A Mathematical Programming Approach for Server Consolidation Problems in Virtualized Data Centers

Today's data centers offer IT services mostly hosted on dedicated physical servers. Server virtualization provides a technical means for server consolidation. Thus, multiple virtual servers can be hosted on a single server. Server consolidation describes the process of combining the workloads of several different servers on a set of target servers. We focus on server consolidation with dozens or hundreds of servers, which can be regularly found in enterprise data centers. Cost saving is among the key drivers for such projects. This paper presents decision models to optimally allocate source servers to physical target servers while considering real-world constraints. Our central model is proven to be an NP-hard problem. Therefore, besides an exact solution method, a heuristic is presented to address large-scale server consolidation projects. In addition, a preprocessing method for server load data is introduced allowing for the consideration of quality-of-service levels. Extensive ...

• Type: Journal
• Year: 2010
• Where: TSC
• Authors: Benjamin Speitkamp, Martin Bichler
• Keywords: Artificial Intelligence, Server Consolidation, Server Load Data, Target Servers, TSC 2010