Shulou(Shulou.com)06/03 Report--

For the concepts we talked about in the previous article, we are here to explain them one by one.

Scene

The so-called scene refers to the profile in which the system is running. Corresponding to the three-dimensional model we established before, we often refer to the section on the timeline, which is determined by the L axis and O axis, that is, the scene has two attributes: load and load object.

Load capacity

Load, which is mainly used to describe the quantitative identification of current system pressure, including user volume, business volume, throughput, PV, etc.

When we carry out load analysis and statistics, the purpose is to obtain the load of a load object in our scenario. For example, for the whole system, we need to count the current number of users and business of the whole system at a certain point in time. What is the throughput of the system and so on. From this, we can get information about how many people should participate in the load in our test model, and how much pressure people expect to put on the system.

Load object

Load objects, including functional modules, pages, page components, interfaces, etc., are mainly used to describe the pressure distribution of the system; load objects are analyzed by tree structure in the process of modeling. As shown in the following figure:

The analysis of the load object is mainly to analyze the load pressure on different load objects under a certain load, and according to this, to determine how the users we test should be distributed.

Running

The so-called operation is the state change of the profile of the system on the timeline. There are two key words involved, one is the timeline and the other is the state change of the profile.

First of all, as the timeline, we can naturally choose a certain time point or a certain time period. However, it is difficult to enumerate the absolute time point, and if the granularity is too fine, it also loses its statistical representativeness. Therefore, generally speaking, we choose a certain period of time as the scale of the profile change.

As for the state change of the profile, as we said earlier: the profile of a system contains two attributes, the load and the load object, according to the different properties we observe, there are two different angles of observation for the operation of the system, that is, the state change of the load in time and the state change of the load object on the timeline.

The state change of the load on the timeline is called the load change trend, while the state change of the load object on the timeline is called the user behavior trend.

Load changing trend

The load trend has been defined previously, that is, the state change of the load on the timeline. The commonly used standard for the interpretation of the timeline is also given, that is, a certain time period.

For the selection of time periods, we first define a standard, that is, minutes as the most atomic analysis scale, that is, the load per minute as a sampling point. The sampling point is defined as the instantaneous load of the system, which is what we usually call absolute concurrency. We call it OT1. The sampling results selected during this period are mainly used to judge whether the real concurrent users during the test are consistent with the pressure of the actual production system. For a relatively concurrent test execution process, it is total user * concurrency = OT1;. For an absolutely concurrent test execution process, it is concurrent user = OT1.

In the analysis of state change, the hour is taken as the most atomic analysis scale of the change analysis, that is, the load change curve described by at least 60 sampling points within an hour can be regarded as a meaningful load change analysis sample. The total load of one hour is defined as the online load of the system, which is what we usually call relative concurrency. We call it OT0. The sampling results selected during this period are mainly used to determine the number of users during the execution of the test.

User behavior trend

The trend of user behavior refers to the state change of the load object on the timeline, that is, how much the load is decomposed into different load objects in a certain period of time, and the load change curve of each load object.

From this concept, we can see that the sampling of user behavior trends includes at least two perspectives:

One point of view is the load change, that is, the duration of the load generated by a load object, such as the access cycle of an object in the system (session survival time) and the load convergence order between different objects (user operation flow), or, for example, the curve of login function changing its load over time. Sampling from this perspective determines which actions should be included in the instantiated script of each test case when the test is executed, the execution time (how much thinking time to set), and the operation cycle (iteration interval).

Another angle of sampling is load distribution, that is, the amount of load at a certain point in time, which load objects are, and how big each load object is, a more popular term is functional distribution. Sampling from this perspective determines which functional test cases to focus on when performing tests, and what percentage of users should be allocated to the instantiated scripts of each test case.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.