Slashdot Mirror
Google Is Testing Its Own Internet Speed Test In Search Results (thenextweb.com)
An anonymous reader writes: Everyone appears to have a speed test of their own nowadays. Netflix launched fast.com more than a month ago; SourceForge released their new HTML5 speed test soon after. Google appears to want a piece of the action as they are trying out a way for people to check their internet speed by simply typing "check internet speed" into search. The tests are performed by Google's Measurement Lab tools, and were first spotted by Pete Meyers, who posted a screenshot of the feature and discovered a Google Support webpage detailing how it works. The feature has not been widely released yet, but it's possible we'll see it made more widely available soon.
internet is fast enough for me!
Must be a regional thing. Not working in Spain.
Countdown until legal action from the speed testing sites that they're stealing traffic and users from. They already had this problem with maps and shopping, they just never learn.
Net Neutrality is the kryptonite of libertarians.
Vive le Linux libre!
I doubt legal action.
Mostly it's major ISPs that offer speed testing, e.g. Comcast.
The trick is that they offer the test from a customer node within their network, to a test server node also within their network, which avoids crossing one or more peering points.
That actually only gives you "last mile" speeds, which don't represent real world expected performance.
A lawsuit would make this information very "in your face" for the general public, and stir up the whole NetFlix/peering controversy again, and that has to be the absolutely last thing a broadband provider wants to see happen.
You could always install your own speedtest server and test your own speed, but of course you wouldn't be able to convince trendy idiots of the validity of your results unless you own a name brand speedtest app, and bragging to trendy idiots about your speed is the only real reason to do a speedtest, now isn't it?
It's not just because of the Sourceforge servers being unreliable, it also depends on what browser you use. With the latest Palemoon, I had 180ms latency according to SF's test, and 50Mbit/s download 60Mbit/s upload, while with Vivaldi I had 108ms latency and 79Mbit/s download, 120Mbit/s upload. With Opera, I got 120ms latency, 85Mbit/s download and 25Mbit/s upload. So, it's totally worthless as a test.
Oh, I should add: Moving my cursor during the latency test added to the latency jitter...
Well, if you don't like Net Neutrality, you're free to set up your own Internet.
Yeah, but when the variability is so high against the same IP, just by switching browsers, and the test performed before the morning surge starts in the US, shows that it's highly flawed. And as I mentioned, moving my cursor around caused latency jitter.
Doing a quick test of speed is fine, but what about ongoing records, automatically recorded? It's a very common story for actual internet speeds not to match advertised speeds (be that truth or exaggeration). While a single speed test might reveal underperformance, charting performance over time would be far more revealing.
Indeed. Netflix is saying 55Mbps, while Sourceforge is claiming 37Mbps down, 87Mbps up (at this time of the evening, it isn't unusual to have to share some of the download bandwidth with my neighbours, but in my experience what Netflix is claiming should be about the right ballpark for downloads compared with other tests). Since I'm in a developing country, Google isn't supporting my location yet.
So the other day, my internet was very slow, I did a speed test on speedtest.net, and it reported about 25Mpbs, which is exactly what my internet is supposed to be.
But it didn't seem right, because every site was painfully slow. So I went to speedof.me and did a speed test, and I was getting like 1.5Mbps, which was definitely more accurate.
The other day I overheard a Comcast commercial where they mentioned that they're number 1 on speedtest.net. So, I came to the very obvious conclusion that Comcast deliberately makes sure to un-throttle your internet when doing a speedtest on speedtest.net (and who knows where else).
The Internet King? I wonder if he could provide faster nudity.
So they make the search term you would naturally type in to find their competitors be the cue to run their own service. Nice job if you can get it.
Some drink at the fountain of knowledge. Others just gargle.
I can understand ISPs wanting to have their own speed test available for their customers within their network, but what benefit does making this available offer for other companies?
What is the ROI for SourceForge having its own speed test?
load "linux",8,1
I see this more as a way to test their servers more so than our own connections... well, at least when testing from Gigabit FTTH
CenturyLink: 535 / 727mbps (the fastest I've ever gotten from my ISP's test server. Usually in the 200mbps range)
Comcast: 470 / 819mbps
Speedtest.net Sprint Seattle: 657 / 751mbps
SourceForge: 282 / 133mbps (usual speeds when testing)
Netflix Fast.com: 44mbps (the fastest I've ever gotten, I usually get around 10mbps from them)
MeasurementLab.net: 71 / 67mbps
SpeakEasy.net: 500 / 896 mbps
AT&T: 325 / 889 mbps
Google: I cant get their speedtest to show up
The question becomes... how saturated are the speed test servers themselves? And then how saturated are the links in each direction between the client and servers? Since download is generally significantly faster than upload with cable/dsl, the reverse is easily seen with fiber. The links are only saturated in one direction from the speed test servers, so pushing content to them has that extra headroom to really push the speed limits.
Also as a side note for those curious, on the tests that report latency, they're all in ~4ms range. The lowest I've ever seen is 2ms with this connection.