Recent sector performance test requirements, but also to do performance testing for a single interface, just tell the truth began to test the performance of a single interface to do when feeling very ignorant to force ah, as do performance tests were conducted before a single specific business scenarios mixed performance test scenarios and scenes, so the beginning more confused, do performance testing interface testing purposes where it? Like a long time did not come to an answer I wanted. Well, I finally just give up, take a direct jemter proceeds pressure measurement, and monitoring lunix server resources.
I just started the whole server resources direct monitoring, to different service where the resource allocation of the current interface, then the pressure measurement for the current resource, very happy to finally be able to work smoothly, and on Mao vigorously dry, so two put 10 days at several different memory interfaces where the pressure measurement is complete, and finishing to excel, you look to the following results:
It is not done very carefully ah, but however. . . . . I later discovered, Nima jemeter monitored instead of the entire server's resources well, my server hung below can be more than one service ah ah ah ah ah ah. . . Instant fast crazy people, what do they need to do it all over
Pass behind Baidu know, you can find the corresponding process with the top command lunix, you can monitor the cpu and memory, according to the process of change id, as follows
ps: before the performance test, to exclude the bottleneck of the network, how to see it? --As shown below
Being only know so you can more accurately monitor the resources required information monitoring service, continue to study further update after follow-up, and we hope that with the exchange of learning and common progress!