Optimization of data access in tiered storage

Abstract

The specifics of data access in multi-tiered data storage are considered. The problem of data blocks optimal allocation with taking in consideration the technical characteristics of storage tiers and applications requirements is formulated. Three models for different variants of data storage types, user requests and expenses minimization policies are proposed. Methods for solving these problems were described. The variant of genetic algorithm to determine the optimal allocation of data blocks onto different tiers of data storage is developed. Principles of implementation for cases with file system virtualization and databases with and without table partitioning are described.