Pooja Jain, Intel Corporation
Cloud performance has been known to be complex and unpredictable. In this paper, we assessed software performance quality by quantifying the performance variation of running applications in the cloud. Using one commercial tool and one free tool, we took on the task of collecting performance data running in a cloud lab. We set up multiple software workloads and varied the configurations to run them in the lab. We set up software load testing tools to measure the performance in the cloud. We also collected performance data at the system level, the software level and the CPU level to help understand changes in software performance quality corresponding to cloud configuration changes.
The contributions of this presentation are:
- The variation of performance data corresponding to software configurations in the cloud.
- Insights and interpretations of statistics collected by the tools
- Pros and cons of collecting performance data from these tools