The debate in the US surrounding net neutrality is fierce but for some reason it hasn’t really taken off as a dinner party debating subject on this side of the pond. With questions around our precious human rights, it seems odd we haven’t linked the way our access to the internet is controlled to freedom of speech and fundamental liberties as they have in the US; equally we are quick to complain about over charging and under delivery of almost anything but the British only (rightly) complain really about failed speed promises and notspots.
Next generation applications will of course mean we need more bandwidth, and in order to take a full and active role in modern society that bandwidth will need to be delivered widely and in many different forms. But bandwidth is not the only crucial measure of network performance that will be important – latency and jitter will also be critical, but so will the right to use the applications we choose as they begin to touch more and more of our everyday lives.
When I spoke at the NextGen09 conference yesterday I opened by reflecting on the different approaches taken by network strategists in the enterprise and telecoms worlds. An enterprise network architect always starts with a real world user requirement and as they develop that into a technical design they have to keep returning to that original requirement to make sure the user has the optimum experience. In the corporate world deep packet inspection and traffic shaping are seen as positive tools to ensure the optimum business experience.
In contrast, telecoms networks seem to be rarely designed by people who have to report to the users, and as a result deep packet inspection and traffic shaping are tools used to minimise the nuisance of pesky “revenue generating units”. Net neutrality is born out of a broken respect for customers.
Today net neutrality debate is centred on a simple argument – your chosen ISP monopolises your connection to the internet, and if they decide to control your legitimate use then you have a problem. Since most ISP’s behave in a broadly similar fashion, there is little you can do about it – migrating to another provider may improve things in one area but you’ll probably find something not performing as you’d like somewhere else. The original Samknows report into broadband performance, which led to the Ofcom performance report, highlighted many of the tricks the different UK ISP’s employ to limit the damaging effect of customers; traffic shaping so only web and email work properly, blocking or choking certain applications like the BBC’s iPlayer, etc.
But if you outlaw deep packet inspection and traffic shaping you may also be banning the tools which could optimise gaming experiences or ensure your healthcare applications don’t stop working because your kids start watching the iplayer in another room – throwing bandwidth at it won’t make this go away. A little over a decade ago Dave Isenberg wrote his seminal paper “Rise of the Stupid Network” as an antidote to overly complex carrier core networks which still prevail. At the time I didn’t quite agree with him, and I still don’t, but I was much more in favour of his manifesto than that of the opposition so have always kept my concerns largely to myself. My own feeling was that we should have respectfully dumb carrier core networks the just do what they’re told by intelligent edge networks under the control of users and their applications – but since we both feel that they core should be pretty dumb it didn’t seem worth raising a criticism to what was a brilliant piece of thinking.
So before we decide the express our distaste at today’s net neutrality injustice perhaps we really out to be doing something more constructive. We are finally making some progress towards next generation broadband in the UK – so how about making sure that the negative impacts of net neutrality are minimised, and some of the positive aspects are developed.
A few weeks ago I attended the NICC’s Ethernet Working Group’s meeting – this is the body formalising the Active Line Access standard which become our main Layer 2 access method for next generation networks, replacing the likes of PPPoA and L2TP that existing broadband systems use; this is a group of dedicated and brilliant network guru’s who meet to shape the way we use the internet. Essentially the next gen world will be composed of Ethernet virtual networks, and the large part of the debate at the meeting centred on how many VLAN’s should be made available to a single customer. Without going into the details of the debate, the proposal was that a minimum of four, and ideally at least 8, VLANs will need to be available for each and every customers.
Why is this important to net neutrality? Multiple VLAN’s means no customer will be forced to choose a single service provider who can monopolise their access to the IP world. While it isn’t a panacea, it paves the way for specialist content providers (health, gaming, video, etc) to take their own VLAN’s and use all the tricks used on enterprise networks to deliver an optimised customer experience; when you subscribe, you are knowingly buying into an optimised solution.
This just leaves the role of the “catch all” Internet service provider. The simplest answer would be to ensure vanilla internet transit, from customer to the server, is just that – plain and simple with nothing added. Implement a standardised set of qualities of service with clear descriptions which mean something to ordinary people but leave the allocation to customers and their applications. This doesn’t need to be complicated – a tick box in the Skype connection options so it tags packets with a suitable quality of service would work, as would a sensible default profile in customer routers which tags ftp and email as bulk traffic, clearing the fats lane for more sensitive applications.
Let move the debate on – and focus on how we want the next generation to work in everyone’s interest.