Middle Variance?

[XANAVirus]

Hi,

 

You know how you do a speed test here on TMN, and then you get sent a results page which contains the TiP summary and the max / average speeds of your connection?

 

Well, further down there is the Middle Variance value, which displays in different colors depending on the value it has (red, blue, green, etc.).

 

What I'd like to know is, what does this value represent and what are good values for it to show?

Is blue color better than green color, where does this facor in to the results of the tests?

 

Thanks!

[CA3LE]

Hi,

 

You know how you do a speed test here on TMN, and then you get sent a results page which contains the TiP summary and the max / average speeds of your connection?

 

Well, further down there is the Middle Variance value, which displays in different colors depending on the value it has (red, blue, green, etc.).

 

What I'd like to know is, what does this value represent and what are good values for it to show?

Is blue color better than green color, where does this facor in to the results of the tests?

 

Thanks!

 

Those colors slide down the RGB scale.  Red being the worse and blue being better.  Although, sometimes little variance is bad... like sometimes if you encounter a bottleneck along your route.  So use your best judgment, don't always go off the color. 

 

e.g. Say you have a 50 Mbps connection but you're scoring only 12 Mbps... BUT the TiP is showing only 5% variance.  It's going to show up blue.  Obviously you know that something isn't right because you're supposed to have 50 Mbps.  What the test saw however is a steady transfer...

 

You can have a steady clean connection that is limited... just make sure to judge your result in its entirety, the color is only a hint to how steady the connection was... not an indication of if it was a good result or not.

 

Here's one in red that I wouldn't consider bad... 67% middle variance

 

... it could have done better, here's a result taken at the same time on a different device. 25% middle variance

 

Is either one of the above bad?  I personally don't think so.  But I think it's obvious which performed better.  Hint: Just like in a drag race, it's usually the one that accelerated the fastest.  As long as there aren't bumps in the road.

 

Funny story, testing at the same time were a pair of identical LG G2's.  One was bogged down with programs running and the other was the better result above. 

 

Here's the result from the bogged down phone.

Now that's a rough ride... something she had running was ruining her connection.  Hint: pattern variance (like this example) is usually a major indication to an underlying issue. 

 

She opened the task manager and closed all programs and retested...

 

... almost mirrored the result of the other phone.  Both beat my i5s

 

Note: In case you didn't know.  TiP is measured from 5% to 95% ... the middle of that (or 10% to 90%) is the middle.  The variance within that is the middle variance.  ... so a large burst or drop off at the beginning or end will not effect the middle variance.

 

I hope this answers your questions and then some about TiP.  It's something I made up and never really explained.  I guess I'm always aiming to build something that doesn't need explaining.  ... I'll keep working at it. 

[XANAVirus]

Great explanation, CA3LE.

 

Plenty of information to help me understand this. You definitely answered my question and then some.

 

Thanks so much!

[gerry5]

Do I understand the explanation and charts correctly to say that when the line starts at a higher Mbps at the very start than the graph that starts low and rises, the graph that starts at the much higher Mbps is the better performer?

[CA3LE]

Do I understand the explanation and charts correctly to say that when the line starts at a higher Mbps at the very start than the graph that starts low and rises, the graph that starts at the much higher Mbps is the better performer?

Yes, out of those examples the 2nd one at 113 Mbps was the best. It almost instantly ramped up and fluctuated little. Even better would be the same fast build up... then a flat top. But as I said in my other reply.. that can also happen if there is a bottleneck limiting your speed. ... only to be considered if you feel like your running slow and have little fluctuation in combination.

[VianetCA]

https://testmy.net/db/83xBhmSi.zEPkLSqD  :  Full Result

 

 

Vianet tells me that everything is 'OK'

True / False / Bad / Good / Ugly ?

[mudmanc4]

https://testmy.net/db/83xBhmSi.zEPkLSqD  :  Full Result

 

 

Vianet tells me that everything is 'OK'

True / False / Bad / Good / Ugly ?

This all depends on what agreement you have made with your ISP, concerning throughput. 

[basjke]

I know this is an old thread, but I experience just that on my PC, a high middle speed variance.

 

I know it is not my router or internet connection, because all other devices and even my phone on Wifi will give me a down speed result double of my PC.

 

I have inserted the HDD from my son's PC, which is the same OS as I have, in my PC and then I also get a high download.

 

Since have I already reinstalled OS several times, tried other OS and other HDD's, but I always get the high middle variance ( +100% by the way)

 

FYI on all my devices connected over LAN or Wifi I'll get ~14Mb to London server, on my PC connected by LAN and very close to the router I get ~7Mb.

 

Right now I have a new disk with new installed OS and almost no programs installed, but still get the low speed, so how do I find out what gives me this high middle variance?

 

When I insert my son's HDD in my PC I'll get 16Mb easily.

[Ben Williams]

If I read what you're saying correctly you swapped HDD and got proper performance. In that case it's almost definitely something to do with the HDD or HDD cabling. Very odd but the only reasonable explanation given the information you've provided.

 

Also: were the test servers identical on each? I ran into an issue where the server I was testing to was showing me 10mbit speeds but when I switched to another server I'm getting 30+.