U.S. patent application number 12/204673 was filed with the patent office on 2009-01-01 for apparatus and method to check data integrity when handling data.
This patent application is currently assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION. Invention is credited to STEVEN T. BROADBENT, MICHAEL H. HARTUNG, CARL E. JONES, KARL A. NIELSEN, JEREMY M. PINSON.
Application Number | 20090006904 12/204673 |
Document ID | / |
Family ID | 36460838 |
Filed Date | 2009-01-01 |
United States Patent
Application |
20090006904 |
Kind Code |
A1 |
BROADBENT; STEVEN T. ; et
al. |
January 1, 2009 |
APPARATUS AND METHOD TO CHECK DATA INTEGRITY WHEN HANDLING DATA
Abstract
An apparatus and method to check integrity when handling data.
The method provides a storage array which includes a plurality of
sectors. The method defines (N) data state identifiers and (N)
parity state identifiers. The method receives a command to handle
data, where that command designates a target sector. The method
determines the data state identifier assigned to the target sector,
determines the parity state identifier assigned to the target
sector, and compares the data state identifier and the parity state
identifier. If the method determines that the data state identifier
and the parity state identifier are the same, the method performs
the command to handle data. Alternatively, if the method determines
that the data state identifier and the parity state identifier
differ, the method generates an error message.
Inventors: |
BROADBENT; STEVEN T.;
(TUCSON, AZ) ; HARTUNG; MICHAEL H.; (TUCSON,
AZ) ; JONES; CARL E.; (TUCSON, AZ) ; NIELSEN;
KARL A.; (TUCSON, AZ) ; PINSON; JEREMY M.;
(TUCSON, AZ) |
Correspondence
Address: |
DALE F. REGELMAN;QUARLES & BRADY, LLP
ONE SOUTH CHURCH AVENUE, STE. 1700
TUCSON
AZ
85701-1621
US
|
Assignee: |
INTERNATIONAL BUSINESS MACHINES
CORPORATION
ARMONK
NY
|
Family ID: |
36460838 |
Appl. No.: |
12/204673 |
Filed: |
September 4, 2008 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10995873 |
Nov 23, 2004 |
7454686 |
|
|
12204673 |
|
|
|
|
Current U.S.
Class: |
714/49 ;
714/E11.025; 714/E11.207 |
Current CPC
Class: |
G06F 11/1076
20130101 |
Class at
Publication: |
714/49 ;
714/E11.025 |
International
Class: |
G06F 11/07 20060101
G06F011/07 |
Claims
1. An article of manufacture comprising an information storage
array which includes a plurality of sectors and an information
storage medium having computer readable program code stored therein
to check data integrity when handling data, the computer readable
program code comprising a series of computer readable program steps
to effect: receiving a command to handle data, wherein said command
designates a target sector, and wherein said target sector
comprises one of said plurality of sectors; determining the data
state identifier assigned to said target sector; determining the
parity state identifier assigned to said target sector; comparing
said data state identifier and said parity state identifier; when
said data state identifier and said parity state identifier are the
same, performing said command; when said data state identifier and
said parity state identifier differ, generating an error
message.
2. The article of manufacture of claim 1, said computer readable
program code further comprising a series of computer readable
program steps to effect: receiving a WRITE command to write data to
a target sector; receiving said data; when said data state
identifier and said parity state identifier are the same:
incrementing said data state identifier for said target sector,
wherein said incremented data state identifier comprises one of (N)
predetermined data state identifiers, wherein N is greater than or
equal to 1; calculating a new parity for said target sector;
writing said data and said incremented data state identifier to
said target sector; incrementing said parity state identifier,
wherein said incremented parity state identifier comprises one of
(N) predetermined parity state identifiers; writing said new parity
and said incremented parity state identifier to said information
storage array.
3. The article of manufacture of claim 2, wherein said data state
identifier comprises a set of two checkbits.
4. The article of manufacture of claim 2, wherein said data state
identifier comprises a set of more than two checkbits.
5. The article of manufacture of claim 2, wherein said target
sector comprises a metadata portion and a data portion, said
computer readable program code further comprising a series of
computer readable program steps to effect: writing said data to
said data portion; and writing said incremented data state
identifier to said metadata portion.
6. The article of manufacture of claim 5, wherein said information
storage array comprises (R) storage medium, and wherein metadata
portion comprises (R) Data State Addresses, said computer readable
program code further comprising a series of computer readable
program steps to effect writing said incremented data state
identifier to one of said (R) data state addresses.
7. The article of manufacture of claim 2, wherein said information
storage array comprises (R) storage media and a plurality of parity
blocks, wherein one or more of said plurality of parity blocks are
written to each of said (R) storage media, said computer readable
program code further comprising a series of computer readable
program steps to effect writing said new parity and said
incremented parity state identifier to one or more of said
plurality of parity blocks.
8. The article of manufacture of claim 2, wherein said information
storage array comprises (R) storage media, said computer readable
program code further comprising a series of computer readable
program steps to effect writing said data and said incremented data
state identifier to (R-1) of said (R) storage media.
9. The article of manufacture of claim 8, said computer readable
program code further comprising a series of computer readable
program steps to effect writing said new parity and said
incremented parity state identifier to the remaining information
storage medium.
10. The article of manufacture of claim 1, said computer readable
program code further comprising a series of computer readable
program steps to effect: receiving a READ command designating a
target sector; operative if said data state identifier and said
parity state identifier are the same, providing information read
from said target sector.
11. A computer program product encoded in an information storage
medium and usable with a programmable computer processor to check
data integrity when handling data using an information storage and
retrieval system comprising (N) predetermined data state
identifiers, (N) predetermined parity state identifiers, and an
information storage array comprising a plurality of sectors, said
computer program product comprising: computer readable program code
which causes said programmable computer processor to receive a
command to handle data, wherein said command designates a target
sector, and wherein said target sector comprises one of said
plurality of sectors; computer readable program code which causes
said programmable computer processor to determine the data state
identifier assigned to said target sector, wherein said data state
identifier comprises one of said (N) data state identifiers,
wherein N is greater than or equal to 1; computer readable program
code which causes said programmable computer processor to determine
the parity state identifier assigned to said target sector, wherein
said parity state identifier comprises one of said (N) parity state
identifiers; computer readable program code which causes said
programmable computer processor to compare said data state
identifier and said parity state identifier; computer readable
program code which, if said data state identifier and said parity
state identifier are the same, causes said programmable computer
processor to perform said command; computer readable program code
which, if said data state identifier and said parity state
identifier differ, causes said programmable computer processor to
generate an error message.
12. The computer program product of claim 11, wherein said computer
readable program code to receive a command further comprises
computer readable program code which causes said programmable
computer processor to receive a WRITE command to write data to a
target sector, further comprising: computer readable program code
which causes said programmable computer processor to receive said
data; computer readable program code which, if said data state
identifier and said parity state identifier are the same, causes
said programmable computer processor to: increment said data state
identifier for said target sector, wherein said incremented data
state identifier comprises one of said (N) data state identifiers;
calculate a new parity for said target sector; write said data and
said incremented data state identifier to said target sector;
increment said parity state identifier, wherein said incremented
parity state identifier comprises one of said (N) parity state
identifiers; write said new parity and said incremented parity
state identifier to said information storage array.
13. The computer program product of claim 11, wherein said target
sector comprises a data portion and a metadata portion, further
comprising: computer readable program code which causes said
programmable computer processor to write said data to said data
portion; and computer readable program code which causes said
programmable computer processor to write said incremented data
state identifier to said metadata portion.
14. The computer program product of claim 13, further comprising
computer readable program code which causes said programmable
computer processor to generate an incremented data state identifier
comprising a set of two checkbits.
15. The computer program product of claim 13, further comprising
computer readable program code which causes said programmable
computer processor to generate an incremented data state identifier
comprising a set of more than two checkbits.
16. The computer program product of claim 13, wherein said
information storage array comprises (R) storage media, and wherein
said metadata portion comprises (R) data state addresses, further
comprising computer readable program code which causes said
programmable computer processor to write said incremented data
state identifier to one of said (R) data state addresses.
17. The computer program product of claim 12, wherein said
information storage array comprises (R) information storage media
and a plurality of parity blocks written to each of said (R)
information storage media, further comprising computer readable
program code which causes said programmable computer processor to
write said new parity and said incremented parity state identifier
to one or more of said plurality of parity blocks.
18. The computer program product of claim 12, wherein said
information storage array comprises (R) information storage media,
further comprising computer readable program code which causes said
programmable computer processor to write said data and said
incremented data state identifier to (R-1) of said (R) information
storage media.
19. The computer program product of claim 18, further comprising
computer readable program code which causes said programmable
computer processor to write said new parity and said incremented
parity state identifier to the remaining information storage
medium.
20. The computer program product of claim 11, further comprising:
computer readable program code which causes said programmable
computer processor to receive a READ command designating a target
sector; computer readable program code which, if said data state
identifier and said parity state identifier are the same, causes
said programmable computer processor to provide information read
from said target sector.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] This application is a Continuation application claiming
priority from the Application having Ser. No. 10/995,873 filed Nov.
23, 2004.
FIELD OF THE INVENTION
[0002] The invention relates to an apparatus and method to check
data integrity when handling data. More specifically, the invention
relates to an apparatus and method to check data states and parity
states when handling data. In certain embodiments, the invention
relates to an apparatus and method to check data states and parity
states when writing data. In certain embodiments, the invention
relates to an apparatus and method to check data states and parity
states when reading data.
BACKGROUND OF THE INVENTION
[0003] Information storage and retrieval systems are used to store
information provided by one or more host computer systems. Such
information storage and retrieval systems receive requests to write
information to one or more information storage devices, and
requests to read information from those one or more information
storage devices. The system is continuously moving information to
and from storage devices.
[0004] Data storage devices, such as hard disk drives, can "drop
write," i.e. fail to write data to a designated sector of the
storage medium, and/or perform off-track writes to that designated
sector. A subsequent read of this sector would receive old, i.e.
"stale," data instead of the data last written. In a RAID-5 encoded
array, if a subsequent write to the parity stripe that causes the
invalidly written sector to be read in calculating the new parity,
the parity will become corrupt with the previous data because the
read used to update the parity will get stale data instead of the
data that was used when creating the parity.
[0005] In order to identify such dropped writes/off-track writes,
the prior art teaches reading back data after every write. Such a
read back verifies that the data is written correctly. Such read
backs necessarily require, however, an additional read operation
for every write operation. The prior art also teaches regularly
checking the parity of the entire storage array. This approach is
cumbersome and requires an inordinate amount of time. Moreover,
while such a parity check is in progress, parity corruption can
occur at a different part of the storage medium.
[0006] What is needed is a method to identify dropped writes and/or
off-track writes without reading back every write, and without
checking the parity of the entire storage array. Applicants' method
identifies dropped writes and off-track writes by checking the data
state and the parity state of a sector when handling data.
SUMMARY OF THE INVENTION
[0007] Applicants' invention includes an apparatus and method to
check data states and parity states when handling data. The method
provides an information storage array which includes an information
storage array comprising a plurality of sectors. The method defines
(N) data state identifiers and (N) parity state identifiers. The
method receives a command to handle data, where that command
designates a target sector.
[0008] The method determines the data state identifier assigned to
the target sector, determines the parity state identifier assigned
to the target sector, and compares the data state identifier and
the parity state identifier.
[0009] If the method determines that the data state identifier and
the parity state identifier are the same, the method performs the
command to handle data. Alternatively, if the method determines
that the data state identifier and the parity state identifier
differ, the method generates an error message.
BRIEF DESCRIPTION OF THE DRAWINGS
[0010] The invention will be better understood from a reading of
the following detailed description taken in conjunction with the
drawings in which like reference designators are used to designate
like elements, and in which:
[0011] FIG. 1 is a block diagram of one embodiment of Applicants'
information storage and retrieval system;
[0012] FIG. 2 is a block diagram showing four data states/parity
states;
[0013] FIG. 3A is a block diagram showing a data sector comprising
a data portion and a metadata portion;
[0014] FIG. 3B is a block diagram of the metadata portion of FIG.
3A comprising a plurality of data state addresses;
[0015] FIG. 4A is a block diagram showing a Data State Identifier
maintained at a first State Address for information written to a
first storage medium;
[0016] FIG. 4B is a block diagram showing a Data State Identifier
maintained at a second State Address for information written to a
second storage medium;
[0017] FIG. 4C is a block diagram showing a Data State Identifier
maintained at a third State Address for information written to a
third storage medium;
[0018] FIG. 4D is a block diagram showing a Data State Identifier
maintained at a first State Address on the (m)th sector written to
a first storage medium and a corresponding Parity State Identifier
maintained at a first State Address on the (m)th sector written to
a third storage medium;
[0019] FIG. 4E is a block diagram showing a Data State Identifier
maintained at a second State Address on the (m)th sector written to
a second storage medium and a corresponding Parity State Identifier
maintained at a second State Address on the (m)th sector written to
a third storage medium;
[0020] FIG. 4F is a block diagram showing a Data State Identifier
maintained at a third State Address on the (m)th sector written to
a third storage medium and a corresponding Parity State Identifier
maintained at a third State Address on the (m)th sector written to
a first storage medium;
[0021] FIG. 5 is a flow chart summarizing the steps of Applicants'
method to write data;
[0022] FIG. 6 recites an algorithm encoding the method of FIG. 5;
and
[0023] FIG. 7 is a flow chart summarizing the steps of Applicants'
method to read data.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0024] Referring to the illustrations, like numerals correspond to
like parts depicted in the Figures. The invention will be described
as embodied in a data storage and retrieval system comprising two
or more clusters, where each of those clusters includes one or more
processors, one or more data caches, and one or more non-volatile
storage devices. The following description of Applicants' method to
assign data states and parity states when writing data is not
meant, however, to limit Applicants' invention to data processing
systems in general, or to data processing system which include a
plurality of clusters, as the invention herein can be applied to
writing data to a plurality of storage media in general.
[0025] In the illustrated embodiment of FIG. 1, Applicants'
information storage and retrieval system 100 includes a first
cluster 101A and a second cluster 101B. Each cluster includes a
processor portion 130/140 and an input/output portion 160/170,
respectively. Internal PCI buses in each cluster are connected via
a Remote I/O bridge 155/165 between the processor portions 130/140
and device I/O portions 160/170, respectively.
[0026] In the illustrated embodiment of FIG. 1, system 100 is
capable of communicating with host computer 125 via communication
link 127. In certain embodiments, communication link 127 is
selected from a serial interconnection, such as RS-232 or RS-422,
an ethernet interconnection, a SCSI interconnection, a Fibre
Channel interconnection, an ESCON interconnection, a FICON
interconnection, a Local Area Network (LAN), a private Wide Area
Network (WAN), a public wide area network, Storage Area Network
(SAN), Transmission Control Protocol/Internet Protocol (TCP/IP),
the Internet, and combinations thereof.
[0027] Information storage and retrieval system 100 further
includes a plurality of host adapters 102-105, 107-110, 112-115,
and 117-120, disposed in four host bays 101, 106, 111, and 116.
Each host adapter may comprise one or more Fibre Channel ports, one
or more FICON ports, one or more ESCON ports, or one or more SCSI
ports. Each host adapter is connected to both clusters through one
or more Common Platform Interconnect bus 121 such that each cluster
can handle I/O from any host adapter.
[0028] Processor portion 130 includes processor 132 and cache 134.
In certain embodiments, processor portion 130 further include
memory 133. In certain embodiments, memory device 133 comprises
random access memory. In certain embodiments, memory device 133
comprises non-volatile memory.
[0029] Processor portion 140 includes processor 142 and cache 144.
In certain embodiments, processor portion 140 further include
memory 143. In certain embodiments, memory device 143 comprises
random access memory. In certain embodiments, memory device 143
comprises non-volatile memory.
[0030] I/O portion 160 includes non-volatile storage ("NVS") 162
and NVS batteries 164. I/O portion 170 includes NVS 172 and NVS
batteries 174.
[0031] I/O portion 160 further comprises a plurality of device
adapters, such as device adapters 165, 166, 167, and 168, and
information storage media organized into two arrays, namely array
"A" and array "B". The illustrated embodiment of FIG. 1 shows two
information storage arrays. In other embodiments, Applicants'
information storage and retrieval system includes more than two
information storage arrays. Each such information storage array
appears to a host computer as one or more logical devices.
[0032] In certain embodiments, Applicants' information storage
media comprise magnetic media, such as for example hard disks
disposed in individual hard disk drive units. In certain
embodiments, Applicants' information storage media comprise optical
media, such as for example CDs, DVDs, and the like. In certain
embodiments, Applicants' information storage media comprise
electronic storage media, such as PROMs, EPROMS, EEPROMs, Flash
PROMs, compactflash, smartmedia, and the like.
[0033] In certain embodiments, arrays "A" and "B" utilize a RAID
protocol. In certain embodiments, arrays "A" and "B" comprise what
is sometimes called a JBOD array, i.e. "Just a Bunch Of Disks"
where the array is not configured according to RAID. As those
skilled in the art will appreciate, a RAID (Redundant Array of
Independent Disks) rank comprises independent information storage
media configured in an array to obtain performance, capacity and
reliability that exceeds that of a single large storage medium.
[0034] In the certain embodiments, array "A" includes hard disk
drives 181, 182, 183, 191, 192, 193, and 194, and array "B"
includes hard disk drives 185, 186, 187, 188, 196, 197, and
198.
[0035] Applicants' information storage and retrieval system, such
as system 100 (FIG. 1), receives new data from one or more host
computers, such as for example host computer 125 and writes that
new data to one or more information storage arrays, such as for
example information storage array 180 (FIG. 1). The illustrated
embodiment of FIG. 1 shows system 100 interconnected with one host
computer. In other embodiments, Applicants' information storage and
retrieval system is capable of communicating with a plurality of
host computers.
[0036] Applicants' information storage and retrieval system
receives new data, and writes that new data to one or more
designated sectors on one or more information storage media. A
number of failure modes exist wherein that new data may not be
encoded to the storage media. For example, data storage devices
sometimes fail to properly write data to the information storage
medium. In addition, data storage devices sometimes perform
offtrack writes to a sector. Regardless of the failure mode, a
subsequent attempt to read the new data instead returns old data.
Using a RAID 5 protocol, if a single sector write is unsuccessful a
subsequent write to the parity stripe that causes the invalidly
written sector to be read in calculating the new parity, the parity
will become corrupt with the previous data because the read used to
update the parity will get stale data instead of the data that was
used when creating the parity.
[0037] In certain embodiments, Applicants' method writes
information to a designated sector using a RAID 5 protocol. A RAID
(Redundant Array of Independent Disks) combines two protocols:
parallelism and redundancy. A RAID uses multiple information
storage media in parallel to provide much higher bandwidth than a
single storage medium. A RAID can also perform multiple operations
in parallel. Redundancy in the form of parity is used to maintain
reliability. By storing parity, data can be fully recovered after a
single storage device failure.
[0038] Using a RAID 5 protocol, data is striped across multiple
storage media as blocks. With (R) storage media, a group of (R-1)
data blocks is striped across (R-1) media. A parity block is
computed by exclusive-oring ("XORing") these (R-1) blocks, and the
parity block is stored on the remaining storage medium. Parity is
distributed meaning that successive parity blocks are stored on
different storage media to avoid the bottleneck of a dedicated
parity medium. Each individual block is called a stripe unit, and a
collection of (R-1) data blocks and a parity block is called a
parity stripe.
[0039] For peak efficiency, a full parity stripe should be written
at once. In this case, the data and the parity can be written out
in parallel. If only part of the parity stripe is modified, parity
must be recomputed from the data already on disk. The old data and
parity must be read, and then the new data and the recomputed
parity are written.
[0040] Each information storage medium in Applicants' one or more
storage arrays comprises a plurality of sectors, such as for
example sector 300 (FIG. 3A), where each of those sectors includes
an information portion 340 comprising about 512 bytes and a
metadata portion 350 comprising about 12 bytes.
[0041] In a storage system comprising (R) storage media,
Applicants' method maintains (R) State Addresses for data written
to sector 300, where (R) is greater than or equal to 2. A Data
State Identifier is maintained in one of those (R) State
Addresses.
[0042] As a general matter, Applicants' method writes both the data
and the corresponding Data State Identifier to the same sector. In
the event sector 300 is written to (i)th storage medium, then
Applicants' method utilizes the (i)th State Address to store the
Data State Identifier for the data written to sector 300.
[0043] As an example and referring to FIG. 4A, in this illustrated
embodiment Applicants' storage system includes 3 storage media,
namely storage media 410, 420, and 430. In the illustrated
embodiment of FIG. 4A, sector 300 is written to first storage
medium 410, and therefore, Applicants' method maintains a Data
State Identifier for the data of sector 300 using the first State
Address, namely State Address 310. In this illustrated embodiment,
that Data State Identifier is "01."
[0044] In the illustrated embodiment of FIG. 4B, sector 300 is
written to second storage medium 420, and therefore, Applicants'
method maintains a Data State Identifier for the data of sector 300
using the second State Address, namely State Address 320. In this
illustrated embodiment, that Data State Identifier is "00."
[0045] In the illustrated embodiment of FIG. 4C, sector 300 is
written to third storage medium 430, and therefore, Applicants'
method maintains a Data State Identifier for the data of sector 300
using the third State Address, namely State Address 330. In this
illustrated embodiment, that Data State Identifier is "02."
[0046] As those skilled in the art will appreciate, an information
storage medium may comprise many thousands, or more, of sectors.
Using Applicants' method, the data state for any one of those
sectors can be determined by examining the Data State Identifier
written one of the State Addresses disposed in the metadata portion
of that sector.
[0047] Referring to FIG. 2, in certain embodiments Applicants'
method defines 4 different data states, namely State0, State1,
State2, and State3. In the illustrated embodiment of FIG. 2, State0
is assigned identifier "00," State1 is assigned identifier "01",
State2 is assigned identifier "10," and State3 is assigned
identifier "11".
[0048] Applicants' method maintains a corresponding parity sector
for each information sector. As a general matter, if Applicants'
method writes information to the (m)th sector on the (i)th storage
medium, then Applicants' method maintains a Data State Identifier
for that information at the (i)th State Address portion of the
(m)th sector on (i)th storage medium, and maintains a Parity State
Identifier for the information at the (i)th State Address portion
of the (m)th sector on the (j)th storage medium, where (i) and (j)
differ. As a further general matter, where a parity sector is
written to the (m)th sector on the (j)th storage medium, then the
(j)th State Address portion of that (m)th sector on the (j)th
storage medium will not be used.
[0049] For example in the illustrated embodiment of FIG. 4D,
information has been written to sector 300, i.e. the "m(th)"
sector, on the (i)th storage medium 410, where (i) is 1. This being
the case, Applicants' method maintains a Data State Identifier at
the (i)th State Address portion 310 of that (m)th sector of the
(i)th storage medium. In the illustrated embodiment of FIG. 4D,
that Data State Identifier is "01." Applicants' method maintains a
corresponding (m)th parity sector on the (j)th storage medium. In
the illustrated embodiment of FIG. 4D, (j) is 3, i.e. parity sector
300 is written to the third storage medium 430. Applicants' method
further maintains a Parity State Identifier in the (i)th State
Address portion, i.e. State Address 310 portion of parity sector
300 on the third storage medium. In the illustrated embodiment of
FIG. 4D, that Parity State Identifier is "01." Thus in the
illustrated embodiment of FIG. 4D, the Data State Identifier and
the Parity State Identifier are the same for the information
written to sector 300 on storage medium 410.
[0050] In the illustrated embodiment of FIG. 4E, information has
been written to sector 300, i.e. the "m(th)" sector, on the (i)th
storage medium 420, where (i) is 2. This being the case,
Applicants' method maintains a Data State Identifier at the (i)th
State Address portion 320 of that (m)th sector of the (i)th storage
medium. In the illustrated embodiment of FIG. 4E, that Data State
Identifier is "00." Applicants' method maintains a corresponding
(m)th parity sector on the (j)th storage medium. In the illustrated
embodiment of FIG. 4E, (j) is 3, i.e. parity sector 300 is written
to the third storage medium 430. Applicants' method further
maintains a Parity State Identifier in the (i)th State Address
portion, i.e. State Address 320 portion, of sector 300 on the third
storage medium. In the illustrated embodiment of FIG. 4E, that
Parity State Identifier is "00." Thus in the illustrated embodiment
of FIG. 4E, the Data State Identifier and the Parity State
Identifier are the same for the information written to sector 300
on storage medium 410.
[0051] In the illustrated embodiment of FIG. 4F, information has
been written to sector 300, i.e. the "m(th)" sector, on the (i)th
storage medium 430, where (i) is 3. This being the case,
Applicants' method maintains a Data State Identifier at the (i)th
State Address portion 330 of that (m)th sector of the (i)th storage
medium. In the illustrated embodiment of FIG. 4F, that Data State
Identifier is "02." Applicants' method maintains a corresponding
(m)th parity sector on the (j)th storage medium. In the illustrated
embodiment of FIG. 4F, (j) is 1 i.e. parity sector 300 is written
to the first storage medium 410. Applicants' method further
maintains a Parity State Identifier in the (i)th State Address
portion, i.e. State Address 330 portion of sector 300 on the first
storage medium. In the illustrated embodiment of FIG. 4F, that
Parity State Identifier is "00." Thus in the illustrated embodiment
of FIG. 4F, the Data State Identifier and the Parity State
Identifier differ indicating that a dropped write has occurred.
[0052] Referring again to FIG. 2, in certain embodiments
Applicants' method defines 4 different parity states, namely
State0, State1, State2, and State3. In the illustrated embodiment
of FIG. 2, State0 is assigned identifier "00," State1 is assigned
identifier "01", State2 is assigned identifier "10," and State3 is
assigned identifier "11" Using Applicants' method, each time data
is written to sector 300, the Data State Identifier and the Parity
State Identifier for sector 300 are incremented. As a general
matter, Applicants' method includes defining (N) states.
Incrementing the (i)th Data State/Parity State for sector 300
includes assigning to sector 300 the (i+1)th Data State and the
(i+1)th Parity State, unless (i) equals (N-1). If (i) equals (N-1),
then incrementing the (i)th Data State/Parity State means setting
(i) equal to 0. For example where (N) is 4, State0 is incremented
to State1, State1 is incremented to State2, State2 is incremented
to State3, and State3 is incremented to State0.
[0053] Applicants' invention includes a method to check data
integrity when handling data. The steps of an embodiment of that
method to write data are summarized in FIG. 5. Referring now to
FIG. 5, in step 510 Applicants' method defines (N) First Data and
Parity States and (M) Second Data and Parity States. The (N) First
States are used when performing less Can a full array wide WRITE.
The (M) Second States are used when performing a full array wide
WRITE.
[0054] In certain embodiments, step 510 includes defining 4 First
States. In certain embodiments, those 4 First States correspond to
States 210 (FIG. 2), 230 (FIG. 2), 250 (FIG. 2), and 270 (FIG. 2),
where those 4 First States are assigned identifiers 220, 240, 260,
and 280, respectively. In other embodiments, Applicants' method
defines more than 4 States. In yet other embodiments, Applicants'
method defines 2 States. In yet other embodiments, Applicants'
method defines 3 States.
[0055] In certain embodiments, the (N) First States of step 510 are
defined by the manufacturer of the information storage and
retrieval system, such as for example system 100 (FIG. 1). In other
embodiments, the N) First States of step 510 are defined by the
owner and/or operator of the information storage and retrieval
system. In yet other embodiments, the (N) First States of step 510
are defined by a host computer, such as for example host computer
125 (FIG. 1).
[0056] In step 520, Applicants' information storage and retrieval
system receives data and a WRITE command to write that data to a
target sector, i.e. a designated Logical Block Address. In certain
embodiments, step 520 is performed by a processor disposed in
Applicants' information storage and retrieval system, such as for
example processor 132 (FIG. 1).
[0057] In step 530, Applicants' method determines if the WRITE
command of step 520 includes a full array wide WRITE, i.e. a full
parity stripe. In certain embodiments, step 530 is performed by a
processor disposed in Applicants' information storage and retrieval
system, such as for example processor 132 (FIG. 1). If Applicants'
method determines in step 530 that the WRITE command of step 520
comprises a full array wide WRITE, then the method transitions from
step 530 to step 532 wherein the method randomly assigns one of the
(M) Second States as both the Data State Identifier and the Parity
State Identifier. In certain embodiments, step 532 is performed by
a processor disposed in Applicants' information storage and
retrieval system, such as for example processor 132 (FIG. 1).
[0058] Applicants' method transitions from step 532 to step 534
wherein the method writes the data received in step 520, and writes
the Data State Identifier assigned in step 532 to the designated
information storage array. In certain embodiments, step 534 is
performed by a processor disposed in Applicants' information
storage and retrieval system, such as for example processor 132
(FIG. 1).
[0059] In step 536, Applicants' method writes the Parity State
Identifier assigned in step 532 to the corresponding parity block
disposed in Applicants' information storage array. In certain
embodiments, step 536 is performed by a processor disposed in
Applicants' information storage and retrieval system, such as for
example processor 132 (FIG. 1).
[0060] If Applicants' method determines in step 530 that the WRITE
command of step 520 does not comprises a full array wide WRITE,
then the method transitions from step 530 to step 540 wherein the
method reads the last assigned Data State Identifier for each
target sector. References herein to writing "a target sector" or
"the target sector" include writes to more than one target sector
where that write operation comprises less than a full stride write.
In certain embodiments, step 540 is performed by a processor
disposed in Applicants' information storage and retrieval system,
such as for example processor 132 (FIG. 1).
[0061] Applicant's' method transitions from step 540 to step 550
wherein the method reads the last assigned Parity State Identifier
for each target sector. In certain embodiments, step 550 is
performed by a processor disposed in Applicants' information
storage and retrieval system, such as for example processor 132
(FIG. 1).
[0062] Applicants' method transitions from step 550 to step 560
wherein the method compares the Data State Identifier read in step
540 for each target sector with the Parity State Identifier read in
step 550 for each target sector. In certain embodiments, step 560
is performed by a processor disposed in Applicants' information
storage and retrieval system, such as for example processor 132
(FIG. 1).
[0063] If Applicants' method determines in step 560 that the Data
State Identifier and the Parity State Identifier differ for any of
the target sectors, then a dropped write has been detected. For
example, in the illustrated embodiment of FIG. 4F the Data State
Identifier and the Parity State Identifier differ. If the Data
State Identifier and the Parity State Identifier differ, then
Applicants' method transitions from step 560 to step 565 wherein
the method generates an error log. In certain embodiments, step 565
is performed by a processor disposed in Applicants' information
storage and retrieval system, such as for example processor 132
(FIG. 1).
[0064] Alternatively, if Applicants' method determines in step 560
that the Data State Identifier and the Parity State Identifier for
each target sector are the same, then Applicants' method
transitions from step 560 to step 570. For example, in the
illustrated embodiment of FIG. 4D the Data State Identifier and the
Parity State Identifier are the same.
[0065] In step 570, Applicants' method increments the Data State
Identifier for each target sector, and saves that incremented Data
State Identifier(s). In certain embodiments, step 570 is performed
by a processor disposed in Applicants' information storage and
retrieval system, such as for example processor 132 (FIG. 1).
[0066] In step 580, Applicants' method calculates a new parity for
each target sector, and increments the Parity State Identifier for
each target sector. In certain embodiments, step 580 is performed
by a processor disposed in Applicants' information storage and
retrieval system, such as for example processor 132 (FIG. 1).
[0067] In step 590, Applicants' method writes the data received in
step 520 to the data portion of the target sector(s), and writes
the incremented Data State Identifier for each target sector to the
metadata portion(s) of the target sector(s). In certain
embodiments, step 590 is performed by a processor disposed in
Applicants' information storage and retrieval system, such as for
example processor 132 (FIG. 1).
[0068] In step 595, Applicants' method writes the new parity for
each target sector and the incremented Parity State Identifier for
each target sector to the parity block for each target sector. In
certain embodiments, step 595 is performed by a processor disposed
in Applicants' information storage and retrieval system, such as
for example processor 132 (FIG. 1).
[0069] Referring now to FIGS. 5 and 6, in certain embodiments
Applicants' method is encoded using algorithm 600. Step 610
comprises elements of step 540. Step 620 comprises elements of step
550. Step 630 comprises elements of step 560. Steps 640 and 650
comprise elements of step 570. Step 660 comprises elements of step
580. Step 670 comprises elements of step 590. Step 680 comprises
elements of step 595.
[0070] Applicants' method further includes embodiments to read
data. Referring now to FIG. 7, in step 710 Applicants' information
storage and retrieval system receives a READ command to read data
from a target sector, i.e. a designated Logical Block Address.
References herein to reading "a target sector" or "the target
sector" include reads of more than one target sector. In certain
embodiments, step 710 is performed by a processor disposed in
Applicants' information storage and retrieval system, such as for
example processor 132 (FIG. 1).
[0071] In step 720, Applicants' method reads the last assigned Data
State Identifier for each target sector. In certain embodiments,
step 720 is performed by a processor disposed in Applicants'
information storage and retrieval system, such as for example
processor 132 (FIG. 1).
[0072] Applicant's' method transitions from step 720 to step 730
wherein the method reads the last assigned Parity State Identifier
for each target sector. In certain embodiments, step 730 is
performed by a processor disposed in Applicants' information
storage and retrieval system, such as for example processor 132
(FIG. 1).
[0073] Applicants' method transitions from step 730 to step 740
wherein the method compares the Data State Identifier for each
target sector with the Parity State Identifier for each target
sector. In certain embodiments, step 740 is performed by a
processor disposed in Applicants' information storage and retrieval
system, such as for example processor 132 (FIG. 1).
[0074] If Applicants' method determines in step 740 that the Data
State Identifier and the Parity State Identifier for any target
sector differs, then a dropped write has been detected. If the Data
State Identifier and the Parity State Identifier for any target
sector differs, then Applicants' method transitions from step 740
to step 750 wherein the method generates an error log. In certain
embodiments, step 740 is performed by a processor disposed in
Applicants' information storage and retrieval system, such as for
example processor 132 (FIG. 1).
[0075] Alternatively, if Applicants' method determines in step 740
that the Data State Identifier and the Parity State Identifier are
the same for all target sectors, then Applicants' method
transitions from step 740 to step 760 wherein the method provides
the requested information. In certain embodiments, step 760 is
performed by a processor disposed in Applicants' information
storage and retrieval system, such as for example processor 132
(FIG. 1).
[0076] The method of FIG. 5, or the method of FIG. 6, or the method
of FIG. 7, may be implemented separately. In certain embodiments,
individual steps recited in FIG. 5 and/or FIG. 6, and/or FIG. 7,
may be combined, eliminated, or reordered.
[0077] In certain embodiments, Applicants' invention includes
instructions residing in memory 133 (FIG. 1), where those
instructions are executed by processor 132 (FIG. 1) to performs
steps 520, 530, 532, 534, 536, 540, 550, 560, 565, 570, 580, 590,
and 595, recited in FIG. 5, or steps 610 through 680 recited in
FIG. 6, and/or steps 710 through 760 recited in FIG. 7. In other
embodiments, Applicants' invention includes instructions residing
in any other computer program product, where those instructions are
executed by a computer external to, or internal to, system 100, to
perform steps 520, 530, 532, 534, 536, 5640, 550, 560, 565, 570,
580, 590, and 595, recited in FIG. 5, or steps 610 through 680
recited in FIG. 6, and/or steps 710 through 760 recited in FIG. 7.
In either case, the instructions may be encoded in an information
storage medium comprising, for example, a magnetic information
storage medium, an optical information storage medium, an
electronic information storage medium, and the like. By "electronic
storage media," Applicants mean for example, a device such as a
PROM, EPROM, EEPROM, Flash PROM, compactflash, smartmedia, and the
like.
[0078] While the preferred embodiments of the present invention
have been illustrated in detail, it should be apparent that
modifications and adaptations to those embodiments may occur to one
skilled in the art without departing from the scope of the present
invention as set forth in the following claims.
* * * * *