Slower results with API?

Hi,

I’m using your API to gather test results but when I test my site without the API it’s always much faster.

Here’s an example without the API http://www.webpagetest.org/result/160226_FK_A6Q/

And with

http://www.webpagetest.org/results.php?test=160225_HD_AK7

I’ve done this test about 20 times and results are pretty much the same.

Am I wrong to expect the same results when using the API? If not what should I be looking for to try and sort this out.

Thanks

They API and WebUI use same test agents so they should be similar.

What happens if you do a nine test run via each method and compare the medians?

And more specifically, the Web UI actually uses the API. Looking quickly it looks like the test parameters were the same (same location, browser and connectivity) so it looks like some level of variation in the load times and as Andy pointed out, doing more runs should give you a good feel for the consistency.

The slower test looked like the CPU was pegged for a longer period but given that the request counts and bytes were the same I don’t think it was from differences in the content.

Hi i am from Leehuges i am developer of him. i found one things ridicules.

I did 4 test form that i did 2 test same time or difference of few second or 1 minutes.

look at first one just http://www.webpagetest.org/result/160226_3S_RRB/ with url of “sitegroundspeedtest.com/

then second http://www.webpagetest.org/result/160226_0Y_RR4/ with url of “sitegroundspeedtest.com

You can see both test have same time just url have little differnce of last “/”?

Before 30 min of those tests i made same things again

look here 1st http://www.webpagetest.org/result/160226_C5_QX9/ and 2nd http://www.webpagetest.org/result/160226_V8_QX7/

in those 4 test i made conculation that when i am doing any test after few minutes let say 30 min. then “sitegroundspeedtest.com” and “sitegroundspeedtest.com/” gives different result for firstbyte.

it’s only happens after long interval like 30 min.

once you do it again in few minutes or second you will get right result for both url.

look at here http://www.webpagetest.org/result/160226_37_QYB/ and http://www.webpagetest.org/result/160226_WQ_QYA/

can you explain me why things are happening ?

Hey guys,

So I ran the test 9 times here are the results just using the website, not through my API setup.

Times are for first byte / load time
http://www.webpagetest.org/result/160226_ME_13N0/ 0.166 4.49
http://www.webpagetest.org/result/160226_D1_13P9/ 0.153 4.83
http://www.webpagetest.org/result/160226_WF_13QJ/ 0.158 3.54
http://www.webpagetest.org/result/160226_BP_13R9/ 0.150 4.8
http://www.webpagetest.org/result/160226_TJ_13SM/ 0.158 4.36
http://www.webpagetest.org/result/160226_CX_13TQ/ 0.153 3.592
http://www.webpagetest.org/result/160226_GS_13VG/ 0.165 4.74
http://www.webpagetest.org/result/160226_XR_13W6/ 0.162 4.59
http://www.webpagetest.org/result/160226_WZ_13WS/ 0.156 3.95

Average first byte = 0.141 seconds
Average load time = 4.32 seconds

I actually have log files of the tests using the API. Here is the first byte wphostinghub.net and the load time wphostinghub.net

Average first byte for API setup = 0.48 s
Average load time for API setup = 6.39 s

It’s clear that for whatever reason, the API setup always brings back slower results even though they have the same settings.

I could run these tests for several days and I’m sure I’d have similar results.

Any ideas where to start looking into this?

Thanks

Is it possible wphostinghub is calculating the times incorrectly?

How do you mean?

Sorry I just noticed wphostinghub links back to the WPT results which still shows slower times.

If the CPU was pegged and this is could be a reason for the slower speed is there a way to stop this?

Cheers