In the process of pressure test, found that ab - 1000 - c n 1000 http://xxxxxxxxxxxxxxx several times before the data are good, the average processing time is within 3 s, but with the increase of test times, the time may also be increased to 50 s, 40 s,,,
why difference so big?
Netstat - an | grep ESTABLISHED | wc -l
Ps - ef | grep HTTPD | wc -l

I observed the above command useful concurrent and processes, such as each test are also down this number later for the next test,


Ever give me answers,,, why, as the test number increase, was so large, stop for a long time later, can get satisfactory data, if so how do I test can get relatively accurate and valuable data,,,,