Jump to content

ISP not stacking up? Just blame the speed test.


CA3LE

Recommended Posts

I am disappointed that you make bold claims about being the most accurate speed test site (like all sites don't claim this too) yet are measuring far below my connections peak speed.

:::.. Internet Speed Test Result Details ..:::

Download Connection Speed:: 34160 Kbps or 34.2 Mbps

Download Speed Test Size:: 32.5 MB or 33280 kB or 34078720 bytes

Download Binary File Transfer Speed:: 4270 kB/s or 4.3 MB/s

Upload Connection Speed:: 30632 Kbps or 30.6 Mbps

Upload Speed Test Size:: 14.6 MB or 14912 kB or 15269888 bytes

Upload Binary File Transfer Speed:: 3829 kB/s

Timed:: Download: 7.981 seconds | Upload: 3.988 seconds or 3.8 MB/s

Tested At:: http://TestMy.net Version 13

Test Time:: 2013-06-01 18:28:38 Local Time

Location:: Sheffield, GB GB >> Destination:: Amsterdam, NL / Europe

Validation:: https://testmy.net/db/lvkBnED.jliaTSJ

TiP Measurement Summary (Download):: Min 23.34 Mbps | Middle Avg 35.16 Mbps | Max 47.75 Mbps | 41% Variance

TiP Data Points:: 23.34 Mbps, 34.33 Mbps, 43.77 Mbps, 47.75 Mbps, 36.42 Mbps, 27.93 Mbps, 28.09 Mbps, 33.69 Mbps, 42.8 Mbps, 40.52 Mbps, 33.11 Mbps, 34.71 Mbps, 40.99 Mbps, 32.94 Mbps, 23.61 Mbps, 24.01 Mbps, 32.35 Mbps, 40.72 Mbps, 26.97 Mbps

More Stats:: https://testmy.net/compID/133205285315

User Agent:: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/27.0.1453.93 Safari/537.36 [!]

My connection shows a far more accurate results on speedtest.net, 89Mbit down and 30Mbit up. Bearing in mind my real-world results are usually slightly higher than that.

Perhaps its a lack of enough regional servers, but even the third-party London server is coming up under 70Mbit.

 

Hi Alex,

 

Distance will not be an issue if your provider is delivering.  GB to NL isn't that far, if your provider and the people they peer with were delivering... you'd be able to easily pull your speeds at that distance.  I have connections that I pull hundreds of Mbps from at distances of thousands of miles.

 

TestMy.net uncovers issues that other speed tests fail to detect.  It could be your TCP settings effecting performance.  It could be poor routing or it could be that your ISP is pulling a fast one.. shaping the way you can draw your bandwidth.  LOTS of people are contacting me telling me that this is happening to them. TMN is purposely designed to render as normal content would which makes it in tune with many issues that go undetected otherwise.

 

Before you're so quick to judge you might want to read this post one of my members wrote, just last night.  https://testmy.net/ipb/topic/30731-multithread/?p=339762 -- He's an aerospace engineer with a lot of experience with bandwidth.  His independent findings as well as others backup my claims.  That's coming from someone who's pulling over 600 Mbps frequently using the same exact test you used... it's not my test, it's not my bandwidth.  Other speed tests are wrong more often than not.

 

Believe the numbers you want to believe but before you rush to judgement you might want to put a meter on your line and compare.  TestMy.net is right on the money every time. Other tests myself and others compare to don't line up and obviously aren't taking the entire test into consideration.  They also often don't use enough test information to be accurate so results can be inflated from bursting.

 

Another way you can achieve higher speeds is by multithreading.  If you go to https://testmy.net/mirror you'll see information on it and a link to toggle it on and off.  Currently however I haven't expanded this to my servers outside of the US.  The way it's setup now you'll be pulling resources from all my servers in the US at the same time.  By opening more threads you may be able to pull more speed, even given the distance.  The resulting score will represent your speed not just to one location... but across the entire US.  Multithreading is not my default option because enabling it can mask issues that only testing a single thread will pickup on.  Again, this is what makes my site different.  Other sites are simply trying to max out your connection.  There is more to it than that.  The truth is, if your connection is running correctly and your ISP isn't shaping your bandwidth... you should be able to max a single thread.   ... personally, I have the same issue on my home connection as you might.  I can't come close to my speeds in a single thread, multithread opens me up an extra 30Mbps. Using the multithread speed test on my connection more accurately depicts the capability but the question remains, why can I only do that when I open more threads?  If there was an issue with the test or servers myself (on other connections I have) and other users wouldn't be able to pull many 100's of Mbps.

 

ISPs are starting to heavily shape the way you can draw your bandwidth.  Often limiting single thread transactions.  Other speed tests fail to pick up on this... maybe they design it that way on purpose because that's who the ISPs use.  ... Ookla brand speed tests in particular are BS.  They ignore part of the transfer and omit a portion of the data, the result is an inflated score. Read more - That information comes from their OWN admission printed in their own wiki.  Dude, you have to admit to me... if the results are adjusted, it nullifies the result.  You can't trust something that makes adjustments for a number that's variable.  ... Say that your connection dips for part of the test.  Chances are, it won't be accounted for in the final result.  You're testing to know that information and they're hiding it from you.  TMN accounts for EVERYTHING that happens during the test.

 

If you look at the TiP details on my results you'll see how much simply looking at different parts of the data can effect the result.  Look at 'middle average', that's removing the first and last reading... Not only does ookla/speedtest.net ignore the start and finish, they remove the worst portion of your results before calculation. Flash speed tests have been proven inaccurate by MANY people much smarter than I am.  I've also been told that ISPs that opt in can actually make adjustments to how much of the results are dropped off.  HAHA, sounds like fraud to me.  Letting ISPs basically make their own results.  BAH!  How they sleep at night escapes me.

 

Keep this in mind too... I was building my test nearly 10 years before ookla released their first test.  Believe who you want to... my program is just calling it how it sees it.  Other people in GB are able to pull well over 100 Mbps... even out to my US servers --- https://testmy.net/country/gb/max --- your connection just isn't delivering like you thought it was.  The truth hurts sometimes but when you searched for a speed test I don't think you were looking for someone to pull wool over your eyes... you were looking for a test.  If you don't believe my results, use the other guys test with results designed to make you feel warm and fuzzy inside.

 

Thanks for visiting, I hope that I've helped shed some light and that you'll trust and continue to use my service.  If I opened your eyes to anything, please help me spread the word.  I'm trying my hardest to get this information to people but I'm only one person.

 

- Cheers and Kind Regards,

- Damon Mueller

Link to comment
Share on other sites

I've been testing on the Emory University Campus in Atlanta Ga and getting results in the 5-10 Megabytes/sec range. Our Campus links are rated are much faster. I'm wondering if there is a limit at your servers due to either your internet links or web service.

How high should I expect it go to if there were totally unrestricted bandwidth to the internet on our end?

 

Hi Ken,

 
I have users that regularly pull over 600 Mbps (>70 MB/s). My servers are ALL very well connected with at least 1000 Mbps symmetrical.  My main server has dual symetrical 1000 Mbps connections connected into Softlayer's multi-homed network >> http://www.softlayer.com/about/network/carriers. I'm in all of their datacenters. I've got a lot of bandwidth to back up my application, before anything... I make sure that my bandwidth is premium for my users. It's a bandwdith testing application, to be real.. it has to be on the best connected networks.  Research my host and you'll find that I'm very well connected and peered.   ... My main server is also very powerful with SSD RAID5 and 24 cores. Overpowering my end means that when you test you're guaranteed a private core, memory and plenty of bandwidth on tap. I'm running at a small fraction of my servers potential... and then I have 5 other servers on top of that to choose from.  All dedicated to the task of speed testing. https://testmy.net/mirror  -- 10 more mirrors hosted by other people too (new project, offering it up to anyone with a web server, you should host one at your school if you can). https://testmy.net/make-a-speed-test
 
I suggest testing to some of the mirrors to see if there may be a routing issue.  If you get limited speeds similar on most servers you might want to look into the problem locally. Either on the client machine or network. Often tweaking TCP settings can improve connections that fast. For Windows, which is the worst culprit, search for TCP optimizer from speedguide.net (I think it's) https://speedguide.net/downloads.php. For other OS search google for gigabit TCP tweaks for your specific OS. Note that TestMy.net is sensitive to many issues that other speed tests don't detect. It's purposely designed to be a hard test to ace... try to score as high as you can, then compare your subsequent results to your own average, highs and lows.  Don't expect to pull your full speed on every connection type you test with... especially gigabit and faster.  I can't... but the tools still makes the health of your connection very obvious in my opinion.
 
TestMy.net is more than just an Internet speed test. I've seen first hand where it scored my girlfriends macbook under 20 Mbps on a 50 Mbps connection... I swapped her hard drive for an SSD and upon reboot her TestMy.net score maxed out my connection. I purposely tested around to prove a point. No other speed test picked up on it... even though my download and Internet surfing performance were obviously degraded. I wrote every line of code and I don't even know what truly makes my test different, it just is. Some known differences between how I do it and how my leading competitors do it are touched on at https://testmy.net/ipb/topic/28902-why-do-my-results-differ-from-speedtestnet-ookla-speed-tests/.
 
If you like my life's work, please share it... it's totally word-of-mouth.
- Damon
Link to comment
Share on other sites

×
×
  • Create New...