Redundant Disk Arrays in Transaction Processing Systems
Mourad, Antoine Nagib
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/72019
Description
Title
Redundant Disk Arrays in Transaction Processing Systems
Author(s)
Mourad, Antoine Nagib
Issue Date
1993
Doctoral Committee Chair(s)
Fuchs, W. Kent
Saab, Daniel G.
Department of Study
Electrical Engineering
Discipline
Electrical Engineering
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Engineering, Electronics and Electrical
Computer Science
Abstract
Disk arrays are a cost-effective approach for building large, reliable and high performance storage subsystems. They provide high transfer rates by striping data over multiple disks and use a parity scheme for recovering from any single disk failure. Transaction processing is a large and growing segment of commercial computing. There is a major need in that environment for large, highly available and fast I/O subsystems.
In this thesis, we address various issues dealing with the use of disk arrays in transaction processing environments. We look at the problem of transaction undo recovery and propose a scheme for using the redundancy in disk arrays to support undo recovery. The scheme uses twin page storage for the parity information in the array. It speeds up transaction processing by eliminating the need for undo logging for most transactions. The use of redundant arrays of distributed disks to provide recovery from disasters as well as temporary site failures and disk crashes is also studied. We investigate the problem of assigning the sites of a distributed storage system to redundant arrays in such a way that the cost of maintaining the redundant parity information is minimized. Heuristic algorithms for solving the site partitioning problem are proposed and their performance is evaluated using simulation. We also develop a heuristic for which an upper bound on the deviation from the optimal solution can be established.
Another part of the thesis focuses on the performance of various disk array organizations in transaction processing environments. Trace data from large scale commercial transaction processing sites are used to evaluate and compare the performance of those organizations. We investigate the use of a nonvolatile cache in the disk array controller to reduce the effect of the high cost of small writes. For noncached systems, we evaluate two redundant disk array organizations and compare them to mirrored disks and nonredundant, nonstriped organizations. For cached systems, we consider the above four organizations as well as a disk array organization that uses a dedicated disk for parity in each array and buffers parity updates in the controller cache before spooling them to the parity disk.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
