It seems BT, inventors of the Cleanfeed system that went spectacularly wrong in 2008 when they tried to filter Wikipedia, causing issues accessing the site for many, now want to go further. Claire “Won’t someone think of the children!” Perry MP has apparently just had a meeting with Mike Galvin, Head of Innovation at BT about the MP’s plans for a porn block and has tweeted to say it was a good meeting.

For those not familiar with Mike Galvin, he was involved in BT’s trials with Phorm, so he has a record with this sort of thing.

Perhaps BT would like to comment?

Since yesterday’s post, I’ve dug up a bit more detail on the new BT service. It’s been attacked in the media as being a “two-tier” system, reminiscent of the Net Neutrality debates.

But far from being able to force content providers to have to pay to provide a decent service to end users, the economics mean that BT will possibly need to give this service away to content providers for little more than a nominal fee to make it work.

The concept is simple – local bandwidth is cheap, internet bandwidth is relatively cheap, (Well under £10 per Mb/s per month in bulk) UK-wide backhaul is expensive. (£40 per Mb/s per month and upwards) A significant portion of the cost of your home broadband (Around half, depending on the package you are on) is not the connection to the exchange but the bandwidth from the exchange to the rest of the world. (Hence fair use caps and lower bandwidth limits on cheaper service providers.)

The solution is to keep popular content as close to the end user as possible. If, for example, I know as a provider that an episode of Eastenders or Coronation Street is going to be watched using an on-demand service by tens of thousands of people there will come a point where it makes sense to just send it out in advance and off-peak (When there is spare capacity) to servers around the country. This does not apply to a YouTube video of a skateboarding dog, unless it’s fantastically popular over a particular timeframe (And you need to know in advance what’s going to be popular), because there is a cost of storing all that data throughout the country to consider.

I do not know the details of Virgin or Sky’s architecture for on-demand services but it’s likely they’ve done the same sums and have worked out if this makes sense for them to implement with their own on-demand broadband/cable services. At least with BT Content Connect, all Content Providers have the ability to submit content but it’s the getting content where BT may struggle.

It’s clearly in the interests of service providers who bear the majority of bandwidth costs to implement this, particularly if they are also a content provider (e.g. Sky/Virgin) but there is less benefit to other content providers like the BBC. Even if the service is free for them, there will still be significant engineering investment to talk to BT’s new Content Connect servers.

So BT will need to sell the idea to the content providers, which they can do by promoting the idea that somehow the content will be better/faster/smoother/gold plated compared to the alternative. But this assumes that there is a congestion problem within the ISP in the first place, which does not currently seem to be the case with BT. Unless BT somehow engineer congestion, which would be a net-neutrality issue, (And no doubt result in complaints of lag from Skype users, online gamers and the like if all services were affected) objecting to content connect on neutrality grounds and claiming a two-tier internet is being created is not quite as easy as some press reports are suggesting: It’s popular on-demand content that’s affected, rather than the internet as a whole and it’s likely similar is already being done in more closed shops (Virgin/Sky) anyway.

In practice, I do not think we will see a wide uptake of this service beyond BT Retail. The technical limitations mean that the majority of ISPs will not be able to use it, although BT Wholesale are (rightly) required to offer any service they provide to BT Retail to everyone at the same price in the interests of competition.

From today’s Telegraph:

BT is starting to sell a new broadband service that allows video content to be viewed in a better quality than other material, it has been reported.

Wait, what? So video could be viewed in better quality than some text? That doesn’t make any sense whatsoever. Five seconds with Google gives this flash-driven marketing abomination from BT about the relevant product, “Content Connect”. From what I can see, it’s just a Content Distribution Network that allows content providers to distribute their content so that it’s as close to end-users as possible without the hassle of running a global network. This isn’t new (Akamai have been doing it for years and they’re not the only one) and as far as I can tell, there is no connection to Net Neutrality at all.

I’m not sure if the Telegraph are trying to spin a story out of nothing or if they have misunderstood a story so badly that they’ve destroyed all rational content in the process.

A couple of days ago, an anonymous individual identifying themselves as “Backdoor Santa” published some graphs showing that Comcast, a major US Internet Service Provider, runs it’s transit links at capacity for much of the day. This sort of tactic has depressing implications for any attempt to legislate for Net Neutrality, although I shall need to go through some basics to explain why.

As a mid-level ISP, there are three groups of people who you send traffic to and from. These are your customers, your peers and your transit providers. Customers, logically, pay you to carry their traffic. More customers means more money although for many ISPs, customers who buy bandwidth off you are probably not particularly profitable compared to those renting servers or space in a data centre.

Next come your peers – people who you are happy to exchange traffic with for free. (Or at least for no more than the cost of running a few fibres or other circuits between your networks) There is a careful balance to be struck here as fibre between sites and ports on network equipment cost money. This means there is a certain minimum amount of traffic that needs to exist between the networks before it’s worth doing, plus you also don’t want to peer with someone when you’re in London, New York and San Jose but they’re only in London as you’ll be paying for all the expensive international bandwidth.

Finally, if you have traffic that’s destined for somewhere that’s not a customer or reachable for free via a peer, you need to have paid transit. Some companies are so big that they don’t have to pay anyone else to take their traffic because they can reach everyone else either as a customer or via peering. The number of such ISPs varies, but is usually less than a dozen.

So why do Comcast run their transit links so congested as to cause problems? Won’t their customers complain? Well, yes but as some sites will run fast it will just be those sites that run slow that seem to have the problem and they’re big enough to get away with this. If someone providing content, say Amazon or Apple, want to peer with them in order to make the site faster they either say “Sorry, No, you don’t meet our requirements” or put the peering on a link running overcapacity too. The result is that those providers then need to pay Comcast as a customer in order to get decent service. If they don’t, Comcast users will go to their competitors for service.

This is definitely not “Net Neutrality” and something that should be stopped – if indeed it does occur in this country – as it is abuse of a dominant market position. Sadly, there is no prioritisation of traffic involved anywhere which is the usual topic up for discussion in neutrality debates. I do not know of any way of separating out genuine engineering requirements and complicated calculations on the cost vs. benefit of peering from this sort of activity.

As with many issues, Net Neutrality is one where I’m somewhere in the middle. Or at least, I was, until Ed Vaizey spoke on the topic yesterday. Prior to that, I would have said I was not a fan of the more extreme “you must never prioritise or block traffic” arguments, simply because in the real world they are not practical.

For example, if my network is subject to a denial of service attack, clearly that traffic is going to get pushed into a black hole. Along similar lines, if I find a zombie machine on my network taking part in an attack or spreading viruses, I might cut that machine off or limit it’s traffic. Even better, as occasionally happens, we hit paydirt and find one of the bad guy’s command and control nodes and take it down, usually after some forensics to get a list of infected machines.

Those are clear-cut cases, but it gets more subtle. During times of peak demand, particularly unexpected demand such as 9/11, networks can become overloaded. Whilst it’s entirely possible to just ignore this and give everyone bad service to all destinations, it’s sensible to try to limit some of the less urgent traffic. So, anything that looks like it might be a routine overnight backup of a PC or Peer-to-peer file sharing traffic such as BitTorrent is going to be lower priority than someone checking out the BBC News web site for example. (Some providers do this all the time, and tend to get in trouble for it)

Even outside of network engineering, we already have a significant amount of “Quality of Service”, run by DSL providers such as BT. Depending on the type of line you order – “Business”, “Office”, “Premium” or similar names are often applied – it’s likely that your traffic will be prioritised over a simple Home user. That’s quite reasonable, given that office users pay more. In fact, it can keep down the price of your home DSL because providers can make their money on business users during the day. With some providers, home users, who will use the network more at night, are just using up capacity that’s already been paid for.

So you’d I’d have thought I’d be on Vaizey’s side, but his excessively market liberal tone has just pushed me further towards thinking we need some kind of regulation. There is clearly an increasing commercial pressure to turn the tools we use to keep the internet running into a source of profit, but not the kind of commercial pressure that helps consumers and keeps prices down and choice up.

He’s proposing no regulation, besides forcing ISPs to publish their policies. I see this as a failure on two counts. Firstly, I do not want to have to publish steps I take against botnets and other sources of evil traffic, even in outline, because it makes their lives easier. They already have some idea what we do, but there is no reason to confirm it. There’s also various underground “good guy” groups where we share information on what’s going on. If we’re forced to reveal what we get up to, those groups in other countries are going to be less willing to share data with us.

Secondly, I doubt people are going to base their buying decisions on published data on how ISPs prioritise data. Instead, they’ll base it on adverts on the TV and cost. The big ISPs, such as Virgin and BT, can quite happily go to the BBC and say “Stand and deliver! Give us some money, or your internet data gets it!”. If the BBC, or any other large organisation they think is good for some money, refuse to pay up then the BBC web site slows down compared to other sites or iPlayer stops working properly. Because the user’s mate down the street has the same problem, it looks like it’s a BBC problem, not their ISP being Evil.

But smaller service providers can’t do that – if a small player such as myself with a user count measured not in millions but tens or low hundreds of thousands tries it on, the content providers will tell us where we can stick it. Because only a small number of users have a problem, and they’re all on our services, the problem appears to the end users to be with us. (Which really is where it is, as we’d be the ones doing the traffic shaping!)

So we could end up promoting monopolistic behaviour in the market. Large providers will be able to cut their prices as they’ve discovered a new revenue stream from the content providers. For those end-users not wanting to be subject to this, they’ll find themselves paying increasing prices as small providers will shrink and lose economies of scale.

It’s interesting to note how the press is running this. Today’s Metro has an anti-Vaizey, pro-Neutrality article in it and the BBC also takes a similar line as there does not seem to be an appetite amongst the content producers to pay to prioritise their traffic over their competitors. It’s the really big providers that want this because they see a source of revenue for no additional cost.

I fail to see, as Vaizey postulates, how this will increase innovation in the market.

P.S. For those of you in the UK, there’s an Early Day Motion tabled by Tom Watson you can encourage your MP to sign.