Slashdot Mirror


US Telcos Are Slowing Internet Traffic To and From Popular OTT Apps Like YouTube, Netflix, and Amazon Prime Video, New Research Finds (bloomberg.com)

The largest U.S. telecom companies are slowing internet traffic to and from popular apps like YouTube and Netflix, according to new research from Northeastern University and the University of Massachusetts, Amherst. Bloomberg: The researchers used a smartphone app called Wehe, downloaded by about 100,000 consumers, to monitor which mobile services are being throttled when and by whom, in what likely is the single largest running study of its kind. Among U.S. wireless carriers, YouTube is the No. 1 target of throttling, where data speeds are slowed, according to the data. Netflix's video streaming service, Amazon.com's Prime Video and the NBC Sports app have been degraded in similar ways, according to David Choffnes, one of the study's authors who developed the Wehe app. From January through early May, the app detected "differentiation" by Verizon Communications Inc. more than 11,100 times, according to the study. This is when a type of traffic on a network is treated differently than other types of traffic. Most of this activity is throttling. AT&T Inc. did this 8,398 times and it was spotted almost 3,900 times on the network of T-Mobile US and 339 times on Sprint's network, the study found.

3 of 168 comments (clear)

  1. No surprise by Fly+Swatter · · Score: 1, Informative

    Phone systems have to place priority on phone and emergency services, that means sometimes entertainment data gets throttled to preserve space for those priority services. Thems the breaks. Now if this was happening on a wired land line then I would consider this news. I still think the whole smartphone thing is ridiculous, everywhere you go someone has their eyes addicted to that little screen.

  2. Re:Cause or effect? by sabri · · Score: 5, Informative
    You don't even understand that you don't understand it.

    They created this mess by blocking multicast and web caching to the home because they couldn't bill it. If multicast had been widely available then multiple people streaming the same thing at more or less the same time would not occupy any more of the net than one.

    Nobody blocks multicast. Multicast simply doesn't work like that: it doesn't mean that people can simultaneously stream Youtube or Netflix. That would only work if two or more subscribers would start the same video at the same quality at the same time.

    Furthermore, multicast addresses are limited to 224.0.0.0/4, or 268,435,454 addresses. Not to mention that there is no global multicast infrastructure in place.

    If caching had remained in place, the bulk of the Internet would have remained clear.

    And who do you think is blocking caching? Hint: it's not the ISP. The ISP wants to cache, but in order to do so the content must be clear-text. Oh wait: everyone is moving to HTTPS, which cannot easily be broken.

    Back in 2013 I was working for a large telecom equipment provider on a joint project with a large CDN provider to build a CDN/TIC solution. Youtube, Netflix and all major streaming sites were supported and cached. Until [b]they[/b] decided to break caching by switching to HTTPS.

    Your ignorance in this matter cannot be understated.

    --
    I'm not a complete idiot... Some parts are missing.
  3. Re:Cause or effect? by TheSync · · Score: 4, Informative

    Nobody blocks multicast.

    To be clear, every ISP blocks multicast transport between Internet AS's except in a very few special circumstances, and typically it is not routed within networks as well. It isn't that you can't bill for it, it is the inherent danger of multicast, and also multicast routing doesn't scale well.

    Some end-user ISPs are considering using highly controlled multicast ABR to efficiently deliver live content to their own subscribers, but it is unlikely that multicast will ever be distributed across the Internet.

    Multicast can be used to efficiently deliver popular non-live content as well (for example, see this paper).

    [FWIW I was involved in a multicast ABR trial]