Obstacles to the implementation of performance tests
- Performance tests tend to be dragged before finally executed
- pro forma phenomenon
- will do some meaningless test
- does not have the perfect test environment
- less than the full level of staff participation
- expensive tools
User perspective
• Response time
Development Perspective
• comprehensive needs analysis
• system architecture is reasonable
• code execution efficiency
Systematic Perspective
• system capacity
• concurrency
• Stability
• Scalability
Fault-tolerance test includes testing two aspects:
- Data input error or abnormal operation of the system to verify protection. If fault tolerance is good, the system gives only a hint or internal digest, without causing system failure or crash.
- Disaster recovery testing. By all means, let the software failure occurs forcibly, and then verify that the system has saved user data is lost, whether system and data can be restored quickly.
Concept of automatic recovery testing can be seen, fault tolerance testing is an antagonistic test process. When testing software failure, how to recover useful data transfer and failure. Failover (failover) to ensure that the test object when a failure occurs, can be successfully transferred in a key part of the operation of the system or the system to continue to operate other devices, i.e., the standby system will seize the opportunity to "replace" a system failure, in order to avoid any loss of data or transactions, does not affect the user. To fully test failover, a good method is to test all of the objects portrayed with a system architecture diagram to figure all point of failure design test cases that may occur. The key object existing single point of failure in the system configuration in FIG, is to design flaws.
Reproduced in: https: //www.cnblogs.com/henryhappier/archive/2010/08/03/1791634.html