It’s been a long time since I decided to focus this blog on technology issues and I haven’t manage to start posting regularly. I’m sorry for that, but reading and listening to different positions on current net neutrality issues I felt the need to write down my position on this issue and it’s a good oportunity to kick off this new stage.
After the small preface, let’s get into today’s topic. When I hear debates about net neutrality, arguments gather around two camps: the big business camp which argue that they defend net sustainability and the freedom camp which argue that proposed messures effectively kill the Internet as we know it. From my view both arguments are right, but none of the solutions proposed by either camps are valid. What I feel is that big bussiness camp utilises a particular problem around bandwith usage to change the whole model under which the Internet works without worring about (wanting to) put in place technical solutions to tackle it. The freedom camp doesn’t want to tackle it as it might bring the debate into technical proposals that might look too similar to big bussinesses’ ones from a non-technical perspective.
But there are real issues to fix and we need to do something about them:
- Internet use cases have grown. Today, the Internet is used for all kind of applications with very different requirements. For example:
- Real-time communications require low latency and low jitter links
- Back-up services or storage synchronization require high throughput rates
- Common web surffing and apps related traffic stand somewhere in the middle between high throughput requirements and the need of an interactive user experience
- Internet traffic has grown and the potential growth is still huge as today already used applications like video streamming or VPNs are ready to consume increases on available capacity. Video might be streammed in 3D and 4K resolutions and “road-warriors” are willing to access their enterprise files seamesly. Following chart has been generated using data from [1].
- No mayor technical improvements are expected in order to increase bandwidth availability by order of magnitudes. Pilling up fibers help but it’s not likely to be the solution [2].
- Furthermore, on mobile world, the expected improvements will not bring huge new bandwidth avaylability [3]. There is another restriction when we talk about mobile comunications: medium is always shared among users.
- There are still many scenarios under scarce bandwith resources. Mobile networks use to be weak in low populated areas and even in cities, there are many times where signal quality is weaken by tunnels, buildings, congestion and other factors.
Every day these issues remain unsolved, we are giving argument to those that are willing to turn the Internet into their own controlled platform. Big players sell their proposals as the only and optimal solution to current situation for all of us. But the “for all of us” part is the one we shouldn’t trust. They are defending their interests as we should defend ours, that’s what free market is all about. We must be very aware about one thing, there is no other real alternative but to keep the stack as thin as possible. VPNs, Tor, GnuNet and mesh networks (among others) are a great fight-back instruments and put preasure into service providers, making them think twice before implementing too restrictive policies. But at the end of the day, these technologies add another layer on top of the “naked” Internet that affects user experience in many aspects: connection reliability, connection recovery time, power consumption, CPU/memory consumption, latency.
We do need to implement solutions that protect the network from being overloaded by especific traffic or users but we should make these decision in an open manner and making sure that network endpoints are the ones which decides their connection parameters. ISPs and backbone networks should only route traffic and comply with signed level of services (SLAs) which should always be neutral in respect to traffic’s origin/destination.
My proposal:
- Technology is out there, lets make use of it (QoS [4], RSVP [5])
- Three standard categories should be deffined:
- High priority traffic/low throughput/quota
- Normal priority traffic/medium throughput/quota
- Idle priority traffic/best-effort mechanism
- Traffic categories must be universal and category selection must be a client-side decision
- Providers can offer different quotas/prices depending on user’s requirements, but is the client (effectively, client’s applications) who decides which traffic is assigned to each category. OS and applications should be adapted to allow users to easly make this decisions.
- Traffic category is only decided by endpoints. It might only be modified by intermediates if their peer node has surpased assigned throughput/quota. In this cases a chain of responsabilities would be triggered, which should preassure providers which are closer to en users to make sure that their peer providers fulfil their SLAs.
- Internet providers decide on the meassures to apply in case certain customers surpases entitled throughput/quota on certain traffic category. These meassures could be: charge extra money and keep traffic’s category, degrade traffic to a lower priority category or block traffic of that category completely for certain amount of time.
Expected consequences:
Proposed mechanisms will improve user experience and will help to utilize Internet resources more efficiently. Applications that nowadays struggle to work consistently will benefit from high-priority marking. These applications might not consume too much bandwidth but they do require low latencies and low jitter to work properly. I’m thinking about realtime communications.
Most users would benefit from a better user experience as their calls, gaming, apps and surfing data will have higher priority over the rest of Internet traffic unless until they surpass their contracted throughput/quota.
No single service provider will have advantage over the rest. From a user perspective, all services are accesed under the same rules.
Services with high bandwith usage like video on-demand would have to deal with being the lowest priority traffic, but they shouldn’t affect other valuable activities in the network. If these services doesn’t work as desired, it would means that the network is not yet prepared for this kind of services (tough luck) and providers should need to increase investments if their customers are willing to pay for them. Here the word “customers” are not only domestic users but also providers that would “transmit” their final consumer “requirements” to their SLAs in order to guarantee their bandwidth availability.
As some final words, our interest as users is to keep the Internet as service neutral as we can and this is certainly not the interest of any given service.
I’ll be very pleased if you share your views on this issue.
Disclaimer: I certainly, wouldn’t define myself as an expert in this field, but as an informed user (as much as I can) that shows his view on this topic.
References: