They could make Skype a peer-to-peer technology, then implementing network rules to throttle this traffic would be far more difficult to implement on the backbone networks.
Another own goal by Microsoft ?
US cellphone networks are all throttling video to some extent, providing lower-quality stream to their customers, and some are purposefully undermining Skype as an alternative to their services. That’s the upshot of a ten-month study by Northeastern University's College of Computer and Information Science set up to see what …
OK, so separate out the control channel for call establishment that probably goes back to the main skype site from the main data flow in the data channel and you just end up with a bunch of ephemeral ports between two seemingly random endpoints. You can't just block ephemeral ports since they are used by everything else too.
Now assume that the control and / or data channels are SSL encrypted, you don't hold the keys to decrypt that, so you don't know what the dynamic port ranges are for any given connection either.
So how do you propose creating rules to do that ?
Even if you could, consider the impact on the backbone infrastructure of trying to manage such a huge state table.
How would that work with devices that are frequently behind firewalls or NATs, or changing from one cell tower to another. Writing a client that can send a video signal pier to pier and receive it is not that complicated. Making a good one might take some effort, but the technical aspects are easy enough. However, as with every other pier to pier thing, you end up dealing more with networking than many will accept. Even things that are mostly pier to pier often rely on a central register of who is involved and how you can find them, and sometimes that register will also have the functionality to assist with connections because there isn't a direct open channel to that device. If you can overcome these problems, you'll have fixed quite a bit.
The solution to peer to peer NAT and firewall traversal others use is to connect both sides to a mediation server so that they can more or less simultaneously open connections to each other. While this isn't 100% reliable, it works most of the time and for the few times it still doesn't work, you already have a mediation server to act as a go between.
How would that work with devices that are frequently behind firewalls or NATs, or changing from one cell tower to another.
I think the NAT stuff was dealt with in the 90s, maybe earlier. Certainly has never been an issue with peer-peer links.
As far as cell tower changes, in 2007 I had free mobile data (don't ask how, I just had it ok? And without any mods to my phone or hacking of the telco). I tethered my laptop to the phone often and went for trips passing multiple cell towers. At no stage did the IP change, even though the tower changed a large number of times in the 20-30 miles travelled (greater distances meant going through no-reception zones). If I passed through those zones, I could re-connect on the other side.
Stuff like Skype (pre-MS), ICQ etc worked fine without so much as a glitch in the video (such as it was on early 3g and sometimes 2g connections).
I think the longest trip we did was close to 100 miles passing through a large number of towers and 2g zones in the country but 3g in some towns. Peer-peer worked fine for video calls back then, and I'm pretty sure nothing has changed in networking that would prevent them still working.
As to the throttling itself, intriguingly it is not consistent across video providers or operators, suggesting that there may be deals between certain mobile phone networks and certain video streaming companies to let their videos pass through unthrottled.
Or, some of the throttling is actuallly due to periods of congestion, which means it's not at all throttling.
I would love to understand how the researchers controlled for the natural mechanisms of the mobile video client to negotiate video quality. Not all phones or apps are created equal, and some like Skype are clearly nothing but utter shite at this stage.
The results are only the result of testing. No smoking gun in the form of documents. So is it intentional or just the result of happenstance on time of day by the researchers or it could even be some towers and not all in a given region or owned by a given operator. Then as another commentard has pointed out, this was funded by Google so what restrictions, etc. were placed on the researchers? Too many unanswered questions here in this report.
"The results are only the result of testing. No smoking gun in the form of documents."
The results of adequate testing beats documents any day of the week. The real question you raise (and it's valid) is was this testing adequate, and is anybody attempting to replicate the results?
I much prefer the carriers throttle some hipster douchebag's video stream, so I can continue to make calls in the same cell. This is not evil.
Speaking of evil, the students acknowledge that: "This material is based upon work supported by ... a Google Faculty Research Award". This perhaps should be mentioned in the report. Any study funded by Google will find what Google wants it to find.
Far more useful to me is finding out who is congesting my local cell, so I can go round and throttle them myself. (Joke).
It is evil. Who are you to decide that your cell call to your girlfriend is more important than someones training by video? (See, it's easy to make sweeping examples to prove your case when generalising so much)
If someone has paid for data, they have as much right to use that data anyway they want, just as you have.
Now, if the terms and conditions say that general connection speeds may be slowed down across the board to prioritise cell voice traffic, then fair enough.. Other network managements techniques can also be used if they are in the terms and conditions.
If not, it's the cell companies fault for not provisioning enough capacity, and/or selling contracts to the public that they can't meet.
And, in keeping with the whole Net Neutraility point of the article, there should be no distinction between data traffic from skype and data traffic from youtube...
Voice streams are handled on different blocks of frequency than data streams. There will never be a time when someone's data streaming prevents you from making a phone call. Voice channels are also much smaller than data (What with voice only requiring a few Kbps for a reasonable quality voice call). If the tower runs out of channels to give voice service to newly arriving users, it will take a data channel away from the heaviest user and split it into voice channels. If it runs out of data channels to take away, it will renegotiate with connected handsets for lower quality voice codecs. And even if all phones connected are using the lowest-quality voice possible and the tower still needs more frequency, it will put voice channels into an emergency-only mode allowing only calls to emergency services. But the possibility of that happen would be almost non-existent (you'd have to move the entire population of an entire metro area into a small area for that to happen).
Some networks are set up so that if a cell gets overwhelmed, it will reduce its power and adjacent, less used towers will increase their power levels so their channels can be used to handle the additional load.
I'm amused to see that the Thameslink rail service (Kettering to Brighton via London) has introduced free wifi on some trains. Quite a surprise really as it's generally a costs-cut-to-the-bone service.
But to my delight, they don't permit video streaming and limit total usage per user per trip.
This. I probably actually talk on my phone once a week. If someone is calling me and they haven't arranged to do so via text first, then I know that it's a telemarketers, friend or family that is having an emergency, or one of my two friends who refuse to adapt to the new social more that you don't call someone without arranging it first unless it's an emergency.
Iridium satellite (the legacy version, not 'Next') is famously 2400 bps, including the same data rate for plain old normal telephone voice calls. Doing some testing today, I was impressed with the perfectly acceptable voice quality and apparent lack of any noticable latency. It was just like a telephone.
Last I heard, the PSTN was still using 16,000 bps on their digital voice connections.
What over-the-air data rate is the cellular network using for voice calls these days? If it's more than 2400 bps, then perhaps there's an opportunity here.
Depends on what codecs the handset and network support. Can be anywhere from 2400 bps on an old network and handset, up to 96,000 bps on a new phone attached to a new network that has been set up for the clearest voice possible (Extremely rare outside of a lab environment). It'll also depend on congestion since the tower is going to downgrade a step at a time until there is sufficiently large block of free bandwidth for handsets entering the cell. The other end of the call would also affect the voice channel (no point in giving someone 32,000 bps when the other end is only sending at 4800 bps)
Short of pretending that human nature is inherently good, so-called "libertarians" such as Pai could simply mandate that carriers publish their applied network management policies.
After all, the "free market" ideal assumes perfect information on the part of consumers. Here we're not dealing with a force of nature that really needs scientists to tease it out, just the ripples emanating from boards of directors.
But Pai won't do that, because he's actually still a lobbyist.
Unfortunately, those "policies" are more like guidelines, which management will tweak constantly and may well violate randomly with ad-hoc decisions from day to day.
Publishing them would reveal to the world how amateurishly it's being run. I don't know of any company, in any industry anywhere, that would be up for that.
That's probably to remove the perception that their network is responsible for later throttling. If you have problems and call for support, they can ask you to test whose fault it is by going to several different video providers. Since they are all fast for the 30 seconds you will test obviously Sprint is not to blame!
> "Since they are all fast for the 30 seconds you will test obviously Sprint is not to blame!"
Just as a hypothetical, what if they discovered that the average vid call user only really looks at the image during the onset of the call, and thereafter spends more time parsing voice and merely using the vid for non-verbal clues? Given that, it might be acceptable (for them) to throttle later in the call and still not get complaints.
I have no idea what the author, Kieren McCarthy, is going on about. The FCC never applied Net Neutrality rules to cellphone networks. They were always off on their own regarding Internet access. Net Neutrality rules were specific to land based Internet access: Cable, POTS and fibre.
Also, what's this silliness?
"And second, net neutrality – while it is important – is not as important as the frothing-at-the-mouth advocates believe it to be."
What 'frothing-at-the-mouth' are you frothing about? REAL Net Neutrality is about consumer rights. Anyone considering that a 'frothing-at-the-mouth' goal could only be:
B) Working for a frothing-at-the-mouth ISP desperate to screw over their customers.
Why would anyone sane want to bow down to an ISP corporatocracy, which is in many parts of the USA equivalent to a monopoly? There is very little choice of ISPs in the USA and incredibly horrible access to broadband Internet outside of urban and suburban areas. REAL Net Neutrality was part of the overall effort to provide equal quality broadband across the USA, despite the efforts of ISPs to do exactly the opposite. But it turns out that even some urban areas have had horrible access to broadband Internet. I'm most familiar with the mess that's been going on in New York City whereby both Verizon and now Spectrum/Time Warner Cable/Charter have screwed over the city with lousy access to their services. The city as well as the state government have attempted to overcome this nonsense to the extent of:
A) Providing Verizon with an added fee on their bills to pay for extended broadband in NYC, which Verzon merely ate and never applied to build out of their network
B) Spectrum/Time Warner Cable/Charter merely ignored, cheating the city out of contracted build out of their network. This is specifically why the Attorney General of NYS sued Spectrum in January of 2018, why Spectrum was fined by NYS in June, 2018, and why the New York State Public Service Commission voted at the end of July, 2018 to revoke its approval of Charter's 2016 purchase of Time Warner Cable, ordering Charter to sell the former TWC system. That booting out of Charter and re-sale of TWC is still pending.
IOW: How about researching what you're talking about before writing uninformed articles such as this one?
T-Mobile is pretty clear that they have throttled and non-throttled multimedia plans, though they're not always both available to purchase. A confusing and constantly changing system of passes can temporarily remove throttling. I have a 3000 day no-throttle pass with the option to turn throttling on and off at will, as if that makes any sense.
2) Ofcom is aware of developments and will intervene in the event of market failure. Or a minister gives them a swift kick.
For better or worse, UK's still mostly self-regulating. The EU's been muttering about Net Neutrality, but not had much of an impact yet. Brexit may have more given all the European traffic going through the UK and if they don't play nicely, perhaps we can throttle their connections.
UK approach generally seems to be having a non-discriminatory approach to traffic. This report's challenge is the researcher's ability to differentiate between congestion, and active traffic management. But that's always been part of the problem, ie using traffic management to prioritise time sensitive traffic during congestion.
As for the wider political landscape, apart from us IT types and NN lobbyists, most of the world its happy to ignore the holy war.
The UK has a reasonably good market for fixed line consumer broadband and mobile phones, as most places within a town or city have at least 10 options for each.
The back-end is the market failure, however as the near-monopoly BT/Openreach are highly visible they generally only abuse their position by refusing to build capacity unless the Government pay for it, and by nixing competitors when they try to build capacity.
In the US, each region has a local monopoly on fixed-line - there is often exactly one supplier! Thus they price-gouge and can get away with almost anything.
Mobile phones in the USA are similar, but not quite as bad.
I pay you to deliver xGB at x speed.....it is that simple....if I want more I will pay and you deliver....it is that simple....if I go over the limit, cut me off.....but just deliver xGB at x speed..because that is what I have paid you to do!
Oh and while I am on about it....Ajit Pai ....no, no...do not get me started, just, just.... */ Hits head on keyboard so hard I now need to go to A&E /*
Love and kisses
All internet users
Verizon, my service provider, is up-front with their throttling. They openly say that they throttle all known video streams to an equivalent of about 420p (I usually get about 360p on YouTube). I can't easily tell the difference between that and a 1080p stream on a cell phone screen, anyway.
It's a bit of a lifestyle issue. Watching HD videos is something that I'll do at home, relaxing with my 3rd glass of wine, over wifi connected by fiber. It's rare that I'd need or want to watch video over a cell phone connection.
(Acknowledge in advance others are not the same as me.)
"[...] others are not the same as me"
That is so true, but you miss the other part: Some people may not be in the same situation as you. I, for one, work abroad with a six-hour commute twice a week and thus watch a lot of stuff when on the
run go. What better way to enjoy yourself, when reading a book becomes too much bother? Or when your book has run out of power?
Biting the hand that feeds IT © 1998–2019