We have had a lot to say about Network Neutrality in the past and the acid test to any policy that seeks to lay down the rules by which internet traffic can be prioritized, is whether or not an operator can deliberately distinguish traffic from rivals and block it.
There is a lot of hot air about operators like AT&T and Verizon stopping Google from getting a free ride, and making it pay for service, but the truth of the matter is that consumers pay for broadband, so whatever services they choose to use, the transport is therefore already paid for. Americans wonder if perhaps it’s alright to cap broadband supply at a set number of Gigabytes and wireless networks the same, but indiscriminately capping a service is fine ‘ if you can get away with it without losing your customers.
What is not alright is to slow down AT&T U-Verse packets if you are Comcast, or vice versa, or to slow down all video packets because you have a competing video service. Let’s face it, video is where the growth is, and video is so data heavy it’s where the savings are. Faultline has always thought data caps are perfectly reasonable in a market that is ‘natural.’ By that we mean in a market which has naturally developed with true competition.
It’s NOT alright for a quiet meeting of AT&T, Verizon, Comcast and the other cable operators to all agree on a cap and unilaterally introduce it, because that’s price fixing. But if there is strong competition, then one player might choose to introduce a cap and others might follow, which is fine.
But in the US there is little broadband competition, because there is no independently assessed broadband wholesale pricing mechanism for the major ISPs. In Europe and most of Asia there is.
So CLECs as they were called, and independent ISPs, find the cost of doing business tough in the US and often the majors can siphon off their profits with punitive pricing. In Europe the process is transparent, in the US wholesale pricing is opaque, and has been defended in the past by the courts. That aside, cable hates telcos and telcos hate cable, and in most markets there are two or more ISP providers. So we can suggest that there is probably some level of competition, even if it’s not fully natural. So capping is probably alright. But as we said, targeted slowing of a rival’s traffic should be verboten, or chaos would ensue.
So take a look at the Verizon Google statements this week over their agreed mechanisms for offering varied internet delivery speeds for different types of traffic, but view it in that light ‘ does it break the golden rule?
The companies are proposing that ISPs should not be able to ‘degrade’ traffic which already travels over the Internet, but might introduce new types of services, which they call managed services, which might be prioritized.
Let’s quickly look at this. If you give priority to traffic using a Quality of Service (QoS) layer, here’s what happens, you can open up superhighways across networks, often using MPLS routing techniques. This involves sending a few packets ahead, telling routers that they should entirely forget about a certain amount of capacity because it is now reserved, and that it can recognize this traffic and manage it at a lower level in the network, which makes it even faster. It’s an instruction which says, ‘any traffic with this label on needs to go this route through the network and it always goes to the front of the queue.’
So what happens to any other traffic which is constantly being pushed to the back of the queue ‘ well it fills up buffers until it is about to overflow and be lost, and then it is intelligently thrown away. The way the internet works, if a packet goes missing, the receiving end asks the sending end for a resend, so discarded packets create more back traffic and calls for lost packets.
This is fine. HMTL packets, for instance, on web pages, arrive eventually and the service is simply best effort anyway. If the network is busy, it takes longer to download a web page.
QoS is used in IPTV systems, but for the most part those packets do not travel over the public internet, but arrive at a DSLAM in the local exchange where they mix in with public traffic. This puts them into the front of the queue on an operator’s own access lines from the exchange. Cable is completely different, but can be made to be roughly analogous.
But whenever QoS is used, it slows services down. So the idea of prioritizing new services includes, by implication, the idea of slowing the ‘rest’ of the traffic down. Now if this is done at a powerful BRAS (Broadband Remote Access Server) in the Telco network, then ALL the rest of the internet traffic suffers equally, so the delay would, initially at least, lead to almost imperceptible delays in packets, but the prioritized packets would shoot across the network at light speed.
As these prioritized packets grew in volume, presumably the network capability would grow and so the ‘other’ traffic would more or less keep up with its old speeds. If a telco is getting paid for this traffic, if has the money to invest in the network.
Verizon and Google are saying this might apply to services such as health care monitoring, the smart grid, advanced educational services, or new entertainment and gaming options.
There are a couple of concerns we have though to the proposed ideas. If a telco or cableco has the tools to dynamically allocate bandwidth and becomes sophisticated in their use, it might find itself unable to resist slowing down the traffic of rivals and hiding the fact.
In the first decade of the commercial internet, packets went slowly because of network inefficiencies and the fact that dial up was the most common connection. In the next decade broadband speeded up access and the networks themselves became better and better managed.
The internet has always been a haphazard connection that varies in speed of page delivery. Over the following decade servers and peering agreements got better and better and capacity was plentiful and network management got to grips with public internet performance ‘ we began to get something which is more predictable. CDNs began to move content to the network edge and Google and other service providers began to put servers all over the US, if not the world.
Internet performance is still complicated, and it is still opaque with some details being held by the ISP only. So if we suddenly have a mild falling away of performance and new classes of performance that never fall away in speed or quality, then there will be a slow perception change that one entertainment service which pays for prioritized access, appears to be a better run service and it will tend to keep customers more readily than a new service which cannot afford to buy priority.
The net effect of that will almost certainly be that the well connected majors will appear to have ‘better internet connections’ than innovative new services. Would Facebook have made it big, if the connections to MySpace had been quicker and slicker? We will never know, but our feeling is that innovation could suffer, and companies which are new will never break through, until they ‘pay an internet tax’ to one or all of the major ISPs.
ISPs have to remember they are providing a service to consumers and they should do the best for the consumer. Killing innovation on the internet does those consumers a disservice.
We won’t tell you that innovation is the be all and end all. The PC stopped being innovated because Microsoft cornered all the key markets in PC software. But we still all rely on PCs and only now are new categories of device coming along to challenge it. The PC has had 20 years of stagnation. So it’s not improbable that the internet could have 20 years of stagnation with a move like this, and still be very useful, as long as the slow lane on the internet doesn’t get perceptibly slower.
Let’s just visit the road analogy for a moment. The Verizon Google agreement in road terms is that they are going to build an extra freeway lane for commercial traffic which pays extra. When they have a lot of people paying for it, they have enough money to build yet another lane, and another. Eventually most traffic would be traveling in the fast lanes, and just traffic from start ups would be in the slow lane. If you want that, then tick the Google Verizon box. But eventually they will stop building slow lanes, and they would be full.