In the actual performance tests, the testers often concerned about the number of concurrent users, that is, from a business point of concern whether we should set the number of concurrent reasonable, the following is a way of estimating the number of concurrent users:
(1) calculate the average concurrent users number: C = nL / T
(2) the peak number of concurrent users: C '≈ C + 3 C root
Equation (1), C is the average number of concurrent users; n is the number of login session; L is the average length of loginsession; refers to the period length T investigated.
Equation (2) gives the calculated peak number of concurrent users, where, C 'refers to the average number of concurrent users of the peak number of concurrent users, C is the formula (1) is obtained. The derived equation is assumed that the user generates a Poisson distribution loginsession estimates obtained.
Example:
Suppose there is a OA system, which has 3,000 users, with an average every day about 400 users to access the system, a typical user, the day the user logged on to withdraw from the system, the average time was 4 hours, time of the day, the system user only within 8 hours.
According to the equations (1) and formula (2) can be obtained:
C = 400*4/8 = 200
Root C'≈200 + 3 * 200 = 242
There is also a wide range of users concurrent formula
C=n/10
C ^ = r * C (typically r = 2 ~ 3)
The maximum number of users accessing the system usually 10% of the average number of concurrent users;
2-8 What principles and so on.
----------------