Browse Prior Art Database

Improved performance on file restore for archive or backup data Disclosure Number: IPCOM000021489D
Original Publication Date: 2004-Jan-21
Included in the Prior Art Database: 2004-Jan-21
Document File: 3 page(s) / 60K

Publishing Venue



This article discusses a method to allow quicker restores of archived data stored to disk. The method involves giving filesystems access to the data and metadata for immediate read use and a design for a transition to full read/write use.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 49% of the total text.

Page 1 of 3


BEA820020041 Abdy Raissinia/San Jose/IBM Rob Basham, Michael Kaczmarski, Dave Wolfe

Improved performance on file restore for archive or backup data


For backup and archival applications, restore time can be critical. Disk storage costs continue to drop dramatically, allowing for an increased use of disk as an archival medium. With this trend in mind, a method has been developed to improve the restore time for data archived to disk.

Presently, backup applications with disk-based archive pools restore data by copying it to a specified location. The most advanced applications can make use of direct storage-to-storage copy techniques that have recently been released. While direct copy may improve copy performance somewhat, the process still requires that a copy be done before the archive data can actually be used by a file system or database. Archived data is not used directly by applications because the data has been formatted and possibly compressed as part of the archiving process. Similarly, virtual tape applications only allow data to be accessed sequentially, making it unfeasible for applications to directly use the archived data.

A key development that makes portability between archived data and the file system more attractive and feasible is the advent Storage Area Network-based file systems. These file systems can organize data for multiple servers that could be physically stored in separate buildings or even different locations within a metropolitan area (typical of archival configurations) without significant loss of performance. These sophisticated file systems also have the necessary controls and interlock infrastructure already in place to directly deal with archived data formats typical of virtual tape applications or disk-based archive pools.

Backup Requirements

There are two key actors in the accelerated restore process: the archive application providing archive storage and retrieval services and the file system that is the client of this service. The presumption is that the archival service backs up the data to a disk medium that is on the same network -- most likely a storage network -- the file system client has access to. The file system does not need access to the data; it is probably best to limit access and only provide it as needed when the restore is requested. The metadata may be compact enough that direct access by the client may not be needed and it does not necessarily need to be on a network the file system client has access to.

In order to meet this requirement, the archive service needs information from the file-system requesting an archive about the set of file system clients (possibly in the former of storage area network world-wide names or IP addresses) that use the set of data being archived. From this information, the archive service can see which archive


Page 2 of 3

storage pools are on the network or ne...