Automatic calibration of database cost model in cloud computing

Cloud computing is a new generation of computing and is based on virtualization technology. More and more applications are being deployed in cloud environments. One important application is Database Management Systems (DBMS). Previous work, namely the Virtual Design Advisor (VDA), bas addressed the problem of optimizing the performance of DBMS instances running on virtual machines that share a common physical machine pool. The VDA needs to calibrate the Inning parameters of the DBMS's query optimizer in order to operate it in a what-if mode to accurately and quickly estimate the cost of database workloads running in virtual machines with varying resource allocation. The work in this paper addresses the calibration process, which is found to be a complex, time-consuming task if done manually. Each time a DBMS has to run on a different server infrastructure and each time the DBMS is replaced with another on the same server, the calibration process potentially has to be repeated. This paper presents the design, implementation, and evaluation of a tool, namely Automatic calibration (ACT), to automate the calibration process. ACT contains two modules; controller and worker, which interact together to construct a calibration model. The tool bas been evaluated in Inning the PostgreSQL Inning parameters. The results show that the running time of ACT increases linearly with the number of calibration sampling points.