Electrical computers and digital processing systems: memory – Storage accessing and control – Specific memory composition
Reexamination Certificate
2001-12-06
2004-08-24
Gossage, Glenn (Department: 2186)
Electrical computers and digital processing systems: memory
Storage accessing and control
Specific memory composition
Reexamination Certificate
active
06782450
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to the field of disk storage subsystems, and more particularly to redundant arrays of independent disks (RAID).
BACKGROUND OF THE INVENTION
Modern, large-scale computer systems are usually configured with client and server computers connected via a network. The network can include local and wide area (Internet) components. The client computers, typically desk- or lap-top computers, provide a graphical user interface (GUI), a relatively small amount of local processing and storage, and user application programs. However, it is the server computers that provide the heavy duty processing, and bulk storage for files and databases. For data integrity purposes, the storage subsystems are usually in the form of a redundant array of independent disks (RAID).
A RAID subsystem protects against a disk drive malfunction. By using many disk drives, and storing redundant data along with user data, a disk drive failure will not cause a permanent loss of data. The manner in which the RAID subsystem provides data redundancy is called a RAID level. A number of RAID levels are known. RAID-1 includes sets of N data disks and N mirror disks for storing copies of the data disks. RAID-3 includes sets of N data disks and one parity disk. RAID-4 also includes sets of N+1 disks, however, data transfers are performed in multi-block operations. RAID-5 distributes parity data across all disks in each set of N+1 disks. At any level, it is desired to have RAID systems where an input/output (I/O) operation can be performed with minimal operating system intervention.
FIG. 1
, in a very general way, shows a model of the interactions between an application program
101
and physical storage media
111
of a computer system, be it a client or a server computer. The application
101
makes non-redundant file I/O requests
102
, or “calls,” to a primary file system
104
to access non-redundant file I/O data
103
. The application can be a foreground application, for example a word processor, or a background application, e.g., a file back-up system. Generally, the access requests
102
can be for data input (read) or data output (write) operations.
The primary file system
104
typically assumes the physical storage media is in the form of a block mode device
111
. The block mode device can be single disk, multiple disks, or tapes, or other high capacity, relatively low latency, non-volatile memories. Therefore, the primary file system makes non-redundant block I/O requests
105
to a block server
107
of a prior art block mode RAID subsystem
100
to read or write non-redundant block I/O data
106
. The RAID subsystem
100
uses a block mode interface
110
and makes redundant block I/O requests
108
to the disks
111
for redundant block I/O data
109
.
Clearly, the primary function of the traditional block mode RAID subsystem
100
is to translate non-redundant block I/O requests and non-redundant block data into redundant block I/O requests and redundant block data. Storing at least two copies of each data block on at least two different physical devices provides this redundancy, so that should one device fail, the block can still be recovered. In some RAID levels, parity blocks provide the redundancy.
FIG. 2
shows interactions in a client-server type of arrangement of computers with a primary file system
104
configured to work over a network
204
. Here, the file system
104
has a client side
201
and a server side
202
. The network
204
transports data between the client side
201
and server side
202
of the file system
104
. The application
101
directly calls
102
the client side
201
of the file system
104
, and the server side
202
makes calls
105
to the traditional block mode RAID subsystem
100
of the server system
203
.
In the arrangements shown in
FIGS. 1 and 2
, the RAID subsystem
100
is used to increase reliability of the system. However, the RAID subsystem
107
protects only against failures in the block mode device
111
. Therefore, there are still many other points of failure in the system, each one represented by the components other than the disks used in these arrangements. To protect against failures by these other components, one must provide redundancy for the other components as well. Some examples of these components are memories, busses, controllers, and processors. The term storage area network (SAN) is typically used to describe this type of redundant arrangement.
FIG. 3
is an example of a SAN
300
. Client computers
301
-
303
communicate with the SAN via the network
204
. The SAN
300
appears as one large server computer to the client computers
301
-
303
. The SAN
300
includes server computers
321
-
323
, connected by a redundant bus
331
to shared RAID controllers
341
-
342
, and the RAID controllers
341
-
342
are connected to a shared block mode device
361
via a shared bus
351
which may also be redundant. Thus, any component in the SAN
300
can fail without losing the ability to serve the client computers.
Large scale SANs are complicated and usually configured for specific mission-critical applications, for example, banking, stock markets, airline-reservation, military command and control, etc. In addition, elaborate schemes are often used to provide redundant block-mode data access via wide area networks (WANs) in case of major disasters. Therefore, SANs usually includes many proprietary components, including much one-of-a-kind software that performs system management. The low-volume, proprietary aspects of SANs makes them very expensive to build and operate.
Another approach to allowing redundancy across major components is to virtualize files at the file system level, and serve a set of files from that, see for example, U.S. Pat. No. 5,689,706 issued to Rao on Nov. 18, 1997 “Distributed Systems;” U.S. Pat. No. 6,163,856 issued to Dion on Dec. 19, 2000 “Method and Apparatus for File System Disaster Recovery;” and U.S. Pat. No. 6,195,650 issued to Gaither on Feb. 27, 2001 “Method and Apparatus for Virtualizing File Access Operations and Other I/O Operations.”
However, these prior art SAN systems still have the following problems. They require the use of a specific proprietary distributed file system. They do not allow the use of file systems that are standard to client processors. They cannot be used with databases or other applications that use a block mode device with no file system. Because of these limitations, systems based on those implementations may never provide the features in widely used file systems, and may be limited to a few expensive operating systems.
Therefore, there still is a need for a system and method that provides data redundancy using standard components, interfaces and networks, and provides block mode access for maximum flexibility of application usage.
SUMMARY OF THE INVENTION
The present invention provides data redundancy at the file level, instead of at the block level as in the prior art. The redundancy is provided in a file mode form, rather than a block mode form as in the prior art. Therefore, file data can be located on any system or server, including a local system, or a server on a local area network, or a remote server on a wide area network. Because files are easily shared over networks through standard high volume, low cost hardware, software, and protocols, the file mode redundancy based on files has a level of data redundancy that is as high or higher than a traditional SAN, with more flexibility than a distributed file system. Using the invention, most costs remain consistent with high volume commodity components.
Depending on where files are stored, high performance and reliability can be achieved through disks on the local system that include file systems, and extremely high reliability can be achieved by using disks on network servers that have file systems. With the invention, disaster recovery is trivial to implement because files can be shared over a WAN, using well-known protocols, among any system which uses any operatin
Arnott Randy M.
Noya Eric S.
Gossage Glenn
Raidcore, Inc.
Sterne Kessler Goldstein & Fox p.l.l.c.
LandOfFree
File mode RAID subsystem does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with File mode RAID subsystem, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and File mode RAID subsystem will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3356220