Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
1999-08-17
2004-05-25
Beausoliel, Robert (Department: 2184)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C711S114000
Reexamination Certificate
active
06742137
ABSTRACT:
BACKGROUND
The invention relates generally to file systems for computer systems, and more particularly, but not by way of limitation, to the application of fault tolerance features to the storage of computer system objects.
As information technology progresses, there is a growing need for more efficient and more reliable data storage. Many companies in the marketplace today employ computer systems (including computer networks) which utilize vast amounts of data storage space. Additionally, as corporations begin to rely increasingly on electronic data storage for record keeping, accounting, document storage, and other important business operations, data storage must be dependable.
Conventional computer systems may include fault tolerant mechanisms such as redundant arrays of independent disks (RAID) technology to compensate for hardware failures and/or to increase storage performance. RAID systems typically include several disk units that are used in conjunction to provide fault tolerant data storage at the volume level. A storage volume may be a specified partition within a single disk unit, an entire disk unit, or a collection of disk units. Various levels of RAID technology may be employed to improve performance and/or to maintain data integrity of a storage volume. For example, RAID Level 0 involves striping data objects across multiple disks. Data striping involves storing blocks of data over a series of disk units at the volume level, such that all blocks on a storage volume are divided into a series of bytes that are distributed across the series of disk units. Although data striping may allow parallel disk operations that improve performance, no redundancy is provided. RAID Level 1 allows for complete redundancy by utilizing a mirroring technique at the volume level. Data mirroring involves storing a duplicate copy of all data on a storage volume on a separate volume (the “mirror”). Although mirroring techniques typically require twice as much storage space to accommodate the mirrored data, they provide complete redundancy of a storage volume. With RAID Level 1, an array of disks is divided into two groups such that one group stores the original data while a second group stores the mirrored data. RAID Level 3 includes data striping at the volume level with an additional disk dedicated to parity data. Parity techniques involve the generation and storage of extra information bits that may be used to reconstruct lost data in the event of a storage failure. A parity block may be generated that produces information (the information constitutes the parity block) which allows the original data on a single disk to be recreated from the parity block and data on other disks. Referring to Table 1, an example of RAID Level 3 is shown such that data objects A-E are striped across disk units
1
-
4
, by storing data blocks (e.g., A
1
-A
4
) of the data object (e.g., data object A) on different disk units. The location of a data block within an individual disk unit is arbitrary, and thus a data block may be stored at any location on a particular disk unit as shown in Table 1. All of the parity blocks P
1
-P
5
for data objects A-E are stored on the dedicated parity disk unit
5
. Each parity block may be constructed from data blocks in a parity stripe, illustrated as a row in Table 1, using any parity technique such as an exclusive OR operation (+). For example, parity block P
4
may be the result of a series of exclusive OR operations with data blocks A
3
, E
4
, A
1
, D
1
as operands such that P
4
=A
3
+E
4
+A
1
+D
1
. In addition, Parity blocks are computed from data blocks without regard to the particular object to which the data block belongs.
TABLE 1
RAID Level 3 Example
Another technique, RAID Level 5, involves striping of both data and parity blocks at the volume level. Data and parity striping are combined to provide increased reliability (due to the use of parity) and possibly improved access performance (due to the ability to access multiple portions of a single stored object in parallel because it is distributed across multiple disk units). RAID Level 5 is demonstrated in Table 2 with data objects A-E and parity blocks P
1
-P
5
striped across disk units
1
-
5
. Because RAID Level 5 does not prescribe the pattern in which parity blocks are stored, parity blocks P
1
-P
5
may be interleaved with the series of data blocks or stored elsewhere on the disk units
1
-
5
.
TABLE 2
RAID Level 5 Example
Disk Units
1
2
3
4
5
C3
A4
P1
B4
B3
E2
B1
D3
A2
P2
D4
C2
C4
P3
E3
P4
E4
A1
D1
A3
B2
P5
E1
C1
D2
Current fault tolerance techniques such as those described above are implemented at the storage volume level, requiring the same level of fault tolerance to be applied to all data stored in a storage volume, regardless of the nature of the data. Storing data in this manner is inflexible and does not allow different fault tolerant techniques to be applied to different data objects on a single storage volume. Also, reconstructing data of an object from a failed disk unit may require retrieving data from unrelated objects which otherwise would not need to be retrieved. Thus, it would be beneficial to provide a more flexible fault tolerant computer storage system.
SUMMARY
In one embodiment the invention provides a method to store data objects in a storage medium. The method includes receiving an identifier associated with a data object and associating a fault tolerance indication with the identifier. The method also includes storing the data object in the storage medium in accordance with the fault tolerance indication. In another embodiment of the invention, a program storage device that is readable by a programmable control device includes instructions stored on the program storage device. The instructions cause the programmable control device to receive an identifier associated with a data object and associate a fault tolerance indication with the identifier. The instructions also cause the programmable control device to store the data object in a storage medium in accordance with the fault tolerance indication. In yet another embodiment, the invention provides a memory for storing data for access by a storage retrieval program. The memory includes a data structure having a plurality of entries for use by the storage retrieval program. The entries in the data structure include an object identifier to identify a stored object and a fault tolerance identifier to identify a level of fault tolerance applied to the stored object.
REFERENCES:
patent: 4993030 (1991-02-01), Krakauer et al.
patent: 5083264 (1992-01-01), Platteter et al.
patent: 5287500 (1994-02-01), Stoppani, Jr.
patent: 5355475 (1994-10-01), Tanaka et al.
patent: 5392244 (1995-02-01), Jacobson et al.
patent: 5444719 (1995-08-01), Cox et al.
patent: 5479653 (1995-12-01), Jones
patent: 5588110 (1996-12-01), DeKoning et al.
patent: 5657468 (1997-08-01), Stallmo et al.
patent: 5857112 (1999-01-01), Hashemi et al.
patent: 5893919 (1999-04-01), Sarkozy et al.
patent: 5960169 (1999-09-01), Styczinski
patent: 6016552 (2000-01-01), Lee et al.
patent: 6016553 (2000-01-01), Schneider et al.
patent: 6038679 (2000-03-01), Hanson
patent: 6098119 (2000-08-01), Surugucchi et al.
patent: 6219814 (2001-04-01), Coker et al.
patent: 6269424 (2001-07-01), Katsuragi et al.
patent: 6571314 (2003-05-01), Komachiya et al.
patent: 2002/0059539 (2002-05-01), Anderson
patent: WO 9812621 (1998-03-01), None
patent: WO 99/08173 (1999-02-01), None
patent: WO 99/09479 (1999-02-01), None
patent: WO 9918507 (1999-04-01), None
Adaptec, Inc.
Beausoliel Robert
Chu Gabriel L
Martine & Penilla LLP
LandOfFree
Object oriented fault tolerance does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Object oriented fault tolerance, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Object oriented fault tolerance will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3201402