Web daddy Tim Berners-Lee calls for net neutrality in Europe Sir Tim Berners-Lee, HTTP server visionary and founder of the World Wide Web Foundation, has penned a blog post on the European Commission web site titled Net neutrality is critical for Europe's future. Sir Tim defines net neutrality as “ the principle that each ‘packet’ of data must be treated equally by the network” and says …
I want my VoIP packets to be given an easier ride through the network than those halfway through my big download of the day.
Now saying that all VoIP packets should be treated equally, and all VoD packets should be treated equally is a different statement. It permits competition in each marketplace, and allows for the network to prioritise it's load according to the traffic type it sees.
I don't care if my latest patch download takes 61 or 62 minutes, but I do care if my VoIP call has a 1 second dropout...
I'm not sure that packet equality id quite what I want (always assuming that bandwidth is limited somewhere)
What you want is QoS. Which you can do on your own network, but providing QoS at a core network level is a little trickier.
For example, if the network is saturated with VoIP data who's VoIP data gets priority?
Ideally the core networks should never really run out of bandwidth, the slowest link in the chain would always be the last link (to your device) which means as long as you determine your priories - you're all good!
Mostly - but I reckon that VoIP >>> Torrent, even in the core.
Streaming audio/video comes inbetween somewhere, with buffering it's less critical than comms traffic
though.
The concept that the core network will never be congested is great, but all that means is that you are paying more for the service than you should be. Network infrastructure should be running near capacity in order to make itself worthwhile - that's just basic economics.
Hence the need for some management, although I am absolutely against the concept of discriminating between different sources of the same data type. Having my neighbour's VoIP traffic get priority over my patch download is fine - because the same will happen in reverse as well...
"Having my neighbour's VoIP traffic get priority over my patch download is fine - because the same will happen in reverse as well..."
I'm not decided either way on prioritising on the core network, there is a flip side to the above. If you don't use VoIP, should your connection suffer because your neighbour does? What if all of your neighbours do and had a conference call together? How badly would that affect your connection with low priority packets? Is that fair? I can't decide.
QOS done reasonably shouldn't have too much of an affect.
On a large download jitter is not relevant, latency is not relevant (below a silly threshold which we can ignore). On VoIP both of those are much more important, but as has been said the actual bandwidth required is quite small, so the download will still take the lions share, it will just see a significant amount of jitter...
The concept that the core network will never be congested is great, but all that means is that you are paying more for the service than you should be. Network infrastructure should be running near capacity in order to make itself worthwhile - that's just basic economics.
We've been led to think that telcos need, and are entitled to, excessive and market detrimental amounts of profit. I don't believe that their greed is holy. Prioritizing traffic is a way for them to oversell what they have, and an excuse for not putting very much of their profit plunders into infrastructure.
As for VoIP taking priority, that's fine on a local network, but even there it shouldn't be necessary. I've got low bandwidth and can still squeeze in a VoIP session here even while downloading and browsing. It really takes an insignificant amount of bandwidth - even with uncompressed codecs. If the backbone is running so close to capacity that it doesn't have room for VoIP then we're really in trouble.
“Imagine if a new start-up or service provider had to ask permission from or pay a fee to a competitor before they could attract customers? This sounds a lot like bribery or market abuse - but it is exactly the type of scenario we would see if we depart from net neutrality.”
I think I might have to try to memorise this quote - it's the most succinct and thought provoking version of the Neutrality argument I've heard so far.
People are used now to ADSL, Fiber and Cable speeds and going against net neutrality will end that, period. Detractors can say all they want but they forget (or conveniently forget) the MSN, AOL, CompuServe et al of the past. If was not for the growth and demand for new stuff of people online, we would still live with similar services. If I start to see my internet access crippled, I don't doubt I would spend less and less time online. The same for millions of people as well I'm sure, or at least the ones without money for a 1st class internet. Less customers, less consumers, less advertisers... It just scares me if Microsoft were successful with their MSN...
Sorry Sir Tim. You need to get your science right, this sort of packet treatment leads to uncontrolled performance collapse for everyone. The science says it, so does the practice.
Network Neutrality is far too weak a goal, it doesn't meet the end user requirements (which is for applications to work) nor the network operators goals (to create value chains from which they can extract a reasonable reward for their capital and effort).
I think you need to enlarge on your point.
Net neutrality as outlined by TBL, doesn't preclude ISP's offering different services, just as they do today: namely Residential/consumer and Business. Nor from providing services from layered infrastructure ie. the various content overlay networks such as Akamai, that have differing costs associated with them. Nor does it preclude traffic shaping/management, provided such shaping is based on traffic dynamics and not economics. However, I can see some people getting upset when their general internet access (including third-party VoIP) gets choked due to their bandwidth hungry subscription IPTV service being given priority on their connection...
Hence I don't see the point you are making other than to imagine that net neutrality means no QoS and no traffic shaping/management.
Nor does it preclude traffic shaping/management, provided such shaping is based on traffic dynamics and not economics.
And herein lies the rub: how do you differentiate the two? As indicated elsewhere, good network management often includes prioritization of traffic, This discussion reminds me a bit of the "benevolent dictator" approach to running a nation. If the ISPs are left to determine how to do this, they will eventually, inevitably decide to maximize profits, especially in the short term, no matter the quality of service provided to customers. If there are consistent standards and rules across the board, there will be other problems, to be sure, but they will be lesser than the worse case of service providers gone wild.
It would just lead to a redefinition - or clarificaiton - of what Internet services are.
People will shift high value traffic from the public Internet and have it delivered over private networks to the edges of that public/private interface. If my ISP wants to offer superior Netflix performance, they'll just invite Netflix to plug the end of their private network directly into a port on their routers.
Net neutrality will exist and be perfect - but people will still be able to pay to have their traffic delivered with higher priority by simply not routing that traffic over the public Internet. How do you think Akamai's business works? They promise better access to streaming content by avoiding the public Internet and giving ISPs direct access to their servers.
I think it simply clarifies what is acceptable business practise.
So my ISP can either buy into Netflix's direct access service or Netflix can provide it 'gratuitous' to my ISP, either are valid. However, in neither deal can there be a direct connection between this and the handling of traffic delivered over the 'public' channel. So if the direct Netflix service is only available via subscription, then non-subscribers should not be discriminated against by having their service deliberately reduced. Likewise in no part of the Netflix deal should subscribers to a competitive service, such as LoveFilm, be impacted.
However, as I indicated above, it would not surprise me to find that ISP's with digital TV services will make bandwidth reservations, just as they do at present with respect to the landline telephone service. (I seem to remember one of the 'tricks' BE did to wring the last bit of performance out of a traditional line was to switch the landline phone over to VoIP and so enable a different approach to line signalling and give customers a significant speed boost - important when ADSL was struggling to achieve speeds of 1Mbps).
"Likewise in no part of the Netflix deal should subscribers to a competitive service, such as LoveFilm, be impacted."
But they inevitably would be - that's the point. It's impossible for a Lovefilm stream riding the best efforts public network to arrive at the edge more reliably than a Netflix stream on a clear, private, guaranteed throughput connection. And then we arrive at the same situation that people are complaining about today - a new film streaming service couldn't compete because they can't afford the private connection Netflix have paid for. Net neutrality doesn't fix the problem, because the established players just move that traffic off the public Internet.
Net neutrality has a gigantic unintended consequence - it provides a huge incentive for established content providers to move the distribution of their content off of the public Internet and onto private connections. That makes offering a good service to customers hard for new content providers and hard for new ISPs.
To use an analogy - it doesn't matter how many rules you write for the postal service demanding that they treat all parcels equally if Amazon then offer to deliver all my parcels by a courier. I still get a better service from Amazon than I do from the rival.
"But they inevitably would be - that's the point. ..."
But here you expose the fallacy at the heart of some people's view of what net neutrality is. The key point, your observation missed is, coupled with the transfer of Netflix traffic off the public internet on to a private delivery service, is there is a 'coincidental' negative change in the QoS of similar services such as Lovefilm that also use the public internet. Yes there most probably will be a user perceivable difference in QoS between the service delivered via the overlay network and that delivered via the public internet, but that is business as usual, unless the ISP and content provider engage in some form of restrictive practise to hinder others doing similar.
The hard case to detect and assess will be between services that both use the same channel and determining whether traffic from one or other is being favoured or hindered.
' Sir Tim defines net neutrality as “ the principle that each ‘packet’ of data must be treated equally by the network” and says he thinks that means “there should be no restrictions based on economic motivations.” '
He doesn't say that bulk services should trump real time. His argument is that "Economic motivations" should not impact a packets network traversal.
EG. Don't hobble Netflix because you own Sky or don't hobble iPlayer/Video streaming because you can't provide the bandwidth that you have over sold to your clients.
"He doesn't say that bulk services should trump real time. His argument is that "Economic motivations" should not impact a packets network traversal."
Fine, my economic motivation will then simply lead me to get my packets delivered more reliably than my rivals by using private network overlays. If I plug my server straight into an ISP's edge router it doesn't matter how fairly the public Internet packets are treated, I still win.
Net neutrality doesn't achieve what people think it will. There's a chance it could actually make things worse - we'll see a flight by established providers to private overlays, followed by reduced investment in the public Internet because all the stuff customers want is coming down the private route.
There is no escape route from TimBL's quoted words, and they are just plain wrong.
The network *requires* differentiated services to work properly: voice packets need low jitter and low bandwidth, video packets require moderate jitter and high bandwidth, other content packets can tolerate jitter and need all the spare bandwidth, and network control packets like routing protocols need to get absolute priority. The network neutrality zealots simply prefer to ignore this inconvenient truth, and we risk a debacle as a result.
I think TimBL needs to read a book on queueing theory.
The man has a point. Google already gatekeeps how much advertising exposure sites can get on the web. Rationing bandwidth and selling to the highest bidder won't help.
On the other hand there is something to be said for traffic SHAPING as long as it remains inherently neutral. Fairly easy to do on your own LAN. But tricky tricky on the Big Bad Web. May the internet saints preserve us
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
