Jump to content

Incorrect speed test?

Recommended Posts

The first two tests (Ookla and nPerf) run their tests multi-threaded by default, i.e. they typically make around 8 simultaneous connections to the test server to try to saturate the connection.  Ookla has a lot of test servers within the ISP networks, so there is a good chance your test traffic is not leaving their network unless you manually choose another server. 


To get a multithread test here on TestMy, click the "Multithread off" at the top-right to turn it on.  Note that multithreaded tests don't realistically show what you would get streaming or downloading, which generally run over a single connection from the server.


Google's built-in speed test is a single connection speed test (not multithreaded), but uses the new TCP BBR congestion protocol.  The TCP BBR congestion protocol is excellent at handling packet loss up to about 5% before it suffers significant speed loss.  As web servers don't have BBR enabled (or installed) by default, most websites use the legacy TCP CUBIC congestion protocol.  For this reason, TestMy does not use TCP BBR, apart from the Colorado Springs server at this time.  A few major web hosts such as Google and Microsoft and CDNs such as Cloudflare are TCP BBR enabled. 


As Google's speed test uses Measurement Lab's servers, you can see how your speed compares with the widely used TCP CUBIC congestion protocol that Measurement Lab uses on its own website test: 




If the "Retransmission" figure is not zero, there is a some packet loss on your connection.  If the speed is close to the 80Mbps what you got on Google, then your ISP may have congestion on its peering with international links as TestMy does not have any test servers in Italy. 

Link to post
Share on other sites

Also, the latency times of roughly 10mSec (Rome), and 30 mSec (Milan), indicate that the route of those tests results of 80+ Mbps weren’t done over a very long distance or complex network like you’re typically getting when you test using TestMy. 

I’m some 700 miles from the NY server. Milan and Rome are about 300 miles apart. I get about 60 mSec latency to NY. 

It sounds from the 9mS delay that you’re close to Rome.  Your 30 mSec lag is roughly 1/2 of what I typically get and you’re roughly 1/2 the distance from Milan than I am from New York. Your real internet performance will be worse than the 80+ Mbps as for most of your online time, you’ll be connected over much greater distances and much more complex data paths.


The test methods you’re using are telling you what the maximum sustained performance you can ever expect over short simple data paths. TestMy is giving you a much more realistic measure of what your internet experience will be.

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Create New...