Net Neutrality - A Lofty Goal

‎12-09-2007

Net Neutrality is proving one of the most divisive, if interesting arguments of the current time. Understanding the debate can be a little complicated. We wanted to set out our position on the topic in a short essay, which we hope will prove a useful read for our customers and others involved in the discussion.

Structure

We begin by explaining what people regard as Net Neutrality what it means, using some off site links with background information and references.

Our wholesale costs govern how we operate and shape our views on this topic, so we've explained how they work.

How we design our products relates to the arguments about Net Neutrality, and we've explained that link; not every gigabyte is equal in cost. As we'll illustrate, each Mbps of capacity (which costs approximately £180 per month per Mbps) can be used differently and be spread amongst different numbers of customers.

We'll conclude by putting all that together and explain what Net Neutrality mean to PlusNet and the how that works, and will change in the future for our products.

Key Points

To sum up our views:

Traffic shaping and prioritisation of traffic are actually required for there to be neutrality on any real Broadband network. Providing an appropriate experience for each type of traffic is what should matter, rather than the source of that traffic (i.e. VoIP should always work well and needs priority over file downloads, regardless of whether you are using the PlusNet VoIP product or someone else's)

There should be no need for content providers to pay ISPs to deliver content, and in doing so it seems likely that other non subsidised traffic would suffer as a result, however there is an economic problem here that needs a solution. If it's not the content provider, consumers need to be prepared to pay a fair price for the data they use.

Broadband products advertised as 'Unlimited' put ISPs in a position where their motivation is to restrict the amount of content customers consume in order to keep costs under control. This is not good for consumers, content producers or ISPs ultimately. That's why we never use the term 'Unlimited' and are honest about what customers can expect.

PlusNet is very keen to find ways of working with consumers and content providers to offer a transparent and fair solution to a problem that is only going to get more serious over the coming months and years.

PlusNet is calling for a debate on this topic which separates the emotions from the facts and centres around the real economics of broadband delivery (especially in the UK), not a theoretical position that puts reality to one side.

What is Net Neutrality?

We'll start off with a Net Neutrality definition and link to Wikipedia and there's also an interesting article here in which PlusNet gets a mention. There's also an article on the Beeb about the US Justice Department's thoughts on Net Neutrality. For some people Net Neutrality equals completely clean feeds, feeds that are completely free of of any "interference" from the ISP or anyone else and where the packets from one site or application are treated exactly the same as any other site or application. Some of the proponents of net neutrality foresee that without certain protection being put in place one company can pay another for a higher priority on their network, or even pay to have a competitor blocked. In the ISP world this could mean that a particular content provider could pay an ISP for priority for their traffic and block out all their competition. This is wrong and something we hope doesn't happen, there should never be a split in traffic in this way where someone with large amounts of money can price or block others out of the marketplace. Going the other way of total neutrality though is wrong too, it's almost a fundamentalist position or an ideology. What we want to do is provide an economically affordable broadband service and that means some differentiation needs to be made for some types of traffic. For us, the key determining factor behind net neutrality is the network cost. We see different types traffic on our network and each different type of traffic can and will have different costs to us. Consider first a mobile phone operator. They don't charge their customers any different if the call was a productive call or a non-productive call because they don't cost them any different (assuming length of call, destination, time of day, etc. are the same) but a call to Mongolia will cost the customer more than a call to a UK landline because it costs the mobile operator more. In the event of a major disaster they may also prioritise for example police phone calls if the network is very busy or areas have low capacity. In these cases it's easy to see why you may not want complete "call neutrality", to make the cost of calling Mongolia the same as the UK landline to the customer is likely going to result in an increase in cost of the UK call to subsidise the more expensive call. Whilst the lack or priority calls for the police in a security incident could put people in danger if the police are unable to make a call. Let me expand further. Consider an hour of peak time usage, the different types of applications that could be used, how much bandwidth they use and the pattern of that usage. We'll consider three different applications; web browsing, gaming and binary Usenet. General web browsing will use a small amount of bandwidth with small "spikes" of usage and larger gaps when the connection is idle. Gaming will use a continuous, but generally small stream of bandwidth that importantly requires low latency. Binary Usenet will try and maximise as much of the available download bandwidth as possible (whilst P2P will try and maximise both upload and download). There are plenty of other traffic types in-between and these won't necessarily fit these as hard and fast rules, someone using MySpace for example may use more than someone using Usenet, it depends exactly what they are doing, but it illustrates the pattern. If you consider three scenarios, the first a central pipe with just customers browsing the web, the second a central pipe with just gaming and the third just Usenet. We'll assume there's no traffic management and all the customers can download at 6Mbps. The central pipes each have a capacity of 155Mbps and the design of the pipe means that as you get to around 90 to 95% utilisation some customers will start to see ping spikes and packet loss. On the Usenet pipe you'll get to about 26 customers downloading at 6Mbps before the bandwidth is used up, but the nature of Usenet traffic means that a little packet loss or ping spikes doesn't have much of an adverse effect on the actual downloads. On the gaming pipe though, if we assume a constant download of 128kbps and upload of 64kbps for each customer then you can have somewhere in the region of 1000 to 1100 gamers using the bandwidth, any more and you start to see packet loss and ping spikes which will cause problems for gamers, so you can't use all of the bandwidth and have to leave some free. For the browsing, let's say that each person views a page once a minute, each page uses a 2Mbps burst of bandwidth for 2 seconds (not completely realistic I know but it's useful to illustrate). You can therefore serve pages to about 75 customers at a time and 30 x 75 = 2250 over the course of a minute. We pay for bandwidth in 155Mbps chunks; a 155Mbps segment costs £290,200 (ex VAT) per year. On top of that is a £7.56 charge per year per customer for bandwidth rental and a monthly rental for the connection of £7.63. These central pipe costs are fixed regardless of how many customers are using it and what they are using it for. So the cost per customer is higher the more bandwidth the applications use. With a finite budget for capacity this is where our traffic management comes in. We don't differentiate on the content itself, it makes no difference in managing the traffic if a P2P download is a Linux ISO or an episode of Lost. One obvious question is will PlusNet prioritise its own services above those of competitors, for example the PlusNet Broadband Phone service. The answer is no, we won't, we want to create a level playing field for all types of VoIP and doing that means using QoS as there's always going to be a finite amount of bandwidth. When looking at the design of our Broadband Your Way products the higher the subscription price the more bandwidth is included with that accounts. BBYW Option 1 at £9.99 has 1GB, Option 2 at £14.99 has 8GB, etc. In setting these allowances we assume that usage is spread evenly across the month (excluding overnight) and so create a "kilobit per second per customer" chart by dividing the included usage allowance by the number of seconds in a month (we actually use a figure slightly lower than the actual allowance because the assumption is not every customer will use all of the included allowance but for illustration we'll just use the included amount and we'll make the calculation over a 16 hour day, i.e. the 8am to midnight chargeable period on BBYW). So the number of seconds equals 60 * 60 * 16 * 30 = 1,728,000 and plugging this into the formulas we get the following table taken from the Broadband Your Way Blueprint: Approximate pricing calculations for BBYW:

BBYW Option + Price	VAT + Line rental + Overheads + Margin	Network Capacity Allowance per month	Average Daily Kbps (16 hours)	Average Designed Monthly Usage
1 - £9.99	£9.64	35p	3.2 Kbps	0.7GB
2 - £14.99	£12.19	£2.80	26 Kbps	5.6GB
3 - £19.99	£12.84	£7.15	65 Kbps	14GB
4 - £29.99	£19.32	£10.67	97 Kbps	21GB

As Ian said in the Blueprint the actual design can be more complex than this, but it does illustrate the point. As the included usage increases so too does the mean kilobit per second speed. With a higher kilobit per second speed we can therefore deliver faster speeds on higher bandwidth applications. Which is why Usenet and P2P speeds are higher for example on Option 3 than Option 2. This design, unlike a one size fits all product, doesn't have lighter users subsidising the usage of heavier usage and allows people to choose a product suitable to them at an appropriate price point. So whilst our products don't block certain types of traffic, on BBYW Option 1 you can still download via P2P or Usenet or iTunes or any other source of content, you may not find that the speeds on each type of download are as fast as say BBYW Option 4. We aren't looking to be a gatekeeper to the Internet or control what people use their connections for and limit choice, all we are doing is passing on the cost of supplying the service we provide in the fairest manner we can. The price point a customer pays is buying not only a certain amount of bandwidth but also a certain level of network performance. And the BBYW product set is designed so that you can buy in at the level you want whether that's a light user at £9.99, someone that's happy to schedule large downloads overnight on Option 2 or 3 or someone that wants a larger allowance during the day on Option 4, We've also designed it so you can change between them or set your own amount of included usage should your usage habits change. We do want to give customers the choice to pay for a "neutral product" and plans are underway, more details will be published in due course. I'm sure there's plenty of debate to be had here and we're more than happy to discuss this further in our discussion forums. I don't think I'd like to see a "two tier Internet" but so long as the products are designed right and the costs charged in a fair way I don't think that will happen. Complete net neutrality is a nice goal to have, but as per the subject, a lofty one because of the economics involved. Dave Tomlinson