US 7,689,869 B2 · Unit, method and program for detecting imprecise data

Abstract

When data are transferred to a cache from disk drives, through a back end I/F unit, a redundant code verifier of a data path controller compares write-history information (a Destaging Counter (DC) value) embedded in a redundant code of the data with a DC value embedded in a redundant code of a corresponding parity, thereby verifying whether the data are imprecise. In addition, in a case where the data are updated, the redundant code verifier makes a similar verification when old data to be used for creating a parity are read. When the data are not imprecise, a redundant code generator of the data path controller adds to new data and the parity a redundant code into which a new DC value has been embedded. When the data are imprecise, a controller informs a host of an occurrence of a problem through a front end I/F unit.

Drawings

Drawings are temporarily unavailable. Open the PDF at USPTO

Drawings

Claims17

Dependent claims

A method to detect imprecise data, comprising:

partitioning a data sector area into a user data and a parity data wherein, the parity data is associated with the user data and wherein the parity data includes a primary parity data and a secondary parity data;
partitioning the data sector area further to include a user data redundant code area, a primary parity data redundant code area, and a secondary parity data redundant code area;
maintaining a Data Destaging Counter in the user data redundant code area, a Primary Destaging Counter in the primary parity data redundant code area, and a Secondary Destaging Counter in the secondary parity data redundant code area;
adding history information, which is updated in accordance with updating of the Primary Destaging Counter, to the Primary Destaging Counter and to the Secondary Destaging Counter, and to the Data Destaging Counter;
the history information comprising:
a counter value, the counter value being increased by one upon each updating of the Data Destaging Counter;
generating a first user data redundant code based on the user data;
generating a second user data redundant code by applying an XOR operation to the first user data redundant code and the counter value;
storing the second user data redundant code and the counter value in the user data redundant code area;
generating a first primary parity data redundant code based on the primary parity data;
generating a second primary parity data redundant code by applying an XOR operation to the first primary parity redundant code;
storing the second primary parity data redundant code and the counter value in the primary parity data redundant code area;
storing the counter value in the secondary parity data redundant code area;
updating the Data Destaging Counter;
writing the Primary Destaging Counter to a first sector in a memory and writing the Secondary Destaging Counter to a second sector in the memory and writing the Data Destaging Counter to a third sector in the memory;
reading sector data stored in the first sector and sector data stored in the second sector and sector data stored in the third sector;
verifying the Primary Destaging Counter, the Secondary Destaging Counter, and the Data Destaging Counter by comparing the history data added respectively to the read-out sector data;
and verifying a Primary Destaging Counter value associated with the Primary Destaging Counter when the Primary Destaging Counter value coincides with a Data Destaging Counter value, wherein the Data Destaging Counter value is associated with the Data Destaging Counter;
detecting a fault when history information added to parity data stored in the Primary Destaging Counter does not coincide with history information added to the parity data stored in the Secondary Destaging Counter and history information added to the parity data stored in the Secondary Destaging Counter does coincide with the history information added to the user data stored in the Data Destaging Counter, then the history information stored in the Data Destaging Counter is written over the history data stored in the Primary Destaging Counter, and the parity data, stored respectively in the Primary Destaging Counter and the Secondary Destaging Counter and the user data stored in the Data Destaging Counter, are verified by comparing any one of the history information added to the parity data stored in the Primary Destaging Counter and Secondary Destaging Counter with history information added to the user data stored in the Data Destaging Counter;
detecting a double fault when the Secondary Destaging Counter value does not coincide with the Data Destaging Counter value, and the Primary Destaging Counter value does not coincide with the Secondary Destaging Counter;
and
deleting the Primary Destaging Counter value from a Destaging Counter buffer when the Secondary Destaging Counter value does not coincide with the Data Destaging Counter value.

19 more elements

A device for detecting a data error in a memory, comprising:

a storage sub-system controller;
a memory operatively coupled to the storage sub-system controller, wherein:
the memory is partitioned into a user data and a parity data,
the parity data includes a primary parity data and a secondary parity data,
the memory is further partitioned to include a user data redundant code area, a primary parity data redundant code area, and a secondary parity data redundant code area;
a counter value stored in the user data redundant code area wherein the counter value is associated with a Data Destaging Counter;
a first counter value copy stored in the primary parity data redundant code area as a Primary Destaging Counter;
a second counter value copy is stored in the secondary parity redundant code area as a Secondary Destaging Counter;
an addition unit for adding history information to the sector data, which is updated in accordance with updating of the parity data to be stored in the Primary Destaging Counter, and to parity data to be stored in the Secondary Destaging Counter in the memory;
a verification unit for
verifying the parity data stored in the Primary Destaging Counter and the parity data stored in the Secondary Destaging Counter by making a comparison between the history information which has been added to the parity data stored in the Primary Destaging Counter and the history information which has been added to the parity data stored in the Secondary Destaging Counter;
detecting a fault when history information added to parity data stored in the Primary Destaging Counter does not coincide with history information added to parity data stored in the Secondary Destaging Counter, and
detecting a double fault when a Secondary Destaging Counter value associated with the Secondary Destaging Counter does not coincide with a Data Destaging Counter value associated with the Data Destaging Counter, and a Primary Destaging Counter value does not coincide with the Secondary Destaging Counter;
a redundant code verifier generating redundant code upon the transfer of secondary parity data to a data path controller;
and
the storage sub-system controller configured to specify to a host system where in the memory a fault or double fault occurs.

13 more elements

A computer program product comprising a computer useable medium including a computer readable program, wherein the computer readable program, comprises:

a set of instructions for obtaining first history information that is updated in accordance with updating of a Primary Destaging Counter stored in a first parity data sector of a memory;
a set of instructions for obtaining second history information that is updated in accordance with updating of a Secondary Destaging Counter stored in a second parity data sector of the memory;
and
a set of instructions for verifying the Primary Destaging Counter or the Secondary Destaging Counter by comparing the first history information or the second history information with a Data Destaging Counter located outside of the first parity data sector and the second parity data sector;
a set of instructions for detecting a double fault when a Secondary Destaging Counter value does not coincide with a Data Destaging Counter value, and a Primary Destaging Counter value does not coincide with the Secondary Destaging Counter value;
a set of instructions for generating redundant code by a redundant code verifier upon the transfer of a secondary parity data to a data path controller;
and
a set of instructions for informing a host system of the detection of a double fault.

5 more elements

Dependencies are inferred from the claim text.

Description9,069 words

Summary1,437 words

#The exemplary embodiments of this invention relate to a unit for detecting an imprecise data generated, for example, by an error in an address where data are written into a memory device.

#A problem that a storage sub-system returns imprecise data to a host system in reply to READ command causes not only loss of user data but also serious failure such as system failure. For this reason, in general, the storage sub-system avoids the problem that the imprecise data is returned to the host system by adding redundant data such as parity, ECC (Error Correcting Code) and CRC (Cyclic Redundancy Check) code to each data and checking the redundant data upon reading each data.

#For example, a middle range or upper class RAID (Redundant Array of Inexpensive Disks) sub-system may add 8 to 16 bytes of redundant code to a sector of data of 512 bytes, which is a minimum unit of the user data upon writing (destaging) from a cache to a disk drive. Then, this redundant code is checked upon reading (staging) from the disk drive to the cache. In this way, error detection is performed in a data path after a backend interface of a RAID controller. Incidentally, in this case, the redundant code is data calculated uniquely out of the sector data of 512 bytes, and a code based on LRC (Longitudinal Redundancy Check), CRC (Cyclical Redundancy Check) or the like is generally used.

Brief description of the drawings9 figures

#The foregoing and other aspects of these teachings are made more evident in the following Detailed Description of the Preferred Embodiments, when read in conjunction with the attached Drawing Figures, wherein:

#FIG. 1 is diagram showing a first example of arrangement of user data, parity data and DC value according to an embodiment of the present invention.

#FIG. 2 is a diagram showing a second example of arrangement of user data, parity data and DC value according to the embodiment of the present invention.

Detailed description7,394 words

#Hereinafter, the exemplary embodiments of this invention will be described in detail with reference to the accompanying drawings (hereinafter, referred to as “aspect of embodiment”). An aspect of this embodiment is that the present invention is applied to a disk array system which performs data protection by use of parity such as RAID5. In the disk array system like this, a plurality of sector data (hereinafter referred to as “user data”) obtained by partitioning data transmitted from a host system by a sector size and a sector data (hereinafter referred to as “parity data”) to detect imprecise user data and restore it are stored in a plurality of disks in a distributed manner. Moreover, in a sector where the user data are stored (hereinafter referred to as “data sector”), an area where redundant codes to check validity of the user data are stored is concomitantly provided, and in a sector where parity data are stored (hereinafter referred to as “parity sector”), an area where redundant codes to check validity of the parity data are stored is concomitantly provided.

#According to the embodiment of the present invention, by paying attention to such a configuration of the disk array system, write-history information to be updated in conjunction with updating of the user data are provided in a redundant code area of the data sector where the user data are stored, and in a redundant code area of the parity sector where the parity data for the user data are stored. Here, as the write-history information, any kind of information may be adopted, while a counter to be increased by one upon updating of the user data is adopted in the aspect of the embodiment. In this case, since the write-history information can be grasped as the counter to show a number of destaging of the user data, such a counter is referred to as a DC (Destaging Counter), and a value set to the DC is referred to as a DC value hereinafter.

#Next, with reference to FIGS. 1 and 2, arrangement of user data, parity data, DC value and the like in a configuration of RAID assumed in the aspect of the embodiment will be described in detail. Note that in the aspect of the embodiment, for convenience sake, the DC which is provided in the redundant code area of the data sector is referred to as a DDC (Data DC) and a value which is set to the DDC is referred to as a DDC value.

Unit, method and program for detecting imprecise data

Life of this patent

Unit, method and program for detecting imprecise data

Life of this patent

Life of this patent

Abstract

Drawings

Drawings

Claims17

Description9,069 words

Drawings

Classification4 codes

Life of this patent

Description9,069 words