Stories
Slash Boxes
Comments

SoylentNews is people

SoylentNews is powered by your submissions, so send in your scoop. Only 19 submissions in the queue.
posted by martyb on Monday May 23 2016, @05:16AM   Printer-friendly
from the speedy-or-not dept.

Netflix has launched a speedtest service at Fast.com

Advantages:

  • no advertising
  • no Flash
  • IPv6 capable

Disadvantages:

  • requires JavaScript
  • begins download test immediately upon page load
  • does not test upload speed

What other sites do you use? How accurate are they? How do you test your connection?


Original Submission

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 4, Interesting) by canopic jug on Monday May 23 2016, @05:27AM

    by canopic jug (3949) on Monday May 23 2016, @05:27AM (#349789) Journal
    In the past, some ISPs have been checking the destination of connections and prioritizing sites known to be speed tests, especially their own. In that way they can throw off the results. So while the site may be useful for a while, it will soon not be a realistic speed test for ISPs that misbehave using that trick.
    --
    Money is not free speech. Elections should not be auctions.
    • (Score: 0) by Anonymous Coward on Monday May 23 2016, @05:53AM

      by Anonymous Coward on Monday May 23 2016, @05:53AM (#349799)

      I guess the news is fast.com is new so it hasn't received preferential treatment from scummy ISPs, yet, and it's not full of ads for malware, yet.

    • (Score: 2) by Hairyfeet on Monday May 23 2016, @06:08AM

      by Hairyfeet (75) <{bassbeast1968} {at} {gmail.com}> on Monday May 23 2016, @06:08AM (#349805) Journal

      This is why I use two different tests (speedtest and testmy.net) and have my DNS going through a third party and NOT my ISP. If both tests are damned near identical? Then I can be relatively sure I'm getting a straight result.

      BTW using my method I tested this new site if anybody is interested and it gave me the same results as the other two, and quite quickly I might add so I'll be adding this to my list...thanks!

      --
      ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
    • (Score: 4, Interesting) by butthurt on Monday May 23 2016, @06:33AM

      by butthurt (6141) on Monday May 23 2016, @06:33AM (#349810) Journal

      On Slashdot, JcMorin [slashdot.org] had written:

      I did a quick lookup and the download are made from:
      https://ipv4_1-cxl0-c273.1.nyc... [nflxvideo.net].... I believe this is the same domain as the real video so that would make it harder to block on "fast.com" since data are not downloaded from that domain name.

      If that commenter and my interpretation of the comment are correct, this may be accurate for checking one's connection to the Netflix servers. Of course, when speed testing sites build a better mouse trap, ISPs may build a better mouse.

      • (Score: 3, Interesting) by maxwell demon on Monday May 23 2016, @07:37AM

        by maxwell demon (1608) on Monday May 23 2016, @07:37AM (#349823) Journal

        Indeed, it's easy to think of ways to trick this. For example, prioritize the first few minutes of any stream from that address, so that the speed test will give great results, while for the actual content only the opening credits are delivered fast. Or temporarily prioritize if previously a connection to fast.com was made.

        If your ISP starts hiring VW engineers, you should get suspicious. ;-)

        --
        The Tao of math: The numbers you can count are not the real numbers.
  • (Score: 2) by Aiwendil on Monday May 23 2016, @06:13AM

    by Aiwendil (531) on Monday May 23 2016, @06:13AM (#349807) Journal

    What is wrong with just wgetting something you was going to download anyway?

    Just remenber to pick a good server to download from (In my case I often end up with something from SUNet (swe), FUNet (fi) (those are university networks) or something from a debian mirror)

    • (Score: 2) by sjames on Monday May 23 2016, @06:48AM

      by sjames (2882) on Monday May 23 2016, @06:48AM (#349814) Journal

      You'll need to know their policies and load for that to be accurate. Many sites rate limit connections in order to better handle the total volume and/or because they need headroom for their own use.

      • (Score: 2) by Aiwendil on Monday May 23 2016, @06:56AM

        by Aiwendil (531) on Monday May 23 2016, @06:56AM (#349818) Journal

        Interesting, since I've never got less than my max bandwidth from sunet I never thought of that - thanks for pointing it out

    • (Score: 0) by Anonymous Coward on Monday May 23 2016, @11:21AM

      by Anonymous Coward on Monday May 23 2016, @11:21AM (#349859)

      Streaming is a bit of a different beast. youtube-dl is a good tool. I suspect somebody out there has an equivalent tool for Netflix.

    • (Score: 2) by Hairyfeet on Tuesday May 24 2016, @12:02AM

      by Hairyfeet (75) <{bassbeast1968} {at} {gmail.com}> on Tuesday May 24 2016, @12:02AM (#350073) Journal

      Uhhh because 1.- The majority of users aren't on Linux, 2.- Aren't messing with command line, 3.- Have plenty of things they want to do where speed would be an issue but it can't be downloaded in the traditional sense, such as online games and streaming?

      Whether we like it or not it appears the future is gonna be SaaS and streaming, the corps love it because it is damned hard to pirate by Joe Average and the users love it because they don't have to give a crap about things like storage or dealing with hard copies, its "push button get stuff" simple and in this day and age convenience tops just about everything.

      --
      ACs are never seen so don't bother. Always ready to show SJWs for the racists they are.
  • (Score: 1, Informative) by Anonymous Coward on Monday May 23 2016, @06:43AM

    by Anonymous Coward on Monday May 23 2016, @06:43AM (#349812)

    This isn't a speed test, it's how fast you can download from Netflix. It's intentionally misleading to profit them, obviously.

    It's also not surprising that there's no upload speed test. Netflix is all about you being a passive receiver, not a full participant of the Internet.

    • (Score: 4, Insightful) by butthurt on Monday May 23 2016, @07:13AM

      by butthurt (6141) on Monday May 23 2016, @07:13AM (#349819) Journal

      It's intentionally misleading to profit them, obviously.

      The most obvious explanation, to me, is that they wish to provide their past, present and prospective customers with information as to whether an Internet connection is adequate for watching their streaming video service. If fast.com accurately shows the speed of one's connection to Netflix, people who would have a good experience may be less hesitant to signing up; people who would have a poor experience may decide not to sign up. The proportion of satisfied to dissatisfied customers may increase, resulting in fewer cancellations, fewer support calls, and a better reputation--and that would profit them, obviously.

      Why they would be motivated to provide false results is not obvious to me. Perhaps you'd like to share your analysis.

    • (Score: 1, Insightful) by Anonymous Coward on Monday May 23 2016, @02:34PM

      by Anonymous Coward on Monday May 23 2016, @02:34PM (#349916)

      They are attempting to show that ISPs are selectively throttling back Netflix traffic. That is the punchline of the whole thing, and a bit sad that a "tech" website misses the point.

      • (Score: 0) by Anonymous Coward on Monday May 23 2016, @10:09PM

        by Anonymous Coward on Monday May 23 2016, @10:09PM (#350044)

        We're too busy arguing about Presidential candidates.

      • (Score: 2) by butthurt on Monday May 23 2016, @10:34PM

        by butthurt (6141) on Monday May 23 2016, @10:34PM (#350055) Journal

        If they make the site "intentionally misleading" it will just take one person with wireshark and a Netflix subscription to catch the lie, by comparing the throughput claimed by fast.com with the actual throughput from that site and from their regular streams.

        If they throttle their bandwidth for users of certain ISPs, I personally wouldn't know how to prove that it wasn't the ISP or a transit AS restricting the throughput, but someone noticed them doing it [archive.is] (I didn't watch the video, which may explain the technique).

  • (Score: 2) by maxwell demon on Monday May 23 2016, @09:42AM

    by maxwell demon (1608) on Monday May 23 2016, @09:42AM (#349839) Journal

    I could not imagine how to make a speed test web site without Javascript.

    --
    The Tao of math: The numbers you can count are not the real numbers.
    • (Score: 2) by Aiwendil on Monday May 23 2016, @09:50AM

      by Aiwendil (531) on Monday May 23 2016, @09:50AM (#349843) Journal

      Page is loading please wait [sample time at server].. [stuff massive amount of uuencoded inline data of a _very_ bloated progress bar (some formats allow for almost endless comments at frequent intervals)] .. time for page to load was [sample time at server] sample2-sample1 for an average of Math..

      Just remember to turn off any serverside caching..

      (Btw, isn't this of all that AJAX is done?)

      • (Score: 2) by tibman on Monday May 23 2016, @08:15PM

        by tibman (134) Subscriber Badge on Monday May 23 2016, @08:15PM (#350004)

        Unfortunately, this can't be done without javascript. For example, how will the browser know when the file is done downloading to call the server for the second time sample? Have to use javascript. A common trick for zero javascript programming is you can refresh the page every few seconds to check the server for updated information. You'll have a cookie or url token to create a unique session with the server. But that scheme can't be used here because the browser is downloading the file, not the server.

        --
        SN won't survive on lurkers alone. Write comments.
        • (Score: 2) by Aiwendil on Monday May 23 2016, @09:10PM

          by Aiwendil (531) on Monday May 23 2016, @09:10PM (#350030) Journal

          Why connect a second time? Just time how long time it took to serve that huge progress bar and send the time near the end of the page - nothing prevent you from generating that huge blob of data in the serverside script (well, if you set a high enough timeout)..

          However, if you really want to connect several times just use frames/iframes and have one frame use the meta http-equiv="refresh", get-parameters can be eausly passes for sessions id..

          And the server is uploading the file - and it tends to be aware of when a transfer is done

          • (Score: 2) by JNCF on Monday May 23 2016, @09:29PM

            by JNCF (4317) on Monday May 23 2016, @09:29PM (#350033) Journal

            I wasn't aware of http-equiv="refresh", but putting that on an iframe should make the whole thing doable. Good one.

          • (Score: 2) by tibman on Monday May 23 2016, @11:08PM

            by tibman (134) Subscriber Badge on Monday May 23 2016, @11:08PM (#350058)

            Yes, that is very doable. The iframe would give you the finished time (from the server's perspective). I should point out though that you aren't connecting two times. You create a new request every time the iframe refreshes (expensive!).

            --
            SN won't survive on lurkers alone. Write comments.
            • (Score: 2) by JNCF on Tuesday May 24 2016, @12:04AM

              by JNCF (4317) on Tuesday May 24 2016, @12:04AM (#350075) Journal

              Another thought: you could have an iframe with a source URL that the server just doesn't immediately respond to. It waits until it knows how long the test download took, and then it returns a full response. A piece of the page takes a moment to load. Not so expensive.

    • (Score: 2) by NotSanguine on Monday May 23 2016, @04:31PM

      Fortunately [speedof.me], you don't [openspeedtest.com] [adblock detector] have [html5speedtest.me] to imagine [dslreports.com] it [testmy.net].

      --
      No, no, you're not thinking; you're just being logical. --Niels Bohr
      • (Score: 3, Informative) by Scruffy Beard 2 on Monday May 23 2016, @04:47PM

        by Scruffy Beard 2 (6030) on Monday May 23 2016, @04:47PM (#349949)

        First one uses JavaScript (HTML5 is the dogwhistle) No JAVA, which is compiled bytecode running in a VM.

        Same for the second one.

        And the third,

        The fourth is also "pure HTML5".

        The fifth also appears to require Javascript.

        TL;DR: To test internet speed in the browser without Javascript, you apparently use Flash or Java bytecode.

        • (Score: 2) by NotSanguine on Monday May 23 2016, @07:45PM

          First one uses JavaScript (HTML5 is the dogwhistle) No JAVA, which is compiled bytecode running in a VM.

          Same for the second one.

          And the third,

          The fourth is also "pure HTML5".

          The fifth also appears to require Javascript.

          TL;DR: To test internet speed in the browser without Javascript, you apparently use Flash or Java bytecode.

          Yes. After I posted this comment, I realized that I had mentally replaced "javascript" with "Flash" and was responding (wrongly) to that instead of OP's original statement.

          That's what I get for posting before I've had my coffee. Sigh.

          Thank you for the correction and the additional information about the links I posted.

          --
          No, no, you're not thinking; you're just being logical. --Niels Bohr
      • (Score: 3, Informative) by maxwell demon on Monday May 23 2016, @05:37PM

        by maxwell demon (1608) on Monday May 23 2016, @05:37PM (#349957) Journal

        OK, I opened those pages with JavaScript disabled.

        speedof.me: Opening the page gets me a very explicit "JAVASCRIPT MUST BE ENABLED!" in red letters.

        openspeedtest.com: Doesn't do anything useful (probably because JavaScript is disabled). Clicking on "How it works" has as first section title in large letters: "JavaScript and HTML5 technology"

        html5speedtest.me: Only a short text, no functionality (probably because JavaScript is disabled). Clicking on the "About" link brings me to a text containing: "This speed test is of the JavaScript variant but adds to it some niceties from the HTML5 chest of goodies."

        dslreports.com: While there's no message and it apparently manages to generate some processor load (though less than 100%), after ten minutes, there's still no result available (according to the page, the result should be available in less than 45 seconds). A cross check with JavaScript enabled gives me 100% processor load, and then the message "JavaScript: This device may be running too slowly". Which confirms that the test indeed uses JavaScript.

        testmy.net: I get a big red box containing red text: "Javascript Disabled
        For full functionality re-enable javascript or add [*.]testmy.net and tmnstatic.com to your browsers javascript exception list."
        Clicking on the "Test my Internet" button does nothing.

        Summary: None of the pages work without JavaScript.

        --
        The Tao of math: The numbers you can count are not the real numbers.
    • (Score: 2) by tempest on Monday May 23 2016, @06:35PM

      by tempest (3050) on Monday May 23 2016, @06:35PM (#349976)

      10Gb Jpeg.

      It'll feel like the 90s all over again as you watch it load.

      • (Score: 2) by JNCF on Monday May 23 2016, @09:03PM

        by JNCF (4317) on Monday May 23 2016, @09:03PM (#350026) Journal

        I was thinking along these lines, but I don't see how it would work with a reasonable interface that doesn't involve a stopwatch. If we could defer the loading of a second image until the gigantic hidden one was done, the second image could be served up differently to each user. But I believe we can only use the defer attribute on external script tags, which is a shame; it would be great for images. Still thinking.

  • (Score: 2, Interesting) by Anonymous Coward on Monday May 23 2016, @01:28PM

    by Anonymous Coward on Monday May 23 2016, @01:28PM (#349895)

    "Bursting" is what most ISPs apply to your TCP connections -- It's throttling on a port by port basis. When a new port is allocated by your router (you know, because IPV6 still didn't give us individual IPs for all our devices and end the NAT insanity), the new connection is in "burst" mode, which allows packets to flow at whatever max speed they've set you at. Then after a little while, usually N bytes downloaded, but sometimes time or packet count based, your connection is throttled.

    DownThemAll is a download accelerator for Firefox. What DTA does (and few other download accelerators do) is that it opens multiple connections to the same resource, using the HTTP resume supported features to start fetching data from byte 0, 1/4 the size, 1/2 and 3/4 (default 4 chunks at once). This is nice because the throttle is per port and will allow you to use S times more bandwidth (where S is the number of splits). However, even this sucks when compared to pausing and resuming a download.

    While a file is downloading, note the initial high speed for "burst" mode, then a little while later the artificial throttle will engage and you'll see the speed drop. Most download status bars, esp. browser builtins, have a large smoothing factor so it's difficult to see exactly when the speed falls off and throttling begins. If you hover your cursor over the DownThemAll progress it will display a real time speed graph and you can see precisely when the speed falls off. After your connection is throttled pause the download. Then hit resume. This allocates new ports during the rebuilding of the TCP streams and thus puts your download back into "burst" mode so you can continue to D/L at the highest speed. When combined with a download accelerator like DTA you can attain speeds far in excess of those that stupid ass speed test websites display.

    My game's networking engine has an experimental feature that constantly detects when a connection is throttled (burst mode wore off) and automatically opens a new TCP and/or UDP socket then seamlessly rolls communication over to the next port. This can utilize a lot more of your server bandwidth, but for P2P transfers it's typically a phenomenal speed boost. This means that web based download tests won't tell you fuck all about the speed you'll get downloading game updates in game vs externally. I've tested this on mobile apps too... but thanks to bandwidth caps I've erred on the side of caution and avoided adding the feature (you're not stupid but most mobile users are, and they'll blame me for their stupidity).

    I have yet to see the automatic connection teardown and rebuild (not just a TCP reset) in modern download accelerators.

    TL;DR: Take a second look at download accelerators. They're not a scam like downloading more RAM. You can actually speed up downloads with this one weird trick that ISPs hate!

    • (Score: 1) by Scruffy Beard 2 on Monday May 23 2016, @04:56PM

      by Scruffy Beard 2 (6030) on Monday May 23 2016, @04:56PM (#349950)

      You are not Developer of a space combat game are you? (I forget the name)

      I have had that games' downloader pull down the Internet, all while the router says not too many connections are in use. Problem went away after throttling (but could not set upload/download separately).

      My brother refused to submit a bug report "because problems are expected with alpha software"

      • (Score: 1) by Scruffy Beard 2 on Monday May 23 2016, @05:05PM

        by Scruffy Beard 2 (6030) on Monday May 23 2016, @05:05PM (#349951)

        I forgot to mention: you also repeat Microsoft's assumption that wired connections are not capped.

        In 2011 Bell Canada got approval to charge Internet resellers extra if an individual customer exceed 300GB in a month. Even though that decision has been reversed, bandwidth caps in the region of 300GB/month are common in Canada.

      • (Score: 0) by Anonymous Coward on Monday May 23 2016, @05:06PM

        by Anonymous Coward on Monday May 23 2016, @05:06PM (#349952)

        Unfortunately, no. I'm not that dev. Please do submit bug reports though. Indies are often deficient in the testing department.

      • (Score: 0) by Anonymous Coward on Monday May 23 2016, @05:46PM

        by Anonymous Coward on Monday May 23 2016, @05:46PM (#349961)

        I remember a game download program for free space SCP. There was something like 50 files or more it had to download. The fun part was that it would open a connection for each in a separate thread and try to download all at once. My router at the time would slow to a crawl trying to handle more than 64 connections, so that was a fun few days.

    • (Score: 0) by Anonymous Coward on Monday May 23 2016, @06:28PM

      by Anonymous Coward on Monday May 23 2016, @06:28PM (#349974)

      Form most end users you get better bang for your buck with https://en.wikipedia.org/wiki/HTTP_pipelining [wikipedia.org]

      Most browsers have a setting for it. Not sure what effect it will have on http2.0.

      you know, because IPV6 still didn't give us individual IPs for all our devices and end the NAT insanity

      Who is your ISP? Mine gives me a full /64 IPV6 address range. Now all 15 of my devices have enough ips to count the atoms in the universe. Also be careful which router you use. There are many out there that are still semi shaky on IPV6 implementation. I have had good luck with ASUS.

      While a file is downloading, note the initial high speed for "burst" mode, then a little while later the artificial throttle will engage and you'll see the speed drop.
      My ISP used to do this. Now they just max it out the whole way. They over-provisioned everyone instead of playing 'turbo' games.

      Now in the mobile area (which I suspect you are developing for) that is a whole different ball game. No one does IPV6 very well usually feeding you out a /128 if they support it at all. Also you may be running up against the local limiter on the phone itself. They are designed to limit themselves so they do not zorch thru the battery in 2 seconds. So the phone manufactures play any trick they can to get the phone to have a battery life of more than 2 hours. They do that by keeping an eye on the power usage and heat from the battery/phone. Also if the carrier thinks you are in *any* way abusing their network they just will not setup the data call on 4g/3g at all and welcome back to CDMA/EDGE. They can bash it down at the AAA RADIUS layer. Also if you are doing this on a cell phone you are kinda a dick. As you are basically causing your phone to use up more time division data slots and causing everyone around you to get less. The system is trying to balance the amount of data needed across everyone. If there is no one else on the tower you probably will get the full towers attention. But put 300 people on it and you dont get as much attention.

  • (Score: 3, Informative) by Zinho on Monday May 23 2016, @02:32PM

    by Zinho (759) on Monday May 23 2016, @02:32PM (#349915)

    I recommend DSLReports [dslreports.com] as a speed testing service; they have given me decent results over the years. I'm not certain what technology they're using for the test now, seems to be javascript; in the past they were using a Java applet, but that seems to have been deprecated.

    Analysis and feedback from the Soylent crowd is appreciated; I'd love to hear whether or not my trust in this group has been misplaced.

    --
    "Space Exploration is not endless circles in low earth orbit." -Buzz Aldrin
  • (Score: 1) by biffulator on Monday May 23 2016, @02:36PM

    by biffulator (5417) on Monday May 23 2016, @02:36PM (#349917)

    Fast.com tops out at 150mbit on my fiber connection. DSLreports is better.

  • (Score: 2, Disagree) by zeigerpuppy on Monday May 23 2016, @04:02PM

    by zeigerpuppy (1298) on Monday May 23 2016, @04:02PM (#349936)

    Https://beta.speedtest.net doesn't need javascript

    • (Score: 3, Informative) by isostatic on Monday May 23 2016, @05:12PM

      by isostatic (365) Subscriber Badge on Monday May 23 2016, @05:12PM (#349954) Journal

      Https://beta.speedtest.net doesn't need javascript

      It really really does.

      It doesn't require flash.

      Javascript is far better than flash. If I'm after a specific bandwidth check I'll break out iperf into one of my servers, but for an idea speedtest.net is great.

  • (Score: 2) by opinionated_science on Tuesday May 24 2016, @02:12AM

    by opinionated_science (4031) on Tuesday May 24 2016, @02:12AM (#350109)

    I have a python script that pings speedtest.net. Since it is a user script , you can run it as a nobody, and so no risks...

      git clone https://github.com/sivel/speedtest-cli.git [github.com]
            python speedtest-cli/setup.py install

    See how you fare ;-)

    • (Score: 0) by Anonymous Coward on Tuesday May 24 2016, @11:25AM

      by Anonymous Coward on Tuesday May 24 2016, @11:25AM (#350246)

      Just a sec while I bypass my innate compulsion to not run scripts from untrusted sources, even if they're not running as root.

      Ah, done. Just created a new separate user to run such shite as. My privilege escalation lobe is still making my eye twitch though.