Automated storage network capacity management utilizing simulation and optimization

Abstract

There is a wealth of research on storage systems, focusing on the storage system architecture, intelligent use of storage resources, and mass storage systems. However, very little research looks at a key component of centralized storage systems, the physical fabric of the network that holds the storage system together. “Best practice” frameworks for the management of storage systems exist, but the design of the network fabric within storage systems has relied heavily on the knowledge and expertise of storage systems architects, administrators and managers. In this paper we present the Network Capacity Management Cycle (NCMC), a new framework for network capacity management. This framework is unique because it was developed around wellestablished Operations Research methods, namely simulation and optimization. Applying these methods to the network fabric of storage systems allows a majority of the framework to become automated, significantly reducing the workload for storage systems architects, administrators and managers. The NCMC utilizes cutting-edge network discovery and monitoring tools, leading network simulation software and new methods for network capacity design. We describe each of the steps of the NCMC in detail and discuss how to automate almost all of these steps. A case study is also presented that demonstrates one iteration of the NCMC applied to an existing storage system in the Department of Engineering Science at the University of Auckland. In addition to illustrating the steps of the process, this case study also outlines the numerous complications we have encountered in our initial use of the NCMC.