Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
2000-09-06
2001-10-09
Iqbal, Nadeem (Department: 2184)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C711S112000
Reexamination Certificate
active
06301677
ABSTRACT:
REFERENCE TO PAPER APPENDIX
This application incorporates by reference the computer program listing contained in the attached paper appendix. The paper appendix includes 471 pages.
1. Field of the Invention
The present invention relates to improvements in the field of computer systems having backup/restore or archive/retrieve subsystems. More particularly, the present invention relates to a method and apparatus to efficiently protect and archive active data to streaming media.
2. Background Information
In a data processing system, a backup/restore subsystem, usually referred to as a backup subsystem, is typically used to save a recent copy of an active file and several earlier versions. There are, for example, three general strategies employed by backup systems. First, full backup periodically copies all files from a client system's storage to a backup server. A second strategy includes incremental backup, where the client system copies only the modified files to the backup server. In a third strategy, a delta backup copies only the modified portions of the modified files to the backup server.
Complete discussions of various backup and storage technologies are disclosed, for example in U.S. Pat. No. 5,479,654 entitled APPARATUS AND METHOD FOR RECONSTRUCTING A FILE FROM A DIFFERENCE SIGNATURE AND AN ORIGINAL FILE, which is hereby incorporated by reference. Applicants co-pending applications entitled COMPUTER APPARATUS AND METHOD FOR MERGING A SEQUENTIAL PLURALITY OF DELTA STREAMS, filed Nov. 14, 1997, and COMPUTER APPARATUS AND METHOD FOR MERGING SYSTEM DELTAS, filed Nov. 30, 1995, also relate to backup and storage technologies and are hereby incorporated by reference. In addition, the Storage Service Kit, ©1996 by Mark Squibb also relates to data storage systems and is hereby incorporated by reference.
It is apparent to those skilled in the art that in any given backup system, the higher the backup frequency, the more accurate the backup copy will represent the present state of the data within a file. Considering the large volume of data maintained and continuously generated in a data processing system, the amount of storage, time and other resources associated with protecting data are very substantial. Thus, those skilled in the art are continuously engaged in searching for better and more efficient ways to provide data protection.
The time lag between the last backup, for example by the backup methods described above, and the current data on an active system represents risk of data loss. This “protection gap” is an active concern among computer users because it represents unprotected information. Mirroring systems, described below, partially overcome this gap.
It is well known in the art to capture write events to a storage system. For example, each time a change is made to a storage device, the change is recorded or logged into a second media. A variety of types of media have been used for recording of logs including, for example, streaming tape, hard disk, and remote hard disk.
A write event comprises, for example, a storage indicator indicating what storage component or device the write applies to, a position indicator within the component or event as an offset telling where in the storage the write occurred, and the data (e.g., event data) which was written. Various embodiments also include time or sequence markers for synchronization to identify or select points in time and to coordinate state information across storage boundaries. A collection of write events is known as an event log, referred to as an event journal when the event log is stored on a storage device.
Since events are recorded just after they occur, event logs are ordered in chronological sequence. Events at the beginning of the journal occurred before events at the end of the journal. Creation of event logs is well known in the art. Replaying event logs to re-enact changes to a storage component is also well known in the art.
Prior art systems use event journals to replay changes to random access writeable media. The word “replay” means, for example, to chronologically re-enact the storage write events that resulted in a particular file given an original file. The replay process begins with a disk file on random access storage and an event log synchronized with the disk file. The initial file's Data State, for example, must correspond to the starting instant of the event log. The events in the event log are repeated on the disk file in the sequence that they occur in the log.
Each event is read from the event journal in sequence from beginning to end. After each event is read, the corresponding event offset is located in the disk file. This location process usually involves repositioning in the disk file to a random position that may be before or after the position of the last event offset. The event data is then written to the file at the new offset. Old information in the file is destroyed because the new data overlays the prior data. This process is repeated until the event log is exhausted. When the process completes, data in the revised file represents the final Data State represented by the event log.
Mirror systems duplicate changes as they occur. Storage devices are usually treated as block devices. When change occurs, a write event is packaged and transmitted to a remote mirror system. Upon receipt, the remote mirror duplicates the change in the mirror storage. Mirror systems sometimes employ event logs to store events.
Event logs are used in mirroring systems for several purposes. For example, event logs are used in mirroring systems to compensate for transmission delays at the source system. At the mirror system, event logs are used to cache events when the mirrored storage cannot keep up with incoming write events. It is also known to temporarily halt incoming events to a mirror system so that the data to the mirror system is constant during backup of the mirror.
From a data protection perspective, there are two types of events that compromise data access. The first and best understood is a hardware failure. Mirroring effectively prevents a hardware failure from compromising business continuance. The second event type is a logical failure. A logical failure occurs when a user, operator or application does something that destroys, corrupts or distorts information. Mirroring systems immediately and irreversibly duplicate logical errors.
Mirroring systems have several deficiencies. For example, mirroring does not protect from logical failures. In addition, the need to re-execute every write event on another random-access storage device shortly after the initial event effectively requires duplication of all active storage in a second storage system. The cost of the extra second storage and the management overhead prevents widespread adoption of mirroring technology.
Mirroring does eliminate the data protection gap inherent with backup technology. Mirrored systems immediately protect new data, unless it is destroyed by a logical failure. The volatile nature of mirrored storage, however, is a deficiency. The cost of duplicating massive storage is also a deficiency. In contrast, the present invention provides the same data protection as mirrored storage at much lower cost and provides recourse for logical errors.
Clustering is a type of mirroring. Clusters are a collection of servers that maintain a mirror distributed among several other servers on the network. Clustered servers share the logical fault intolerance and storage doubling characteristics of mirrored systems.
File mirroring systems are another type of mirroring. It is known in the art to have a many-to-one mirror, for example when a single system provides a logical mirroring service for a number of network servers. The single system collects change from the servers on a network. The change is stored or applied to a hard disk cache of active data files. Periodically the data file versions are backed up to tape. File mirror systems, however, share the same deficiencies as mirror systems because all active data must be stored on
Baker & McKenzie
Delta-Tek Research, Inc.
Iqbal Nadeem
LandOfFree
System and apparatus for merging a write event journal and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and apparatus for merging a write event journal and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and apparatus for merging a write event journal and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2613710