performance - Calculation of system response time -
i've developed software , i'm going monitor performance of system, system working well, except in periods of working faces slowness. have prepared o log related system response , time response has been generated, log time below: process#1 step1: 300 millisecond step2: 700 milliseconds step 3: 10 millisecond process#2 step1: 10 millisecond step2: 50 milliseconds step 3: 4 milliseconds process#3 step1:2 millisecond step2: 40 milliseconds step 3: 80 milliseconds
now want detect process has suspicious response time, rule in concepts of software engineering? mean have tried values greater average + 3* standard deviation, did not work, few data has been marked know data more 100 millisecond not normal rule purpose in software engineering, how detect bottle neck of system?
there's no real rule of thumb finding bottlenecks. rule is, measure objectively parts of application slow, , have definitive answer.
much engineering time wasted trying optimise code turns out not slow in first place. measure, , act. make sure can reproduce slow performance, can repeat test , test changes reliably.
Comments
Post a Comment