|
Sponsored Links: |
* Client / Server check requires hardware, application & human resources. Before the tests, you ought to know what resources are obtainable & select how to use effectively.
* Load testing means that the evidence of their client / server method under typical working conditions, & verification of method performance, reliability, capacity, etc.
* Objectives of the selective tests & check criteria to ensure that the tests successfully. For example, it is to define the vague goals like "the server response time under heavy load check." A success criteria would be more focused "Make sure that 50 customers can check their account balance simultaneously, & the server response time shall not exceed two minute.
Analyzing the Client / Server
The first step in the planning of the burden of proof is to analyze the client / server method. You ought to familiarize yourself with the hardware components & application, method configuration, & the typical usage model.
Identification of method components
Draw a diagram to illustrate the structure of the client / server method. If possible, extract a schematic diagram of the existing documentation. If the method under check is part of a larger network method, you ought to identify the method component to be tested. Ensure that the method includes all method components, such as customer machines, network, middleware & servers. The following diagram illustrates a chain of 60 shops. Clothing retailers & distributors of clothing for each connected to the same database to update the stock information & to check prices. Distributors connect to the database server through the branch office LAN, & clothing stores connected by wide area network to the LAN.
This analysis ensures that the check environment is created using LoadRunner accurately reflects the environment & configuration of the method under check.
Defining Testing Objectives:
Stating the objectives in measurable terms
The first step in the planning of the burden of proof is to analyze the client / server process. You ought to familiarize yourself with the hardware components & program,One times you select on its overall objectives of load check, it should provide more focused on the objectives set out its objectives in measurable terms. To provide a basis for evaluation, choose exactly what constitutes acceptable & unacceptable check results. For example:process configuration, & the typical usage model. This analysis ensures that the check environment is created using LoadRunner accurately reflects the environment & configuration of the process under check.
General aim - Product Evaluation: select the hardware for the database server.
Focused Aim - Product Evaluation: run the same group of 300 virtual users on one different servers, HP & Sun SPARC 6000. When all 300 users at one times make one predefined queries, determining what hardware provides a faster response time.
Planning LoadRunner Implementation:
You can use LoadRunner to measure the response time at different points in the client / server method. Decide where to run the Vusers and Vusers to run according to the objectives of the check:
Define the scope of measures
You can measure the response time experienced by a typical user walking a GUI Vuser or RTE Vuser in the front. GUI Vusers emulate real users by providing input and output of receipt of the customer's request; TEN Vusers emulate real users input for the submission and receipt of the output of a character based on the application.
Outcome measurement-to-end response time:
You can run GUI or Vusers TEN at the front to measure the response time across the network, including a terminal emulator or front finish GUI, network and server.
Examining Load Testing Objectives:
Your check plan will be based on clearly defined objective analysis. This section presents an overview of the common objectives of testing:
* Defining the optimal hardware configuration
* Checking the reliability
* Checking hardware upgrades or application
* Evaluation of new products
* The ability of the measurement technique
* Identifying bottlenecks
* Finish of measuring user response time
No comments:
Post a Comment