Performance Modeling of Storage Virtualization01.11.2008
in co-operation with IBM Germany Research and Development GmbH
Project Period: Nov. 2008 - Apr. 2009
People involved (from Descartes Group): Christoph Rathfelder, Nikolaus Huber
Virtualization enables shared usage of hardware resources. Hence, it is important factor especially for server consolidation, promising to save management, space, and energy costs. One research area is the I/O virtualization, which e.g. provides shared storage hardware access. During design and implementation of systems, the influences of different parameters or design decisions on the system's performance are often unknown or hard to estimate. Performance models offer the ability to reflect a system's performance behavior for analysis and evaluation.
In this project a performance model for different design alternatives for storage hardware virtualization was developed. The model was calibrated and validated by means of measurements of an existing system. With this model, the influences of different factors on the system performance can be analyzed and design alternatives can be compared. Furthermore, this project is a case study of PCM's applicability in the field and its capabilities in other domains besides business information systems.