the ISPs that pay (Score:5, Insightful)
Live by the marketing hype, die by the same.
Regards,
I will! (Score:5, Insightful)
"unlimited" was part of the title of my service plan; so, unlimited bits at the contract rate or I get to sue!
There is no neutrality issue; what we are debating is greed(or incompetence coupled with back tracking and lying) in newspeak!
Re:I will! (Score:4, Funny)
Pure moaning (Score:5, Insightful)
It really wasn't that interesting. He mostly just shows you a bunch of network traffic graphs.
raises timely questions about the Net Neutrality debate.
His argument basically boils down to "Waaa! Customers are actually using their internet connections! The BBC has lots of money, give some of it to us! Waaa!".
This particular ISP may be bitching and moaning but frankly that's because they're discovering they can't compete. Virgin Media (Cable) recently announced a UK-wide upgrade for all of it's customers. My currently 4MB connection is going up to 10MB. I don't hear the any bitching from them, and they clearly wouldn't be doing it if bandwidth was really a problem.
The usual suspects, one would hope... (Score:5, Insightful)
The only issue I can see is that dishonest ISPs want to keep charging their customers the "Unlimted* Fast** internet for the low low price of $X a month!", whilst either denying them the service being advertised by throttling some traffic, or charging the server side twice, once for the real cost and once for "access to consumers".
It's greed and weaseling out of advertised services, pure and simple.
Re:The usual suspects, one would hope... (Score:4, Insightful)
60 ish million folks in the UK.
This sort of thing will only get more common as time goes on a people use the net for ever more and bigger media. Personally I think ISPs need to do more to bite the bullet and price their services honestly, rather than pricing them cheap and then coming up with a million and one reasons you can't have what you thought you'd paid for.
We'll all be throttled (Score:5, Informative)
I'd hesitate to call The Reg "good" on this one... (Score:5, Insightful)
the ISPs promised you and then seriously overbooked for a major profit. It's not that the claims weren't true on the networking
solutions being better overall- it's that greedy people didn't implement what they claimed and pocketed the extra, we can't seem
to get people to move to things like IP Multicast to shed most of that load, and things like the aforementioned.
I don't go boo-hoo for the ISPs. They knew this was going to eventually happen. They didn't prepare for it. They had the
chance to do the right thing and they didn't- and still aren't. All in the name of large profits- something that nobody can
sustain for long, ever. Nobody gets rich quick save by stealing or dumb luck. Once people start remembering that concept
perhaps sanity will resume...naaahhh...we would never have that, now would we?
Misguided (Score:5, Insightful)
The idea that everything must be monetized to have value is irksome and tiring. This fallacy permeates the article and is, in my opinion, why the article sometimes misses the mark.
I think it's also interesting to note that the main point of the article is "ISPs, who are in the business of selling connectivity and bandwidth, are doomed because the demand for connectivity and bandwidth is large and getting larger." Imagine how silly it would be to say "grocery stores, who are in the business of selling food, are doomed because the demand for food is large and getting larger."
The fact that demand is increasing would be a good sign for most industries. (Perhaps the ISPs view it as a bad thing only because they are so accustomed to over-selling their networks and not having customers actually use what they pay for?) This is not the death knell for ISPs, this is an opportunity for them to compete, expand, and sell more of their product. Until they wake up and understand this, they will keep complaining and deliver shoddy service, I guess. But make no mistake: the consumer thirst for high-bandwidth Internet applications is a good thing.
good games? (Score:5, Funny)
Are they some kind of guitar hero/FPS mashup?
The UK's problem is two fold (Score:5, Insightful)
The second issue is the UK's internet backbone, it's simply not up to scratch and doesn't meet todays requirements in terms of bandwidth. Many people laugh when there are articles about how the internet is going to run out of spare bandwidth, but the fact is in the UK it's happening, the whole reason ISPs over the past few years have gone from true unlimited to heavily capped is because bandwidth is having to be rationed, there just isn't enough room on the backbone for everyone's requirements in an unlimited world.
As such, the UK also needs investment in it's internet backbone and whilst BT is bringing implementing 21CN, whilst I don't know the technical details it seems a mere band-aid fix as some people in the industry have commented that there will still be similar bandwidth caps as today.
It's not an unsolvable problem, on the contrary the solution is there - Japan with a population double that of the UK quite happily handles 100mbps connections to end users with the requirement for caps and their internet backbone falling over as a result. There are plenty of other examples like Sweden, however some may argue that as Sweden only has around 1/10th the UK's population that they don't have enough end users to clog the pipes up, hence why Japan is a much better example. South Korea is a decent example also at around 5/6ths of the UK's population. The core issue is politics and who's going to give up short term profits temporarily for vastly improved long term profits.
The UK simply needs investment in it's internet infrastructure, but it needs everyone work together. BT are semi-interested in updating their backbone but quite rightly they think why should they when it's ISPs and content providers that are going to make the money off of it? The fact is that a one off investment (to ensure net neutrality) by the major players is required - BT, ISPs, the Goverment and yes, possibly even the BBC and other major content providers.
It's all very well ISPs complaining it's costing them a fortune currently, but when they're not willing to give up that money to BT for infrastructure improvements then they can't realistically expect a solution.
One final point is that it doesn't help the goverment wasting ISP's time and money with their threats about getting rid of file sharers. It's all very well the goverment, ISPs and BT whining about the problems the UK has with internet access, but when they're all doing nothing about the problems, or in the governments case, making the problem worse then they can quite frankly shut up and put up. The only downside to that is, it's us, the end users that suffer.
It's the architecture (Score:5, Informative)
There are two types of ISPs,
BT / Virgin / Easynet + a few others who have unbundled kit in exchanges and their own pipes to exchanges
Everyone else who resells capacity from the above, who pays a fixed price for capacity irrespective of where in the country it came from.
All that capacity goes back to telehouse where LINX is and all the content and internet exchange takes place.
There is no peering at the local exchanges, or apart from London or Manchester.
So when a two BBC users with the P2P iplayer service but different ISPs, all the traffic goes to London and back again. Even if it's the same ISP the ISP doesn't see it until it leaves the resellers pipes in London at which point it gets shipped back down the pipe it came from. When I downloaded a programme on my laptop that was already on my desktop PC I got a download rate of 500Mbits as it streamed across my internal gigabit LAN - if we had peering at the exchanges and decent ADSL uplinks we should be able to do that within metropolitan areas.
Now this may work itself out - there aren't any really long distances in the UK, so we should be able to run 10Gbit ethernet backhaul between exchanges relatively quickly and cheaply for unbundled providers, but to really do it well we need peering in every major city between the majority of ISPs rather than the current model where every ISP ships all their traffic to London.
Re:Multicast? (Score:5, Informative)
This is distributed in two ways: the first is a flash video player, modelled on youtube, that shows the videos low-res in a browser window. The second is a via a kontiki P2P system, which allows users to download DVD quality DRMed videos onto their (currently Windows, Mac soon, Linux almost certainly never) computer.
The BBC also do multicast via several ISPs, but this is almost completely unpublicised, and apart from news, nigh-on content free.
Re:Multicast? (Score:4, Interesting)
In my suspiciously successful attempts at using this aspect of iPlayer outside the UK, I discovered the actual video data being sent from an Akamai-controlled IP address. So presumably, if ISPs want to control bandwidth usage from this source, they'd just need to host an Akamai node thingy?
The video quality for this 'lesser' iPlayer is still pretty good. I clocked it at about 100kB/s (i.e. ~800kbit/s) - it looks okay fullscreen if you're using the computer as a telly. Haven't tried the Kontiki thing yet - I've been doing this on my Macs...
Re:Multicast? (Score:5, Interesting)
Multicast would be a good idea for live broadcasts, though.
Not that I actually use any of it - my wireless and 2GB cap wouldn't cope. A co-worker found the "always running, even when iPlayer isn't" service recently, though.