JP2008250777A - Medium checking load reducing device in array system, its method, and program therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To quickly perform the periodical medium check of a defective sector that occurs in a magnetic disk device and to reduce the load required for a matching check. <P>SOLUTION: A defective sector where an error occurs in the magnetic disk device is stored in an abnormality place table for storing error occurrence places. The matching check is performed by narrowing down only to the neighborhood of the defective sector. About a part with no defective sector occurring, a ReadVerify command that can execute a periodical medium check with a lower load is used to thereby perform the medium check. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to reducing the load on an array system during a medium check. More specifically, the present invention performs a periodic medium check for a bad sector generated in a magnetic disk device at a lower load, and The present invention relates to reducing the time for checking a medium.

  In order to always maintain reliability and redundancy as an array system, it is effective to periodically check the consistency of the medium of the magnetic disk device.

  In this regard, as a means for checking the consistency of the medium of the conventional magnetic disk device, a Read command is used, and a full consistency check is performed, which is a consistency check for all surfaces of the storage medium of the magnetic disk device. It was carried out.

The overall consistency check by the Read command is a data check that reads all the data recorded on all magnetic disks that make up the array system and checks whether the integrity of the array system is maintained. (See, for example, Patent Document 1).
JP 2000-330732 A

  However, when the full consistency check is performed by the Read command, there are the following problems.

  The first problem is that a recent magnetic disk has a large capacity, and if all data is read, the consistency check itself takes time.

  The second problem is that the performance of the entire system is reduced by the load necessary for the consistency check while the consistency check is performed.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a medium check load reducing apparatus, an array method, and a program thereof in an array system capable of quickly performing a consistency check and reducing a load necessary for the consistency check. And

  According to the present invention, there is provided a medium check load reducing device in an array system for checking the consistency of an array disk, and for detecting an error location that has occurred in the array disk, and for detecting the detected error location. A medium check load comprising: error location information holding means for holding position information; and consistency check means for checking consistency of the array disk based on the location information of the error location held. A mitigation device is provided.

  In the above-described medium check load reducing device, the error location finding means may be a medium check load reducing device that operates using a Read Verify command.

  According to the present invention, it is possible to quickly perform a periodic medium check for defective sectors occurring in a magnetic disk device.

  Next, the best mode of the present invention will be described with reference to the drawings.

[Embodiment 1]
The configuration of the embodiment of the present invention will be described with reference to FIG. The present embodiment is an array system using three SCSI (Small Computer System Interface) disks.
This is an example.

  In order to implement the present invention, the disk standard is not limited to SCSI, and the number of disks may be a plurality other than three.

  First, referring to FIG. 1, the array system 1 includes an array disk 2 and a host unit 3.

  The array disk 2 includes a magnetic disk device 21, a magnetic disk device 22, and a magnetic disk device 23.

  The host unit 3 includes a consistency check unit 31 for performing a consistency check focusing on the vicinity of a defective sector that has occurred during operation, a medium check unit 32 for periodically performing a media check using a Read Verify command, and an error occurrence. An abnormal location table 33 for storing locations and a load monitoring unit 34 for monitoring the load on the array disk 2 are provided.

  Here, the operation of each unit included in the host unit 3 will be briefly described.

  The load monitoring unit 34 monitors the load on the array disk 2.

  When there is no load on the array disk 2, that is, when it is determined that there is no problem in the operation of the array disk system even if the consistency check is performed, the load monitoring unit 34 matches the consistency check unit 31. Request that a sex check be performed.

  When there is a request from the load monitoring unit 34, the consistency check unit 31 refers to the abnormal part table 33, and matches the sector in the vicinity of the part where the error has occurred recorded in the abnormal part table 33. Perform a sex check.

  Here, the neighborhood can be exemplified by 5,000 sectors before and after the error-occurring sector as a range. Specifically, a total of 10,000 sectors including 5,000 sectors before the error occurrence sector and 5,000 sectors after the error occurrence sector are targeted.

  Considering this point from the viewpoint of storage capacity, a magnetic disk device having a storage capacity of 512 bytes per sector is in the range of about 5 megabytes.

  Since 10,000 sectors are merely examples, any number of sectors can be selected in accordance with the performance of the disk array system, the number of error occurrence locations, the nature of the magnetic disk device, and the like.

  The medium check unit 32 periodically checks the entire surface of the medium of each magnetic disk device for any sector in which an error has occurred.

  When an error occurrence location is found, the error occurrence location is recorded in the abnormal location table 33.

  The data content is not checked. Therefore, it is not necessary to read data into the host unit once. That is, there is no need for data transfer.

  Therefore, the Read Verify command, which is a command for finding errors, is used.

  The Read Verify command is defined as a standard by ANSI (American National Standard Institute).

  The abnormal part table 33 has a function of recording position information of an error occurrence part when an error occurs in the magnetic disk device 21, 22 or 23 during operation of the system.

  When the consistency check unit 31 performs the consistency check, the error occurrence location recorded in the abnormal location table 33 is referred to.

  Next, the operation of this embodiment will be described in detail with reference to FIG. 1 and FIG.

  First, the operation of recording the location of the sector where the abnormality has occurred in the abnormal location table 33 will be described below.

  When an error occurs in the magnetic disk device 21, 22, or 23, the magnetic disk device 21, 22, or 23 notifies the host unit 3 that an error has occurred along with the error occurrence location.

  Upon receiving the error occurrence notification from the magnetic disk device 21, 22, or 23, the host unit 3 records the error occurrence location information in the abnormal location table 33.

  Next, an operation for performing a consistency check by the consistency check unit 31 will be described below.

  The load monitoring unit 34 monitors the load on the array disk 2 and, when detecting that there is no load, requests the consistency check unit 31 to perform a consistency check.

  Upon receiving the request, the consistency check unit 31 refers to the abnormal part table 33 and performs a consistency check on a sector in the vicinity of the recorded error part.

  If no error location is recorded in the abnormal location table 33, the consistency check is not performed.

  After completion of the consistency check, the error location corresponding to the error location for which the vicinity has been checked is deleted from the abnormal location table 33.

  Next, the operation of performing the entire medium check by Read Verify by the medium check unit 32 and the operation of the conventional consistency check will be described with reference to FIG.

  First, the consistency check by the conventional Read command will be described (see the flowchart on the right side of FIG. 3, A11, A12, A13, and A14).

  In the consistency check by the conventional Read command, first, the host unit issues a Read command to the magnetic disk device (A11).

  Next, the magnetic disk device reads data on the storage medium (A12).

  Note that the reading range at this time has a limitation that reading can be performed only in units corresponding to the stripe length of the array.

  Then, the magnetic disk device transfers the read data to the host unit (A13).

  The host unit checks data consistency based on the transferred data (A14).

  Next, the entire medium check in the present invention will be described (see the left flowchart in FIG. 3, A21, A22 and A23).

  In the overall medium check in the embodiment of the present invention, first, the medium check unit 32 issues a Read Verify command to the magnetic disk device 21, 22 or 23 (A21).

  In the entire medium check by Read Verify in the present invention, no parity comparison is performed at this time.

  Therefore, unlike the prior art, the reading range is not limited by the stripe length.

  Next, each of the magnetic disk devices 21, 22 or 23 reads the data on the recording medium (A22), and responds to the medium check unit 32 only with the presence or absence of an error, not the content of the data (A23).

  As described above, there are the following two differences when comparing the entire medium check in the present invention with the conventional consistency check.

  The first difference is that the full medium check in the present invention can issue a Read command regardless of the stripe length (FIG. 3, B1). This is because, unlike the prior art, no parity comparison is performed.

  Next, the second difference is that data on the recording medium is not transferred to the host unit (FIG. 3, B2).

  Furthermore, the third difference is that no parity comparison is performed (similarly, FIG. 3, B2).

  Due to the second and third differences, the load on the system can be reduced and the check time can be shortened.

  Also, only when the magnetic disk unit is not loaded, by performing a consistency check on the location where an error has occurred, the consistency of the array disk can be improved without hindering the essential purpose of operating as an array disk. The purpose of keeping can also be achieved.

  Furthermore, by combining with a full medium check by periodic Read Verify, it is possible to shorten the check time and reduce the system load while maintaining the same function as the conventional consistency check.

[Embodiment 2]
As another embodiment of the present invention, an embodiment in which the functions of the present invention are realized as a program on an OS will be described in detail with reference to FIGS.

  First, referring to FIG. The array system 4 in this embodiment includes an array disk 5, a host unit 6, and an array disk check program 7.

  The array disk 5 includes a magnetic disk device 51, a magnetic disk device 52, and a magnetic disk device 53.

  Reference is now made to FIG. The array disk check program 7 is software on the OS including the functions of the consistency check unit 71, the medium check unit 72, the abnormal part table 73, and the load monitoring unit 74.

  The array disk check program 7 monitors whether or not an error has occurred in the disk device 51, 52 or 53 and records the location where the error has occurred via the host unit 6.

  In addition, the load on the array disk 5 is monitored, and when there is no load, the host unit 6 is requested to perform a consistency check for the vicinity of the recorded error location.

  In addition, the host unit 6 is requested to periodically perform a full medium check by Read Verify.

  The host unit 6 performs a consistency check and a full medium check by Read Verify in accordance with a request from the array disk check program 7.

  The medium check load reduction system according to the present invention can be realized by hardware, software, or a combination thereof.

  Next, the effect of this embodiment will be described.

  The first effect is that the time required for the consistency check can be shortened.

  Causes of read errors on the disk medium include medium scratches that occur when the head contacts the medium over a long period of operation, or the medium cannot be written normally due to vibration or shock, and the corresponding part is read. In some cases, an error occurs and the surface appears.

  For this reason, there is a tendency that a plurality of read errors occur in a sector in the vicinity where the read error has occurred.

  Therefore, the range of the consistency check can be narrowed down to the nearby sector where the read error occurs.

  The range of consistency check is narrowed down, and the entire medium check time is shortened while maintaining the function of the conventional consistency check by performing the entire medium check by Read Verify, which requires less time for the entire medium check. Is possible.

  The second effect is that the performance degradation of the disk array system can be suppressed.

  First, it is because the absolute workload is reduced by narrowing the scope of the consistency check. Next, by performing the consistency check while monitoring the system load, an excessive load is not applied to the system. Further, this is because data transfer and parity comparison do not occur in the entire medium check by Read Verify.

It is a figure showing the basic composition of the present invention. 6 is a flowchart showing a comparison between a full medium check of the present invention and a consistency check by a Read command which is a conventional technique. It is a figure showing the structure of embodiment which implement | achieved this invention as the program on OS. It is a figure which shows the structure of the program of embodiment which implement | achieved this invention as the program on OS.

Explanation of symbols

1, 4 Array system 2, 5 Array disk 3, 6 Host unit 7 Array disk check program 21, 22, 23, 51, 52, 53 Magnetic disk unit 31, 71 Consistency check unit 32, 72 Medium check unit 33, 73 Abnormal point table 34, 74 Load monitoring unit

Claims (11)

A medium check load reducing device in an array system for checking the consistency of an array disk,
An error location finding means for finding an error location that occurred in the array disk,
Error location information holding means for holding location information of the found error location;
Consistency check means for performing a consistency check of the array disk based on the positional information of the error location to be held,
A medium check load reducing device comprising: The medium check load reducing device according to claim 1,
The apparatus for reducing a load on a medium check, wherein the error location finding means operates using a Read Verify command. The medium check load reducing device according to claim 1 or 2,
An apparatus for reducing a load on a medium check, further comprising means for acquiring position information of an error location that has occurred in an array disk based on a notification from the array disk. The medium check load reducing device according to any one of claims 1 to 3,
Load monitoring means for monitoring the load on the array disk;
The medium check load reducing apparatus, wherein the consistency check unit operates based on a monitoring result of the load monitoring unit. The medium check load reduction device according to any one of claims 1 to 4,
The medium check load reducing apparatus, wherein the consistency check means operates only in the vicinity of the error location. A method for reducing a medium check load in an array system, which performs an array disk consistency check,
An error location finding step for finding an error location in the array disk;
Holding the location information of the found error location, error location information holding step;
A consistency check step of performing a consistency check of the array disk based on the positional information of the error location to be held;
A medium check load reduction method comprising: The medium check load reduction method according to claim 6,
The medium check load reducing method, wherein the error location finding step operates using a Read Verify command. The medium check load reduction method according to claim 6 or 7,
A method for reducing a medium check load, further comprising the step of acquiring position information of an error location that has occurred in an array disk by notification from the array disk. The medium check load reduction method according to any one of claims 6 to 8,
A load monitoring step for monitoring the load on the array disk;
The medium check load reducing method, wherein the consistency check step operates based on a monitoring result in the load monitoring step. The medium check load reduction method according to any one of claims 6 to 9,
The medium check load reduction method, wherein the consistency check step operates only in the vicinity of the error portion.   A program for causing a computer to function as the medium check load reducing device according to any one of claims 1 to 5.

Images