Electrical computers and digital processing systems: memory – Storage accessing and control – Specific memory composition
Reexamination Certificate
1999-12-29
2002-12-31
Peikari, B. James (Department: 2186)
Electrical computers and digital processing systems: memory
Storage accessing and control
Specific memory composition
C711S113000, C711S112000
Reexamination Certificate
active
06502166
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to data storage systems having user configurable levels of input/output (“I/O”) performance and fault tolerance. More particularly, the present invention relates to a system, apparatus, and method for distributing data across multiple disk drives that provides exceptional levels of I/O performance and one-hundred percent data redundancy.
BACKGROUND OF THE INVENTION
Disk drives in all computer systems are susceptible to failures caused by temperature variations, head crashes, motor failure, controller failure, and changing voltage conditions. Modem computer systems require, or at least benefit from, a fault-tolerant data storage system, for protecting data in the data storage system against instances of disk drive failure. One approach to meeting this need is to provide a redundant array of independent disks (RAID) system operated by a disk array controller (controller).
A RAID system typically includes a single standalone controller, or multiple independent controllers, wherein each controller operates independently with respect to the other controllers. A controller is generally coupled across one or more input/output (I/O) buses both to a an array of disk drives and also to one or more host computers. The controller processes I/O requests from the one or more host computers to the rack of disk drives. Such I/O requests include, for example, Small Computer System Interface (SCSI) I/O requests, which are known in the art.
Such a RAID system provides fault tolerance to the one or more host computers, at a disk drive level. In other words, if one or more disk drives fail, the controller can typically rebuild any data from the one or more failed disk drives onto any surviving disk drives. In this manner, the RAID system handles most disk drive failures without interrupting any host computer I/O requests.
Fundamental to RAID technology, is the concept of“striping,” or dividing a body of data, from a host computer, into data segments and distributing the data segments in a well-defined manner across each disk drive in the disk drive array. In this manner, the disk drive array becomes, in effect, one logical storage unit as far as a host computer is concerned. There are a number of well known data striping techniques, or RAID levels, including RAID levels
0
-
6
. A numerically higher RAID level does not imply an increase to the disk subsystem's fault tolerance (reliability), I/O performance and scalability. Instead, the numerical levels refer to different techniques that balance various levels of reliability, I/O performance and scalability.
To illustrate this balance, consider that RAID level
0
has exceptional I/O performance because, as data is written to or read from the disk drive array in response to a group, or an ensemble of I/O requests, each disk drive, or spindle in the array comes into play to satisfy the I/O requests. Optimal I/O performance is realized in systems that use RAID level
0
, because each disk drive, or spindle in the array comes into play to satisfy the ensemble of I/O requests.
However, RAID level
0
is redundant in name only, and offers no fault tolerance. If RAID level
0
were fault tolerant, the techniques typically used to provide fault tolerance would slow down the I/O performance typically available through the use of RAID level
0
. Because RAID level
0
is not fault tolerant, it is not a viable solution in systems that require reliability.
Fault tolerance in case of disk drive failure is typically provided by a number of different techniques. These techniques include disk drive mirroring and data mirroring. Disk drive mirroring involves duplicating an original datum that is stored on a first disk drive, and storing the duplicate datum on a second disk drive. RAID levels
1
and
0
+1 use disk drive mirroring to provide fault tolerance to a data storage subsystem. Disk drive mirroring also provides one-hundred percent redundancy of data that virtually eliminates RAID system interruption due to a single disk drive failure.
There are a number of problems with data striping techniques (RAID levels) that use disk drive mirroring to increase fault tolerance. One problem is that disk drive mirroring sacrifices I/O performance for fault tolerance. For example, consider that in a data storage subsystem implemented with either RAID level
1
or RAID level
0
+1, only one-half of the disk drives are used to satisfy any read request from a host computer. The disk drives that are used to satisfy a read data request are the disk drives have original data stored on them. (The other one-half of the disk drives only come into play only if a primary disk drive fails, wherein the duplicate data is used to satisfy the read request). As noted above, optimal I/O performance is only realized if each disk drive, or spindle in the array comes into play to satisfy the I/O request. Therefore, RAID levels that use disk drive mirroring are not viable solutions for systems that require fast response to read data requests.
RAID level
6
data striping techniques use data mirroring, as compared to disk drive mirroring. Data mirroring also means that each original data is mirrored across the disk drives. However, using data mirroring, original data is typically not mirrored on a dedicated mirror disk drive, as is done in RAID levels that use disk drive mirroring. This means that it is possible to distribute the data across the disk drives in a manner that provides optimal read data request performance.
To illustrate data mirroring according to RAID level
6
, refer to Table 1, where there are shown aspects of RAID level
6
data striping techniques according to the state of the art.
TABLE 1
Example of RAID-6
Drive 1
Drive 2
Drive 3
A
B
C
stripe 0
C’
A’
B’
stripe 1
D
E
F
stripe 2
F’
D’
E’
stripe 3
The first three vertical columns represent disk drives
1
-
3
and are respectively labeled “Drive 1”, “Drive 2”, and “Drive 3”. Horizontal rows, stripes
0
-
3
, represent “stripes of data,” where original and duplicate data are respectively distributed across the disk drives
1
-
3
in the disk drive
1
-
3
array. Original data is stored on disk drives
1
-
3
respectively in data segments A, B, C, D, E, and F. Mirrored data, or duplicate data are respectively stored on disk drives
1
-
3
in data segments A′, B′, C′, D′, E′, and F′. For example, data segment A′ contains a duplicate of the original data contained in data segment A, B′ contains a duplicate of the original data contained in B, C′ contains a duplicate of the original data contained in C, and the like.
Stripe
0
includes original data in data segments A-C, and stripe
1
contains respective duplicates of original data in data segments A′-C′. Stripe
2
includes original data in data segments D-F, and stripe
3
contains respective duplicates of original data in data segments D′-F′. As can be seen, RAID level
6
stores duplicate data in data segments A′-F′ on different disk drives
1
-
3
than the corresponding original data in data segments A-F. To accomplish this, the RAID level
6
data striping algorithm will rotate to the right by one data segment, a copy of the original data in each respective data segment in the immediately proceeding stripe.
This rotation to the right by one data segment before writing the duplicate data introduces an undesirable amount of rotational delay into a data storage subsystem that uses RAID level
6
. Such rotational delay slows down the data storage subsystem performance in response to sequential write data requests. To understand why this is the case, it is helpful to understand how a write data request is handled by a disk drive
1
-
3
.
Each disk drive
1
-
3
is organized into a plurality of platters, each of which has two recordable disk surfaces. (Individual platters and disk surfaces are not shown) Each platter surface is divided into concentric circles called “tracks”. Each track is in turn divided into a p
Flehr Hohbach Test Albritton & Herbert LLP
Peikari B. James
Samodovitz Arthur J.
LandOfFree
Method and apparatus for distributing data across multiple... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for distributing data across multiple..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for distributing data across multiple... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2955753