The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Performance evaluation of various allocation methods in a heterogeneous disk array architecture
Author:	Branzoi, Bogdan Alexandru
View Online:	njit-etd2005-004 (xi, 49 pages ~ 2.9 MB pdf)
Department:	Department of Computer Science
Degree:	Master of Science
Program:	Computer Science
Document Type:	Thesis
Advisory Committee:	Thomasian, Alexander (Committee chair) Nassimi, David (Committee member) Gerbessiotis, Alexandros V. (Committee member)
Date:	2005-01
Keywords:	Data allocation Disk array architecture
Availability:	Unrestricted
Abstract:	Dataset attributes, such as data availability levels and access patterns, make their mapping to certain RAID levels more desirable than others. On the other hand, it is not economically viable for an installation to acquire multiple disk arrays to satisfy diverse data storage requirements. A Heterogeneous Disk Array (HDA) architecture is proposed, which allows device heterogeneity as well as RAID level heterogeneity. In other words, various disks of different types can be incorporated in a single HDA and multiple RAID schemes can coexist in the same array. The goal of this architecture is to utilize the resources of all its disks to the maximum possible extent by using appropriate RAID levels to meet the varying availability requirements for different applications. An improved best-fit allocation algorithm is proposed and various data allocation methods are tested against it. In an HDA system, each new object is associated with an appropriate RAID level and the allocation is carried out in a way to keep disk bandwidth and capacity utilizations balanced. The data structures of the HDA architecture are described and the flowcharts for the most frequent operations are depicted. Then a data allocation algorithm is formulized and a possible solution is given. Finally, the HDA architecture is prototyped based on the DASim simulation toolkit developed at NJIT and comparison results of various data allocation algorithms are presented.