please explain "my rt" ("my response time")

[bratteemailtest]

when you click on results, the rightmost button under the graph is labelled as "My RT" (only appears when signed in as a user). when you click on it, it provides a history of "response times". can someone please explain what exactly is being tested and reported as "response time"? i would venture a guess that it is measuring some type of transaction request. but what exactly constitutes a "transaction" and what are the delimiting factors that define the start and end times?

[CA3LE]

It's measuring the time it takes for you to respond to a request. 

Your browser sends the request, it hits one of my servers and is immediately returned back to you.  I then cut that time in half to represent half of the round trip... and call it Response Time.  Generally it should be close to what you can expect to see when you run a ping but don't confuse it with ping, I've found that there are many more things that can influence the Response Time Test.

I'm still trying to make sense of it myself... it will without a doubt need much more development.  I've been reluctant to release it because I don't fully understand everything but I figure the best way to understand it is to get feedback from my users.

Sorry to my Beta Testers for not releasing this to you first... but there really aren't that many active beta testers and I need a wide range of input from many people.

Please help and provide feedback, good or bad. 

[bratteemailtest]

excellent "response time" to the question and thanks very much! a very good answer to the question. i'm not sure that i would cut the time in half however, since the upstream and downstream can be rather different, though i don't know how you would measure the difference. and yes, i believe that many other things can affect your response time test. packet and/or transaction size, upstream vs. downstream pipe, data manipulation (if any), etc. all things considered, i think this is an excellent addition to the status report. good luck with the further development.

p.s.: many thanks for this web site in general. it is indeed one of the best test sites (if not the best) available. i have found it's accuracy and consistency to be far above the others and i greatly appreciate your efforts.

[CA3LE]

I like your train of thought but I think this test is different from what you're thinking.  It's not a bandwidth test and has less to do with your upstream and downstream connection speed and more to do with its quality.  You're not really transferring data, put it this way... minus headers you're only transferring a little over 100 bytes of data each time that test runs.  What the results tell you is how quickly you're able to respond to requests.   Quicker is better, RT results are shown in milliseconds so we want a low value here.

Testing from the same location to the same locale you can have vastly different results depending on the device used and connection quality.  For instance...

iPhone 6, Verizon Wireless - ~140 ms (in some areas less, in some areas it can be 100's of ms higher)
iPhone 6, Comcast cable - ~30ms

Desktop on Comcast (same connection above) - ~20ms
Laptop wifi - 30-60 ms

As I get into a worse connection area or further from my router (or my providers towers when I'm on LTE) the time goes up.  If the connection has to jump through more hoops to get to you (e.g. more hops, lower quality hops or hops/servers that are further away)... it will slow down the Response Time.

I've found that these numbers always directly correlate with performance.  One that really sold me (it made me want to release this) was recently when I did a modem swap.  When I put the (SB6183) 16 channel modem on my RT dropped by about 4ms... it was consistent and to me an obvious change.   When I swapped back to the 6141 (which I actually prefer overall) the RT immediately jumped back up to exactly what it was before.  (I chose the slower responding modem because it ramped up faster... if I was more of a gamer I may have gone the other way for the better RT, which would reflect a better ping.)

Here's what you don't want to see >> https://testmy.net/rt/tktnsma1 or https://testmy.net/rt/markjfine or... anyone else on any satellite connection (those examples are Hughes Net customers -- they have such poor RT because their connection has to travel into space and back.  Light is fast... but it's not instant.)

Again, this isn't ping.  I've seen where ping is unaffected but RT is obviously not what it should be.  ... it always reflects performance.  The times when I've seen that my browser is responding slowly, which is also reflected in speed test results.  But then I do a normal ping and it doesn't show me anything.  (e.g. leading me in some cases to find that one of my million open browser tabs was bugged out, causing the browser to respond much slower.  Close and re-open, RT returns to where it should be.)   So, this test is fundamentally different from a ping test.  It's similar, if the RT if affected the ping is also often affected but sometimes not.  Just keep in mind, a discrepancy is most often a clue to an underlying issue.

[CA3LE]

p.s.: many thanks for this web site in general. it is indeed one of the best test sites (if not the best) available. i have found it's accuracy and consistency to be far above the others and i greatly appreciate your efforts.

You're welcome, it's my pleasure.  I appreciate the kind words.  Looks like I'll have to work harder to help you understand that this already IS the best testing site... not just one of them.  Where do you think the other speed test sites got the idea from?  

[CA3LE]

Watch how it correlates in this example.

RT goes up...

Speed goes down...

Thanks for the results richardz!