Associating Performance Measures with Perceived End User Performance: ISO 25023 compliant Low Level Derived Measures

摘要

This paper applies a measurement procedure to predict the degraded state of a private cloud application using only available data center log low level derived measures (LLDM). Our intent is to improve the discussion of service level agreements of a widely used private cloud computing application (i.e. 80,000 users on 600 servers world-wide). In organizations, cloud application performance measuring is often based on subjective and qualitative measures with very few researches to address the large-scale private cloud perspective. Furthermore, measurement recommendations from ISO proposals (i.e. ISO 250xx series, ISO/IEC 15939 and more recently the ISO/SC38-SLA framework) are poorly adopted by the industry, mainly due to the absence of proof of concept and the high degree of complexity associated with implementing the measurement concepts described in these international standards. To try to demonstrate these concepts, the ISO 25010 performance efficiency characteristics are used with a number of LLDMs to model the state of a large private cloud computing application using indicators such as: normal, abnormal, adequate or degraded. This application still cannot be generalized due to its nature as research in progress.