2015 IEEE International Conference on Data Science and Data Intensive Systems (DSDIS) (2015)
Dec. 11, 2015 to Dec. 13, 2015
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/DSDIS.2015.94
Today customers tend to supersize their infrastructure requirements in order to readily accommodate the maximum number of concurrent users they envision using their applications on any given day. This leads to a proliferation of servers that largely go unused throughout the year. Therefore, analytical models and tools are needed to allocate servers in a resource-efficient way. This paper intends to study the approaches of measuring and predicting the application loads in a more precise way so that the appropriate number of servers can be determined dynamically and optimally. In this strategy, the usage patterns from users are investigated firstly and then the application load pattern are discovered, which serves as the basis for future load prediction. After the load prediction information is obtained, a dynamic approach for capability planning is investigated. Experiment results show that when applying our capacity planning strategy, we are able to have a 30% reduction in server numbers.
Servers, Capacity planning, Time series analysis, Planning, Pattern matching, Computer science, Resource management
L. Wang, J. Cao and Y. Qu, "A Prediction Based Capacity Planning Strategy for Virtual Servers," 2015 IEEE International Conference on Data Science and Data Intensive Systems (DSDIS), Sydney, Australia, 2015, pp. 46-52.