JPH04188463A - Disk driving apparatus - Google Patents

Abstract

PURPOSE:To make it possible to select a disk drive at a position where the number of drives which can be accessed at one time is not decreased in the arrangement of the disk drives constituting a group even after the switching of a disk drive when a fault occurs in the disk drive by providing a standby drive in a logical drive group. CONSTITUTION:A disk-array controlling device 70 controls a disk array device 80 in accordance with the instruction from a CPU 60. At this time, a logical drive group has a standby drive. When a fault occurs in any drive and the standby drive is present in the same group as the faulty drive, or when the standby drive which is controlled by the same control means as the control means controlling the faulty drive is present, the first mode is started, and the drive is switched. When the standby drive is not present, the second mode is started, and the drive is switched. Such a function is provided. Thus, the performance of each group wherein the standby drive is used by the switching when the fault occurs in the drive can be maintained at the same performance as one before the switching.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a disk drive device using a disk array.

[Prior Art] In recent years, disk arrays in which a large number of small disk drives are arranged have been devised. This is intended to improve data transfer speed and reliability by switching between disk drives.

A technique using this disk array will be explained with reference to the drawings.

FIG. 13 shows an example of a drive group having an array structure.

This device includes hard disk controllers 301 to 30
4 and disk drive 310~312°320~32
2,330 to 332, and 340 to 342, that is, a total of four hard disk controllers and 16 disk drives.

Now, one logical group in the array disk, disk drives 310, 320, 330, 340, disk drives 311, 321, 331, 341, and disk drives 312, 322, 332, 342.
There will be 3 groups of 4 machines each.

Of the disk drives belonging to each logical group, the hard disk controllers 301 and 30
The drives under Hard Disk Controller 303 are assigned as data storage drives, the drives under Hard Disk Controller 303 are assigned as redundant data drives, and the drives under Hard Disk Controller 304 are assigned as spare drives.

Note that the drive group surrounded by point l1Aa indicates one group.

Now, when a failure occurs in the disk drive 310, the disk drive 340, which has been set aside as a spare, is targeted for switching and is switched to the disk drive 310.

After switching, the logical group (hereinafter referred to as group) will be
It is composed of disk drives 320, 330, and 340 (the drive group surrounded by dotted lines). After that, when another failure occurs in the disk drive 320, the disk drive 341 is similarly assigned as a spare.

As a result, the group has disk drives 330.
340 and 341 (drive group surrounded by dotted line C).

Other examples will be shown.

FIG. 14 shows only two groups of drives having an array structure. Group A consists of disk drives 350a to 355a, and group B consists of disk drives 350b to 355b. Disk drives 350a to 353a and disk drives 350b to 353b are assigned as data storage drives. Further, disk drives 354a, 355a, 354b, and 355b are assigned to redundant data drives.

In this case, even if a failure occurs in a disk drive, data will not be lost as long as there are no more than two disk drives in one group.

[Problems to be Solved by the Invention] However, as shown in FIG. 13 above. In conventional technology, when switching disk drives when a disk drive failure occurs, the disk drives that are initially allocated as spares are sequentially used as disk drives to be switched. , there were multiple instances under the same hard disk controller.

As a result, the number of drives that can be accessed decreased at the same time, leading to a decline in the performance of the group.

Specifically speaking in the above case, two disk drive devices 340 and 341 are assigned to − hard disk controllers 304 . Therefore, the disk drives that the group can access at one time are disk drive 330 and disk drive 340.
or 341, that is, only two.

Therefore, one drive can only be accessed sequentially, leading to performance degradation.

In addition, in the example shown in Figure 14, in one group, if even one more disk drive fails in the same group, data cannot be restored, whereas in other groups, all disk drives fail. Even if the drive is normal and two other disk drives fail, it may be possible to restore data. In other words, a situation occurs in which groups with a high risk of data loss and groups with a low risk of data loss coexist, resulting in a problem of lowering the overall reliability.

Specifically, if a failure occurs in disk drive 350a and disk drive 351a, group A
If even one more drive fails, it will be impossible to restore the data. On the other hand, in Group B, data could be restored even if two more drives failed, and there was a difference in reliability among the groups.

Furthermore, even when the performance and reliability required for each group differ, switching is not done to take these into consideration. In other words, groups that require high performance are configured with multiple disk drives under the same hard disk controller, and groups that require high reliability are at high risk of data loss.

It is an object of the present invention to locate disk drives in such a way that even after disk drive switching when a disk drive failure occurs, the arrangement of the disk drives constituting the group is such that the number of drives that can be accessed at one time does not decrease. The objective is to provide a disk drive device to choose from.

Another objective was to distribute the risk of data loss among groups due to disk drive failure by allocating drives from low-risk groups to high-risk groups. , to provide a disk drive device.

Another purpose is to determine whether or not to switch disk drives in the event of a disk drive failure, depending on information specified by the user such as the performance and reliability required for each group, and whether or not to switch disk drives to meet the requirements. An object of the present invention is to provide a disk drive device that selects an optimal switching target.

Another object of the present invention is to provide a disk drive device having a function of setting the trigger for performing a disk drive switching operation for each group.

[Means for Solving the Problems] The present invention has been made to achieve the above object, and its aspects include a disk drive group having a plurality of logical drive groups each consisting of a plurality of drives;
In the disk drive apparatus, the logical drive group has a spare drive, and the control means controls the failed drive to be a spare drive in the same group as the failed drive or the failed drive. a first mode in which the drive is switched to a spare drive controlled by the same control means as the control means controlling the drive, and a failed drive is switched to a spare drive controlled by the same control means as the control means controlling the failed drive; A drive belonging to a group in which there is a spare drive that has not been used is set as a drive to be switched, the drive to be switched is switched to a spare drive in the same group as the drive to be switched, and then the failed drive and the same control means as the failed drive are switched. and a second mode in which the drive to be switched is controlled by a second mode, and when a failure occurs in any drive, a spare drive in the same group as the failed drive, or a control means for controlling the failed drive. If there is a spare drive controlled by the same control means as the controller, the first mode is entered and the drive is switched, but if the spare drive does not exist, the second mode is entered and the drive is switched. A disk drive device is provided.

Another aspect includes a disk drive group having a plurality of logical drive groups each including a plurality of drives;
In the disk drive device comprising a control means for controlling the drives of each group, the control means checks a margin, which is the number of drives until data loss occurs if a drive failure occurs, for each logical group; A disk drive device is provided which is characterized in that a drive belonging to a group with a large margin is switched to belong to a group with a small margin.

In another aspect, in a disk drive device including a plurality of drives and a control means for controlling each drive, the control means has a function of switching the drive to be used, and a function of switching the drive to be used only during a preset time period. A disk drive device is provided which is characterized by having a function of performing the switching.

In each of the above embodiments, the drives constituting the group are logically arranged such that drives belonging to the same logical group are located in the same row, and drives controlled by the same control means are located in the same column. It is preferable that the drives be arranged in a matrix, and the drives may include a data storage drive for registering data and a redundant data drive.

Further, the control means may include one or more of the following parameters.

■The data storage drives that make up each group,
A group table consisting of a physical address of a redundant data drive and a failure flag indicating the state of the drive.

■A spare drive management table consisting of the current number of spare drives, the physical address of the spare drive, and an invalid flag indicating whether the spare drive is unused.

■Physical drive management table that shows the physical address and the attributes of the drive at that address. The physical drive management table contains a failure flag indicating whether the drive is normal or faulty, the group number to which the drive belongs, and whether the drive is a data storage drive, redundant data drive, or spare drive. It consists of the drive identifier shown.

■Required level management table showing the required performance and reliability level for each group. The required level management table is composed of the performance level required for the group, the number of failed drives in the group, and the reliability level required for the group.

■A switching scheduling table that indicates the trigger for switching operations in the event of a drive failure for each group.

■Switching reservation table that allows you to reserve switching operations if switching is not performed immediately in the event of a drive failure. The switching reservation table is set for each group and consists of a failed drive address and a registration flag.

(The following is a blank space) [Operation] When the control means detects a failed drive, it determines whether or not to switch the drive at present. This calculates the group number from the physical drive management table based on the address of the failed drive.

The switching trigger for the group is determined from the switching scheduling table and is determined by comparing it with the current time. At this time, the failure flag in the physical drive management table is also turned on. If it is currently time to perform switching, the switching will be performed immediately, but if it is determined that it is not the time, the address will be registered in the switching reservation table corresponding to the group and the registration flag will be turned on. Registered failed drives are monitored at certain regular intervals, and when the switching time is reached, the registration flag is turned off and switching is performed.This allows switching to be performed according to the trigger set for each group.

When switching, depending on the status of spare drives, etc.
There are three states:

Note that the first state and the second state correspond to the first mode, and the third state corresponds to the second mode.

The first state is a state where a spare drive exists and is located in the same row or column as the address of the failed drive, and the second state is a state where a spare drive exists but is located in the same row or column as the failed drive. This is a state where no spare drive exists in the row or the same column, and the third state is a state where no spare drive exists.

The current state is determined by the following procedure. First, the number of spare drives is determined using the spare drive management table, and if the number of spare drives is O, the third state is determined. If the number of spare drives is other than O, it is in the first or second state. To determine whether it is the first or second drive address, enter the failed drive address (1
2, m), if there is a spare drive address with the invalid flag off in the spare drive management table that has Q in the row or m in the column, it will be in the first state, otherwise it will be in the second state. It is judged that the condition is as follows.

The first state switching operation involves finding the data storage drive and redundant data drive of the group in the group table from the group number of the failed drive, and restoring the data of the failed drive from the data of those drives. , copy to the spare drive. Then, the failed drive address in the group table is changed to a spare drive address, the valid flag of the spare drive is turned off, and the number of spare drives is decreased by 1. Furthermore, in the physical drive management table, copy the group number and data identifier of the failed drive to the replacement drive.

In the second state, simply switching to the spare drive will cause the performance of the group to deteriorate after switching. Here, the performance level required for the group is determined based on the required level management table, and if the performance requirement is low, the drive is directly switched to the spare drive. The switching operation is similar to the first state.

When performance requirements are high, the first step is to search for a drive to switch to. Search for a normal drive that has a spare drive in the same row and is in the same column as the failed drive. This calculates the spare drive address whose valid flag is on in the spare drive management table. This (i
, j), and the address of the failed drive is (Q, m), it is determined from the physical drive management table whether address (X+m) is a normal drive or not, and if it is normal, the drive becomes the switching target. If there is a failure, a primary spare drive is found and the process is repeated until a normal drive that satisfies the above conditions is found. As a result, if there is no drive that satisfies the above conditions, the spare drive is targeted for switching. The switching operation at this time is similar to the switching operation in the first state. The switching operation when there is a drive that satisfies the conditions will be explained. The drive to be switched (hereinafter referred to as C, DRV) is first set as a spare drive (
Copy it to S, DRV below. Restore the data on the failed drive (hereinafter referred to as B, DRV) in the same way as in the first state,
C, Copy to DRV. Next, in the group table, S, DRV7 address is updated to C, DRV7 address, and B, DRV address is updated to C, DR, V address. In the spare drive management table, turn off the valid flag of the relevant S and DRV, and decrement the number of spare drives by 1. Next, copy the group number and drive identifier of S, DRV to the physical drive management table at 2°C, then: B, D.
Copy the group number and drive identifier of R■ to C0DRV. As a result, the performance of each group can be maintained at the same level as before switching by switching using a spare drive when a drive fails. Further, it is possible to provide drive switching suitable for the performance requirements of each group.

The third condition is that the failed drive is unrecoverable, so the risk of data loss is distributed according to the level of reliability required for the group. As a distribution method, increase the number of failed drives in the group by 1 from the request level management table.
and compare it with the confidence level. When the demand for reliability is high and the margin until data loss is small, the drive is more flexible than the group where the margin until data loss is large and the demand for reliability is low. this is.

A group number to be accommodated is determined from the reliability level and the number of failed drives, and from among the group, drives in the same column as the failed drive are determined and made to be accommodated. This is to prevent performance deterioration after accommodation. If there are no drives in the same column, drives in any column are targeted. In the switching operation for accommodation, the data of the failed drive is restored and copied to the 5 switching target drives in the same way as in the first state. Next, enter the address of the failed drive in the group table.
Update the address of the drive to be switched, and turn on the failure flag of the drive to be switched. Next, from the physical drive management table, copy the group number and drive identifier of the failed drive to the drive to be switched.6 In the request level management table,
The number of failed drives in the group that includes the failed drive is decreased by 1, and the number of failed drives in the group that includes the drive to be switched is increased by 1. This makes it possible to distribute the degree of risk in accordance with the reliability required of the group.

(Margin below) [Example] An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows a computer system including a disk drive device to which the present invention is applied.

This computer system has a central processing unit, CPU60.
and a disk drive device mainly composed of a disk array control device 70 and a disk array device 80.

The disk array control device 70 controls the disk array device i80 according to instructions from the CPU 60.

FIG. 2 shows the internal configuration of the disk array control device 70.

The microprocessor unit 74 executes programs stored in the random access memory 75 while sequentially decoding them, thereby controlling the entire disk array control device 70 . The channel control circuit 71 includes the CPU 60
Controls data transfer with. A control circuit 72 controls data transfer with each drive. data buffer 7
3 is used during data transfer between the channel control circuit 71 and the control circuit 72. The redundant data creation circuit 76 adds redundant data to the data sent from the CPU 60. A clock 77 is provided in this subsystem.

When reading, the disk array device f! The data read from each drive of 1f80 is temporarily stored in the data buffer 73, synchronized, and then sent to the channel control circuit 7.
1 to the CPU 60.

FIG. 3 shows the drive configuration that makes up the disk array device 80.

The data transfer control circuits 81 to 87 are for transferring data with the disk array control device 70. Each data transfer control circuit 81-87 has four drives 81a-81-d, 82a-82d, . . . , 87.
a to 87 (l is connected.

The drive groups under the data transfer control circuits 81, 82, 83 and 84 are used as data storage drives, and the drive groups under the data transfer control circuits 85 and 86 are used as redundant data drives. The subordinate drive group is assigned as a spare drive.

This group of drives is also a data recovery group (hereinafter referred to as
There are four sets of drives 81a to 87a, drives 81b to 87b, drives 810 to 87c, and drives 81d to 87d. Note that from now on, each group will be referred to as groups Gb, Gc, and Gd in order from the one closest to the data transfer control circuits 81 to 87.

In addition, regarding the addressing of each drive, the drive arrangement is considered as a matrix, with the row direction being 0 to 69 and the column direction being O to 3,
Let (0, O) to (3, 6) be addresses.

The microprocessor unit 74 of this embodiment is as follows.

It is configured to take into consideration the reliability required for each group and drive, select the most suitable drive that satisfies the requirements, and perform drive switching. Specifically, the random access memory 75 or the like is provided with various tables to be described later in which various requests, conditions, etc. are input in advance, and the drive to be switched is selected while referring to this table.

This table will be explained using FIGS. 4 to 9.

The group table 90 shown in FIG. 4 has address information of the drives that constitute each group.

In this table, row 100 corresponds to group Ga, row 101 to group Gb, row 102 to group Gc, and row 103 to group Gd. Then, corresponding to each group, data storage drive address information 91-94.
Drive address information 95,96 for redundant data is provided.

Each drive address information 91 to 94 includes a failure flag 104 indicating that the data is inaccessible due to a drive failure, and a row address 1 of the drive.
05 and a column address 106.

Next, FIG. 5 shows a spare drive management table 110.

This spare drive management table 110 shows address information of current spare drives. This includes a valid flag 111 indicating that the spare drive is not in use;
Row address 112 and column address 11 of the spare drive
It consists of 3. It also has a spare drive number 114 indicating the current number of spare drives.

FIG. 6 shows the physical drive management table 120.

This physical drive management table 120 shows the management information of the drive at the address shown by the matrix (I, J) consisting of the numbers ■ and J shown on the top and left side of the table in the figure.
This is to indicate that the drive in question is a failed drive.

For each drive, there is a failure flag 121, an identifier 122 indicating whether the drive is a data storage drive, a redundant data drive or a spare drive, and a group number 1 to which the drive belongs.
It has 23.

FIG. 7 shows a request level management table 130.

The requirement level management table 130 shows the performance and reliability requirement level of each group, and in this table, row]7
31 is group Ga, row 132 is group ab, row 13
3, the group table 134 corresponds to group Gd.

Each group also has a performance level 136 required for the group and a number 137 of failed drives within the group.
, the reliability level 138 required for the group.

The performance level 136 is defined as "0" as "low" and "l" as "high". When the performance level 136 is high, switching due to a disk failure is assumed to be performed to a drive that does not cause performance deterioration even after switching.

When low, it shall simply switch to the backup drive.

In addition, the reliability level 138 indicates "0" as "low" and "I
J is defined as "high". And reliability level 1
For a group with a high reliability level of 138, if the number of failed drives in the group becomes two, the group with a lower reliability level of 138 is switched over and accommodated. Groups with a low reliability level of 138 are also switched under the above conditions, but the drives to be switched are of a reliability level of 13.
Select from the lowest group of 8.

FIG. 8 shows a switching scheduling table 140. This table is used to specify when to perform drive switching for each group.

In this switching scheduling table 140, row 141
is in group Ga, and row 142 is in group Gb.
3 corresponds to group Gc, and row 144 corresponds to group Gd. In addition, the data for each group is the parameter 14
5, a start time 146, and a final time 147.

If the value of the parameter 145 is O, drive switching is executed immediately. When the value of parameter 145 is 1, drive switching is performed only in the time period specified by start time 146 and final time 147. Further, when the value of the parameter 145 is 2, execution is performed only on days of the week indicated by the start time 146 to the final time 147. If the value of parameter 145 is 3,
It is executed only on the day specified by the start time 146 and the final time 147.

FIG. 9 shows the switching reservation table 150.

In the switching reservation table 150, the value of the parameter 145 is O.
This is for making a reservation for switching in the time period specified by the start time 146 and final time 147 of the above-mentioned switching scheduling table 140 for a failed drive that does not perform drive switching immediately, that is, for a failed drive that does not perform drive switching immediately. It is.

Reservation is performed by registering the row address 152 and column address 153 of the failed drive and turning on the registration flag 151.

Furthermore, when the switching of the relevant drive is completed, the registration of the relevant drive is deleted by turning off the registration flag 151.

Next, the search operation for a drive to be switched when drive (x, y) fails is shown in Figures 10 (a), (b), (e), (
This will be explained using the flowchart shown in d).

First, it is determined to which group the failed drive (x, y) belongs using the physical drive management table 120 (step 200). Next, the switching execution time for the group is determined from the switching scheduling table 140 (step 20
1).

As a result, it is determined in step 20 whether the parameter 145 is 0 or not, that is, whether or not the switching is to be executed immediately.
2) When there is no need for immediate switching, the switching is reserved by the processing of steps 220 and 221 shown in FIG. 10(d).

In step 220, the microprocessor unit 74 searches for an area in the switching reservation table 150 where the registration flag 151 is off. Next, the registration flag 151 of the area is turned on and the address of the failed drive is registered (step 221).

Regarding the failed drive for which the reservation has been made, the clock 77 will thereafter, for example, at certain regular intervals, judge whether the current time is between the start time 146 and the final time 147, and when the switching time has been reached, , the switching is performed according to the logic from step 203 onwards.

In step 202, if the switching is to be performed immediately, the spare drive management table 110 is searched to determine whether there is a spare drive whose valid flag 111 is on, that is, there is a spare drive that can be used (step 202). 2o3).

As a result, if there is at least one usable spare drive, then among these available spare drives,
It is determined in steps 204 and 205 whether there is a drive located at a position where performance does not deteriorate.

In step 204, the same line as the failed drive (x, )
It is determined whether or not a spare drive exists in the spare drive management table 110 by searching the row address 112 of the spare drive management table 110. If the drive exists, the process proceeds to step 213, where the searched drive is set as the drive to be switched and switched.

If there are no spare drives in the same row, then in the same column (
y) whether there is a spare drive.

This determination is made by searching the column address 113 of the spare drive management table 110 (step 205). If the drive exists, the process advances to step 212, and the searched drive is set as the drive to be switched and switched.

If there is no spare drive in either the X row or the y column, simply switching to the spare drive cannot prevent performance degradation.

Therefore, in order to know the performance required of the group, the performance level is determined using the performance level 136 of the required level management table 130 (step 206), and the value is confirmed (step 207).

As a result, when the performance level 136 is 0, that is, when the required performance is low, any one of the existing spare drives is set as a drive to be switched, and this is switched (step 211). On the other hand, when the performance level 136 is 1, that is, when the required performance is high, the process proceeds to step 208.

By the way, in order to prevent performance deterioration, it is necessary not to reduce the number of parallel drives. Therefore.

It is preferable to select the drive to be switched from the same column (y) as the failed drive. However, at this time, care must be taken to ensure that the performance of the group containing the selected drive does not deteriorate. Therefore, in step 208,
We are currently searching for a drive that meets these conditions.

Specifically, for drives in the y column of the physical drive management table 120, in other words, in the same column as the failed drive, whether or not there is a spare drive in the same row is determined using the identifier 122 of the physical drive management table 120. Use it to judge. For example, if the failed drive is at (2,1) and the spare drive is at (6,3), then (2,3) is the drive that satisfies the above conditions.

When a drive is selected in step 208, the process advances to step 209.

In step 209, first, the drive determined in step 208 is switched to a spare drive existing in the same row as the drive. Then, the failed drive and the drive determined in step 208 are switched (step 2)
).

In this way, by first switching the other drive in the same column as the failed drive to the spare drive in the same row as the other drive, and then switching the failed drive to the other drive, the same drive as the failed drive can be replaced. It becomes possible to switch to a column drive.

By adopting this method, it is possible to provide performance maintenance in the event of drive failure to groups that require high performance.

Note that the actual switching operation will be explained from step 230 onwards.

Next, the case where it is determined that there is no spare drive in step 203 will be explained using FIG. 10(c).

In step 214, 1 is added to the value of the number of failed drives 137 in the request level management table 130. As a result, the number of failed drives 137 indicates the risk of data loss in the group after the failure occurs.

In this embodiment, each group has two redundant data drives, so the margin until data loss is two failures. Therefore, in step 215, the number of failed drives is determined, and if the number of failed drives is two or less, it is determined that the risk of data loss is low, and the operation is immediately terminated.

On the other hand, when there are two failed drives, that is, there is no margin, there is a very high risk of data loss, so drives to be switched are selected so as to distribute the risk throughout the disk array device 80 (step 216).

Note that the optimization logic for drive selection in step 216 will be explained from step 240 onward in FIG. 12, which will be described later.

Subsequently, the failed drive is replaced with the drive selected in step 216 (step 217).

As a result, the group determined in step 200, ie, the group to which the failed drive belongs, has one drive removed from the group to which the drive selected in step 216 belongs. Therefore, step 21
In step 8, the group to which the failed drive belongs decreases the value of the number of failed drives 137 by 1, while the group on the accommodating side increases the value of the number of failed drives 137 by 1 (step 219). Then, the operation ends.

Next, the drive switching operation will be explained using the flowchart shown in FIG.

This operation is performed in steps 209.210.211.212
．． 213.217.

Note that the following description will be made regarding the operation when switching between drive A whose address is (x, y) and drive B whose address is (x' + y').

There are three types of combinations of drives that perform this operation: a normal drive and a spare drive, a failed drive and a spare drive, and a failed drive and a normal drive.

After starting the operation, the microprocessor unit 74 first selects the drive A and drive B that are the targets of the switching operation.
It is determined whether any of the drives is a failed drive (step 230). If any of them is a failed drive, first turn on the failure flag 121 of the failed drive in the physical drive management table 120, and then turn on the failure flag 104 corresponding to the failed drive in the group table 90 (step 231). Thereafter, the process proceeds to step 232.

On the other hand, if it is determined in step 230 that a failed drive is not included, the process directly proceeds to step 232.In step 232, the physical drive management table 120
The identifier 122 and group number 123 of drive A and drive B are exchanged. This completes the switching of the physical address.

Next, it is determined whether or not a spare drive is included (step 233). If it is included, the process proceeds to step 234; if not, the process proceeds to step 238.

In step 234, the spare drive management table 110
The contents of the row address 112 and column address 113 are changed to the normal drive or failed drive address to be switched, and the contents of the failure flag 104 of the group table 90 are also changed to the valid flag 111 of the spare drive management table 110.
Copy to.

Next, it is determined whether the validity flag 111 is 1 (step 235). If the valid flag 111 is 1, that is, if the spare drive and the failed drive have been switched, the process proceeds to step 236) in which the value of the number of spare drives 114 is decreased by 1 (step 236) and step 237. On the other hand, if the valid flag 111 is 0, that is, if the spare drive and the normal drive have been switched, step 237 is performed directly.
Proceed to.

In step 237, the row address 1 of the spare drive is
12, column address 113, row address 105, column address 106 of the drive to be switched in group chifur 90
and complete the switching operation.

On the other hand, if it is determined in step 233 that no spare drive is included, the address information of the drives to be switched is stored in the group table 90, that is, the failure flag 1
04, the row address 1.05 and the column address 106 are exchanged (step 238), and the switching operation is completed.

Through steps 234 to 238 described above, the switching of the address information of the drive addresses constituting one group and the spare drive is completed.

Next, the method for selecting an optimized drive performed in step 216 described above will be explained using the flowchart shown in FIG. 12.

When the number of failed drives in the group reaches 2, the risk of data loss in this group is extremely high because this system only has redundant data for 2 drive corrections.

Therefore, it is first necessary to select a group through which one drive will pass. For this purpose, first, in step 240, a search is made for a group whose reliability level 138 in the required level management table 130 is O, that is, a group whose required reliability is low.

This is to ensure that drives are provided by groups with relatively low reliability requirements.

Next, it is determined whether a group with reliability level O exists (step 241), and if so, step 242)
Proceed to. On the other hand, if it does not exist, the process advances to step 244.

In step 242, a group is searched for where the reliability level 138 is 0 and the number of failed drives is O.

This is determined by whether or not the value of the number of failed drives 137 in the group in question in the request level management table 130 is O.

This allows you to find the group with the lowest risk.

Next, it is determined whether a group that satisfies the condition of step 242 exists (step 243), and if there is,
The group to which the drive will be provided is determined (step 247). However, if it does not exist, proceed to step 244.

In step 241 or step 243, if there is no group that satisfies the conditions required in each step, that is, there is no unreliable group, or even if there is, the group has already failed. If the drive is present, proceed to step 244, as described above.

In step 244, first, the reliability level of the group including the failed drive is determined.

As a result, when the reliability level is low, that is, when the value of the reliability level 138 is 0, it is determined that there is no group to be switched (step 248). Conversely, when the reliability level is high,
In other words, when the value of the reliability level 138 is 1, a group in which the number of failed drives is 0 is found (step 245).
).

Then, a group that satisfies the above conditions, that is, a group that has a high reliability level and the number of failed drives is O,
If so, the process proceeds to step 247 and the group is set as the switching target group.

Conversely, if the group does not exist in step 246, it means that the group to be switched does not exist (step 248).

Through steps 240 to 248 in FIG. 12, the presence or absence of a group to be switched is determined, and if so, at least one group number thereof is determined.

Then, which drive in that group is most suitable for switching is determined at steps 204 to 208 in FIG. 10(a).
The determination can be made by performing the same process as above for the drives in the group.

As explained in the above embodiments, when a drive failure occurs, a switching drive suitable for the performance and reliability required for each group is selected, whether or not there is a spare drive. I can do it. Therefore, it is possible to maintain a balance of performance and reliability of the disk drive device as a whole.

(The following is a blank space) [Effects of the invention] As explained above, according to the present invention, when a drive failure occurs,
If at least one spare drive exists, performance can be maintained even after switching. Furthermore, even when there is no spare drive, reliability is improved because the drive flexibility is done in such a way that the risk of data loss is dispersed among all the drives.

Furthermore, it is possible to specify the time period in which the above processing is to be performed, allowing for flexible responses to user requests. Furthermore, the present invention
Effects such as being able to be achieved without changing the hardware and only using microprograms can be obtained.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a computer system using a disk drive device according to an embodiment of the present invention, FIG. 2 is a block diagram showing the configuration of a disk array control device according to this embodiment, and FIG. An explanatory diagram showing the configuration of a disk array device according to an embodiment of the present invention, FIG. 4 is a configuration diagram of a group table used in this embodiment, and FIG. 5 is a configuration diagram of a spare drive management table used in this embodiment. Figure 6 is a configuration diagram of the physical drive management table used in this embodiment, Figure 7 is a configuration diagram of the request level management table used in this embodiment, and Figure 8 is a diagram of the switching schedule used in this embodiment. Table configuration diagram,
FIG. 9 is a configuration diagram of the switching reservation table used in the present invention,
Fig. 10 is a flowchart showing the search procedure for a drive to be switched, Fig. 11 is a flowchart showing the switching operation procedure, Fig. 12 is a flowchart showing the switching drive search operation to calculate the degree of risk as a fraction, and Fig. 13 is a conventional FIG. 14 is an explanatory diagram showing the concept of a conventional disk array group. 60: central processing unit, 7o: disk array control device,
71: Channel control circuit, 72: Control circuit, 73: Data buffer, 74: Microprocessor unit, 7
5: Random access memory, 76: Redundant data creation circuit, 77: Clock, 80: Disk array device, 81-8
7: Data transfer control circuit, 81 a, b, c, d-87
a, b, c, d Ni drive, 9o Nigel loop table, 91-94: Drive address information for data storage, 9
5: Drive address information for redundant data, 96: Drive address information for redundant data, 100: Line, 101: Line, 102: Line, 1o3: Line, 104: Failure flag, 10
5: Row address, 1o6: Column address. 110: Spare drive management table, 111: Valid flag, 1
12: Row address, 113: Column address, 114: Number of spare drives, 120: Physical drive management table, 121: Failure flag, 122: Identifier, 123 Nigel group number, 1
3o: Requirement level management table, 131: row, 132: row, 1
33: row, 134: row, 136: performance level, 137:
Number of failed drives, 138: Reliability level, 140: Switching scheduling table, 141: Row, 142: Row, 143
: row, 144: row, 145: parameter, 146: start time, 147: final time, 15o: switching reservation table, 151: registration flag, 152: row address, 153
: Column address, 301~3°4 Nihard disk controller, 310~312.320~322, 330~3
32,340-342: Disk drive, 350a
, b ~ 355a, b: Disk drive. Applicant Hitachi Manufacturing Co., Ltd. Representative Patent Attorney Tomi 1) Kazuko Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Spare dry pipe j! Table: Figure 6: Physical Dry Pipe 11 Table: Figure 7: Request Level Management Table: Figure 8: Switching Scheduling Table: Figure 9: Switching Reservation Table: Figure 10 (a) Figure 10 (b) Figure 10 (c) Figure 10 ( d) Figure 11 Figure 12

Claims (1)

[Claims] 1. In a disk drive device comprising a disk drive group having a plurality of logical drive groups each consisting of a plurality of drives, and a control means for controlling the drives in each group, the logical drive group is a spare drive group. a drive, and the control means switches a failed drive to a spare drive in the same group as the failed drive or a spare drive controlled by the same control means as the control means that controls the failed drive. mode and failed drive,
Among the drives controlled by the same control means as the control means that controls the failed drive, a drive belonging to a group in which an unused spare drive exists is set as the drive to be switched, and the drive to be switched is the same as the drive to be switched. Switch with the group's spare drive, then
It has a second mode for switching between a failed drive and a switching target drive controlled by the same control means as the failed drive, and when a failure occurs in any of the drives,
If there is a spare drive in the same group as the failed drive or a spare drive controlled by the same control means that controls the failed drive, the first mode is entered and the drive is switched. A disk drive device having a function of entering a second mode and switching drives when there is no spare drive. 2. In a disk drive device comprising a disk drive group having a plurality of logical drive groups each consisting of a plurality of drives, and a control means for controlling the drives in each group, the control means controls the drive for each logical group. A disk drive device characterized in that when a failure occurs, a margin, which is the number of drives until data loss occurs, is checked, and drives belonging to a group with a large margin are switched to belong to a group with a small margin. . 3. In a disk drive device equipped with a plurality of drives and a control means for controlling each drive, the control means has a function of switching the drive to be used, and performs the switching only during a preset time period. A disk drive device characterized by having functions.