Enabling Effective End-To-End Design and Configuration Spanning the Database and Storage System Tiers |
Modernstorage systems are sophisticated. Simple direct attached storage devices aregiving way to storage systems that are shared, flexible, virtualized andnetwork-attached. Today, storage systems have their own administrators, who usespecialized tools and expertise to configure and manage storage resources.Although the separation of storage management and database management has manyadvantages, it also introduces problems. Database physical design and storageconfiguration are closely related tasks, and the separation makes it moredifficult to achieve a good end-toend design. In this paper, we attempt toclose this gap by addressing the problem of predicting the storage workloadthat will be generated by a database management system. Specifically, we showhow to translate a database workload description, together with a databasephysical design, into a characterization of the storage workload that willresult. Such a characterization can be used by a storage administrator to guidestorage configuration. The ultimate goal of this work is to enable effectiveend-to-end design and configuration spanning both the database and storagesystem tiers. We present an empirical assessment of the cost of workloadprediction as well as the accuracy of the result.