Background introduction:
because " XX reading Platform "is a a public welfare project, from product design, development to testing are volunteers to participate in, the project organization structure is loose, to such a product performance analysis research did not judge the requirements of the benchmark, only according to the common performance analysis strategy to carry out work, according to the characteristics of the project products, performance analysis test strategy is as follows:
Product Phase I
Product target: complete 60% of the 1 level and sub-function, two platform client development only for the function demo, used to express the product prototype, more articles and book display mainly
Performance test strategy: Do not perform performance testing, ready to test data
Product Phase Two
Product target: Complete the 80% function
Target User: XXX monks and mages in the monastery
Performance Test Objectives: perform performance testing and tuning on the server before you go online
Performance Test Strategy:
1. User common interface and operation
2. DB interface with frequent interactions
3. Consumption type operation: Chip load consumes CPU, book download test bandwidth, online read support maximum number of connections
Product Phase Three
Product Goal: to the xxx Temple volunteers and lay open, depending on the situation of the product to provide application market download use
Target Users: Real users
Performance Test Objectives: Complete client performance testing and tuning before you go online
Performance Test Strategy:
1. Time-consuming critical operations for clients
2. Power consumption for a certain period of time for continuous use of content
3. Flow loss for critical operations
4. To ensure the smoothness of the application, to the memory consumption of the analysis
5. Due to the development from the all corners, the statistics hope to establish the Flash back log upload function for development analysis
The performance indicators are as follows:
1. The execution of the method is time-consuming in milliseconds
2. The operation level is time-consuming according to industry standard 1,3,5,8,10. The operation level is at least 3s or less
3. Server response time and TPS
4. The stress test observes the server overhead and gives recommendations for best use of the product:
Base: CPU 60% Max to 80%
This article is from the "Yekoven" blog, make sure to keep this source http://leafwf.blog.51cto.com/872759/1618878
XX reading platform-performance analysis research