Sorry for the lateness of the reply. I forgot about this thread.
So if it’s not currently illegal, it must be a good idea? Can you explain that logic to me?
The fact of the matter is that QoS does degrade the connection for traffic marked “lower priority.” That’s the whole point, which is why nobody uses it on the Internet.
Correct me if I’m wrong, but Skype is p2p, so I don’t think QoS is going to work that well. If it doesn’t completely overwhelm the packet queues, it’ll be, at best, useless, and you’ll be wasting your money.
But yeah - if you want a rock solid, reliable link between your corporate HQ in the US and your factories in China, you run a cable (or you rent one from somebody else - who is presumably a telco that’s already reserved a certain portion of their pipes for virtual dedicated links) You’ll pay an assload for it, natch, but that’s the cost of doing business.
VPN’s are pretty much the way that corporate WAN’s are implemented. They won’t be breaking out onto the ‘public’ internet so QoS is viable (and a good business practice for corporate WAN’s).
As far as NN goes (across the public internet), it’s not something I’d heard about before this thread. I can see that there will probably need to be some option available at some point for controlling traffic volumes, but whether NN is the way forward, I’m not sure. There are a lot of broadband providers that offer tiered services based on traffic volumes but what about the unlimited services that you can currently buy? Say I have an unlimited service (i.e. I am not restricted in the amount of traffic coming into or going out of my access point), if I hammer this connection with as much traffic as I can 24 hours a day I am most likely getting a lot more from my connection than I am paying for. This traffic is not just traversing the portion of the public internet owned by my service provider but also the portions owned by a number of other service providers as well (especially in the case of PtP services such as Skype). I must admit I don’t know whether service providers currently cross charge each other for access to the public internet. Does an a company providing unlimited ADSL access to the internet get charged for the traffic traversing other network providers networks (whereby their access pricing schemes differ)?
How will this be managed in the future? Higher charges for your access lines to allow the network providers to upgrade their backbone capacity or, as seems to be suggested, a QoS style service?
Skype is an internet telephony service (VOIP), much like Vonage, only it uses software on your computer as the phone instead of a hand-held device.
Large service providers all connect at together at some point. For example, the Chicago Network Access Point. Companies pay to connect to this point. Once you’re a member, companies set up peering agreements - they agree to pass eachothers traffic for little-to-no cost because of the mutual benefit (example: a google presence and ameritech. Ameritech benefits by speeding up access to google for their customers, google benefits in having the high availability and low latency that the peering agreement allows, improving the customer experience).
Now at this point, all traffic is the same. Everything’s cruising along nicely right? Say Ameritech decides to ‘tax’ google and skype for not only the use of its lines, but to ‘guarantee’ a certain level of service. Only google and skype tell Ameritech to go pound salt. Ameritech signs a deal with Disney to provide ‘guaranteed service’ streaming video over its network at a hefty cost. Ameritech customers could see streaming video from disney before, and rarely experienced any problems, but now they never experience problems. However, when the traffic gets heavy in mid-day all of a sudden, my google searches come up with ‘Server Cannot be found’ and my VOIP calls are being dropped or the delay from end to end is insufferable, all because they didn’t want to pay twice for their network connection.
Now fast-forward 3-5 years. There are multiple levels of service between each provider. None of them are the same or carrying the same traffic. Its not standardized because in different areas different companies are more successfull than others. AT&T, Verizon, Comcast, Qwest and Charter have eliminated 3rd party VOIP companies by either flat out refusing service or ‘offering’ an acceptable level of service at such predatory rates as to bankrupt them. Charter and Comcast customers can’t reliably use their VOIP technology to call AT&T and Verizon customers because AT&T and Verizon won’t pass the traffic reliably, preferring to strong-arm them (the customers) into using AT&T/Verizon’s services, knowing in a protracted war they’d win because they hold and can reach the most customers. Now billy-joe down the street, studying hard for his computer science degree creates a killer-app for his masters project - software that can do IPTV, VOIP, DVR and surf the net on very cheap hardware. He decides its so cool he’ll create a startup. Only his company fails - not because of poor planning or lack of an exciting product, but because of the artificial barriers to entry in the market. None of the carriers give his traffic the time of day because it directly threatens their business model.
Thats an excellent argument, but thats not what they’re looking for. Thats the thing here. No one I know of is complaining about their levels of service from their ISPs. So why is this so neccessary? Its a solution in search of a problem. They’re looking for additional revenue streams, not to give their customers better service.
But then again, we don’t know how loaded the backbone currently is. This may just be thinking about the future. If, for example, the current backbone is running at 60% capacity, with the increase in bandwidth intensive applications it may be that the infrastructure can’t keep up with the demand.
What happens then? Is a QoS type service implemented to prioritse the traffic or is the backbone infrastructure upgraded? If you want to upgrade the infrastructure, who pays for that? Is it passed on to the end users directly? Or maybe a third/forth/fifth solution?
Uh did you even read those links? The first one is talking about how the bandwidth glut is almost over. In any case, the excess capacity was caused by the irrational exuberance of the internet bubble and is not an inherent quality of the network. The early 21st century can be viewed as an anomaly in terms of capacity.
network providers will do what provides the best service for the cheapest price. Laying down new fibre is horrendously expensive, you have to dig up city streets and block traffic and all that jazz. Nobody ever wants to lay down new fibre unless they absolutely have to. If QoS can provide better service without needing to constantly have gobs of spare bandwidth all over the place, then we should be investing in QoS.
As Shalmanese said and I mentioned in my previous post, who is going to pay for this? Just saying that the backbone providers should up their capacity is not really a useful answer unless you specify where the cash for these upgrades is going to come from. How much more are you prepared to pay on your access costs to fund these upgrades?
I still think that just putting in more capacity is not the only way to go. It’s just a temporary solution to the capacity issues, it will have to be upgraded again and again if a ‘smarter’ way to handle traffic is not developed. Sort of like saying that the only way to solve traffic congestion is to build bigger and bigger and bigger motorways instead of looking for alternative solutions to the congenstion in the first place.
Depends on what they’re prepared to offer in return. Obviously the money is going to have to come from end users, and one would hope they’d continue to be charged in proportion to the amount of bandwidth they need, just as they are today.
QoS isn’t any less temporary - its usefulness decreases as the amount of high-priority traffic increases. Adding capacity has benefits for everyone, not just video and voice users, and as long as there’s enough capacity for everyone’s traffic, QoS is meaningless.
Whilst the use of QoS on the public internet is considerably more complicated than on a VPN, I don’t think you can call QoS ‘meaningless’. It’s true that it will suffer from the same issue as just increasing the capacity would do…eventually it will be overloaded by the increase in the traffic volume (as you noted). However, why do we have to follow ‘just’ capacity increase or ‘just’ QoS? Following both strategies, an upgrade in capacity and implementation of a QoS type service, would most likely leave longer intervals between the upgrade cycle for the backbone.