Difference Between Speed Test Sites?

[Jay Wasack]

So I guess my point is getting good results doesn't necessarily reflect actual performance. Is that fair to say?

[mudmanc4]

20 hours ago, Jay Wasack said:

So I guess my point is getting good results doesn't necessarily reflect actual performance. Is that fair to say?

Actually it's the inverse.

The result you get at testmy.net will correlate along side the performance you will see when accessing data in the same area as the test is run.

[Jay Wasack]

I agree with you. But I can't be sure I'll be accessing data on the same area as the test. I'm not questioning the validity of the test just confirming that the test doesn't necessary reflect what might actually be happening in day-to-day usage since I can't predict what route the service is on.

 

Thanks for your patient responses.

[mudmanc4]

On 4/19/2016 at 8:15 AM, Jay Wasack said:

I agree with you. But I can't be sure I'll be accessing data on the same area as the test. I'm not questioning the validity of the test just confirming that the test doesn't necessary reflect what might actually be happening in day-to-day usage since I can't predict what route the service is on.

 

Thanks for your patient responses.

Considering so many sites are pulling from even countless resources, your point is well taken.

Day to day or minute by minute would be something a user gets a feel for after a time, for instance we get used to seeing a pageload depending n the site, at a certain speed, when there is an issue with one or more of the resources the site relies on, we notice that. This is where testmy.net can assist, by sector, what might be the issue. For instance if test results are all coming out low, this would lead us closer to home, or the ISP, if tests on servers one side of the country are low but not the other, this would lead us to a peering issue. Same for an overseas test.

[Jay Wasack]

Just did a test. 162.3 up 1.4 down - really?

did speedtest: 342 up 16.5 down

 

i dont intend to believe what makes me happy. but the disparity is incredible. 

 

any suggestons?

[Jay Wasack]

To further this: how do you deal with Comcast support when they want you to use speedtest? Again, in my case, both testing methods are in the ballpark for downloads but upload on speedtest are 10x higher than tmn. any suggestions?

[rrroger555]

with the latest policy changes , has there ben an increase in "overhead data" wcross the board.? im suspecting that dnld could be waiting for upload(affecting speeds for all sites and pages...) same speed but theres more overhead. if ypour upload is 100kbps(like mine) do you think that maube that BW cant handle the up-side thruput and is slowing down the entire speed test? secondly, compressed data will mess up your testing pretty good so dont do it unless you are in the clear. finially the realtime data graph at testmy.net .. if you average all the points then you will get a number similar to other sites(higher) but ith internet does not work on averages, neither does engineering. it works between the peaks and valleys. ive seen tests wjhere the REALTIME data shows 200kbps for 90% of the test and then a spike of 60 or so for 10% of the test(60 mbps is not possible on my connection, not even close) and yhe average of it is around 1 mbps, the average isnt totally irrelevant but once your realtime drops below functionality, the average  IS irrelevant. zero is zero is zero and no amount of averaging will make the zeros npot zeros.

[Pgoodwin1]

6 hours ago, rrroger555 said:

with the latest policy changes , has there ben an increase in "overhead data" wcross the board.? im suspecting that dnld could be waiting for upload(affecting speeds for all sites and pages...) same speed but theres more overhead. if ypour upload is 100kbps(like mine) do you think that maube that BW cant handle the up-side thruput and is slowing down the entire speed test? secondly, compressed data will mess up your testing pretty good so dont do it unless you are in the clear. finially the realtime data graph at testmy.net .. if you average all the points then you will get a number similar to other sites(higher) but ith internet does not work on averages, neither does engineering. it works between the peaks and valleys. ive seen tests wjhere the REALTIME data shows 200kbps for 90% of the test and then a spike of 60 or so for 10% of the test(60 mbps is not possible on my connection, not even close) and yhe average of it is around 1 mbps, the average isnt totally irrelevant but once your realtime drops below functionality, the average  IS irrelevant. zero is zero is zero and no amount of averaging will make the zeros npot zeros.

Good points. Here at TMN you get an average, a max, and a graph of what happened over time. 

[mvbmac]

I am wondering why Netflix's "FAST" speed test web tool would report such inflated results.  I am paying for a coax cable level targeting 105 or more Mbps download.  Actual results on testmy.net have, over time, run anywhere from 100 to 125 Mbps, give or take.  Testmy.net has typically returned the lowest speed reports (in which I have the most faith) of various popular test sites like speedtest, ookla, etc., all of which tended to return results in the 150 Mbps neighborhood.  (All this being download of course.). I can understand why some of these test sites, which are controlled by ISP's, would inflate reported results.  However, when I try Netflix's FAST test, I get unbelievably (literally) high reported results -- wild figures like 300 Mbps or more.  Anybody got an opinion on why Netflix would operate such a positively exaggerated test?  They certainly have no love loss for the ISP's after the edge provider/net neutrality battle?