JP4426939B2 - Storage device - Google Patents

Description

The present invention relates to a storage apparatus that constitutes a RAID, and more particularly to a disk array apparatus in which a plurality of drives are used as one unit and a plurality of such units are mounted.

As a method for mounting drives in high density in a space-saving manner, a disk array device has been proposed in which a plurality of physical drives are mounted on one unit (disk blade) and a plurality of these units are mounted (for example, see Patent Document 1). ).

JP-A-9-16343

In the conventional disk array apparatus described above, a plurality of drives in one unit are shown as one drive. However, in such a configuration, since one unit is handled as one drive, a plurality of RAID groups cannot be configured on one unit, and there is a problem that the degree of freedom of configuration is low.

Furthermore, although it is technically possible to configure a plurality of RAID groups on one unit, management of the relationship between the units and the drives in the units has not been taken into account. It is difficult to know how the other RAID groups will be affected by the replacement, and the maintenance and replacement of the unit is difficult.

An object of the present invention is to provide a storage apparatus that can easily replace a unit even if a plurality of drives in the unit are handled separately and a RAID group is freely configured.

The present invention is seen containing a plurality of drive housing unit in which a plurality of drives are mounted detachably is plural detachably mounting a controller housing to the disk control unit is provided, a plurality In the storage apparatus in which the RAID is configured by the disks mounted on the unit, the disk control unit includes a drive status management table that records the status of the drive and a RAID group status management table that records the status of the RAID group. And a background processing number management table indicating whether or not data reconstruction processing by RAID and data replication processing between drives are in progress, and when a failure of the drive is detected, the drive status management table Change the status of the failed drive to “failed” and manage the RAID group status The failure detection unit for changing the status of the RAID group to which the failed drive of the table belongs to “failure”, and the redundancy of the RAID group to which the failed drive belongs is smaller than the set request value of 1 or more and 0 or more And whether or not the redundancy of the RAID group to which the other drive mounted in the unit in which the failed drive is mounted is smaller than the set request value of 1 or more and 0 or more is set. If the redundancy of the RAID group to which the failed drive belongs is smaller than the set request value of 1 or more and 0 or more, the unit other than the unit in which the failed drive is mounted Change the status of the found spare drive to "Rebuild", rebuild the data of the failed drive in the spare drive, and rebuild this During processing, the data reconstruction unit that reflects the status of the spare drive in the background processing number management table, and the redundancy of the RAID group to which the other drive belongs is smaller than one request value that is set, and If it is 0 or more, the status of the spare drive retrieved from the unit other than the unit in which the other drive is mounted is changed to “Evacuated”, and the data of the other drive is copied to the spare drive. During the replication process, the data replication unit reflecting the status of the spare drive in the background processing number management table, and the processing by the data reconstruction unit and the data replication unit based on the background processing number management table are completed. When these processes are completed, the unit in which the failed drive is mounted and the It has a change instruction unit which defective drives notifying that became interchangeable, the.

According to the present invention, a plurality of physical drives mounted on one unit can be handled separately, and a RAID group can be freely formed. Moreover, the removal for every unit,
Exchange becomes easy.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of a disk array device according to an embodiment of the present invention.

The disk array device 1 according to the embodiment of the present invention includes a plurality of drive housings 10a, 10b,
A controller housing 19 is included. The disk array device 1 is a management LAN
5 is connected to the management terminal device 2 via 5. The disk array device 1 is a SAN
4 is connected to a plurality of SAN hosts 3.

The drive housing 10a includes a plurality of units (disk blades) 101, 102, 10
3, 104... Are implemented. Each disk blade has a plurality (4 in this embodiment).
Drive). For example, the disk A 101 includes a disk A 0.
-A3 is mounted. The disk A0 or the like mounted on the drive housing 10 by the disk blade 101 or the like is connected to the connection interface 180, and data can be transmitted to or received from the disk array controller 191 or the like. This connection interface 18
For 0, an interface such as ATA (AT Attachment), SAS (Serial Attached SCSI), or Fiber Channel can be used.

The controller housing 19 is provided with disk array controllers 191 and 192. In the disk array controller 191 or the like, a control program is operating to control data input / output with respect to the disk A0 or the like. Further, it manages a RAID configuration constituted by the disk A0 and the like mounted on the disk array device 1. The controller housing 19 is provided with a plurality of disk array controllers 191 and 192, and the SAN 4
A lot of data can be input and output at the same time.

Depending on the device configuration, a single disk array controller 191 may be provided, or the controller housing 19 and the drive housing 10a may be a single housing.

The management terminal device 2 is a computer device that includes a CPU, a memory, a storage device, an interface, an input device, and a display device. A management program is running on the management terminal device 2. The management program grasps the operating state of the disk array device 1 and controls the operation of the disk array device 1. The management terminal device 2 operates a client program such as a web browser, and the operation of the disk array device 1 is controlled by a management program (Common Gateway Interface, Java (registered trademark), etc.) supplied from the disk array device 1. You may make it control.

The SAN host 3 is a computer device having a CPU, a memory, a storage device, and an interface, and makes it possible to use a database service, a web service, and the like by using data provided from the disk array device 1.

The SAN 4 is a network that can communicate with a protocol suitable for data transfer, such as a fiber channel protocol.

The management LAN 5 can communicate data and control information between computers using, for example, the TCP / IP protocol. For example, Ethernet (registered trademark, the same applies hereinafter) is used.

  FIG. 2 is an external view showing the configuration of the disk array device according to the embodiment of the present invention.

The disk array device 1 of the present embodiment is housed in a 19-inch rack, and a plurality of stages of drive housings (enclosures) containing a plurality of disk blades on which disks are mounted are provided on the top of the disk array device 1. It has been. On the back side of the drive housings 10a to 10i, a power supply unit (not shown) for supplying power to the drives in the housing and a connection interface 180 are housed. The connection interface 180 and each disk blade are physically connected via a wiring board (back plane) provided inside the housing.
As shown in FIG. 4, the disk array controller 191 or the like, the drive housing 10a or the like is connected.

Below the drive housing, a controller housing 19 containing the disk array controllers 191 and 192 and a cooling fan unit for cooling the disk array controller are provided.

FIG. 3 is a perspective view of the drive housing 10a constituting the disk array device according to the embodiment of the present invention.

A plurality of disk blades 101 to 113 can be mounted on the drive housing 10a. The disk blade is accommodated in the drive casing by sliding along a rail provided in the drive casing.

Unit replacement instruction LEDs 121 to 133 are provided on the drive housing 10a above the disk blades 101 to 113. When the disk blade is ready for removal, the unit replacement instruction LED corresponding to the disk blade is lit to notify the user, thereby facilitating maintenance and replacement work.

FIG. 4 shows a disk blade 101 constituting the disk array device according to the embodiment of the present invention.
FIG.

A plurality of drives 2010 to 2040 are attached to the disk blade 101. The drives 2010 to 2040 have a holding frame that can be attached to and detached from the disk blade 101 (
The canister is attached to the disk blade 101 together with the canister. A connector is provided at the rear end of the drives 2010 to 2040. This connector is engaged with a connector 2110 provided on the disk blade 101 to connect between the drives 2010 to 2040 and the disk blade 101.

The disk blade 101 is provided with a drive replacement instruction LED 2210. When the drive (canister) 2010 to 2040 is in a removable state, the drive replacement instruction LED 2210 corresponding to the drive is lit to notify the user, thereby facilitating maintenance replacement work. The drive replacement instruction LED 2210 is driven by the LED control circuit 2300. The LED control circuit 2300 includes disk array controllers 191, 19
2, the drive replacement instruction LED 2210 is turned on.

In addition, the drive replacement instruction LED 2210 indicates that the disk blade 101 is connected to the drive housing 1
Even after being removed from 0a (that is, after the supply of power from the disk array device 1 is interrupted), it can be lit. For this reason, the LED control circuit 2300 is a storage circuit (for example, flip-flop) that maintains instructions from the disk array controllers 191 and 192.
And a power supply circuit that supplies power to be applied to the drive replacement instruction LED 2210 (for example,
Secondary battery, battery). In this way, the drive replacement LED 2210 is lit even after the disk blade 101 is removed from the drive housing 10a, so that the drive to be replaced can be confirmed even after the disk blade 101 is removed from the drive housing 10a.

A connector 2400 is attached to the back end of the disk blade 101, and the disk blade 101 and the backplane are connected by fitting the connector 2400 with a connector provided on the backplane.

The disk blade 101 is provided with a rotary switch 2500. By setting the unit ID using this rotary switch 2500, the disk blades mounted in the same drive housing can be electrically identified. Thus, the rotary switch 2500 functions as a unit ID setting unit.

FIG. 5 is a block diagram showing a configuration of the disk array controller 191 according to the embodiment of this invention.

The disk array controller 191 includes a CPU 1901, a memory 1902, a data transfer controller 1904, a front-end connection interface controller 1905,
A back-end connection interface controller 1906, a data buffer 1907, and a LAN interface controller 1908 are provided.

The memory 1902 stores a control program 1903 (see FIG. 6), and the CPU 1
Various processes are performed by the 901 calling and executing the control program 1903.

The data transfer controller 1904 includes a CPU 1901, a front-end connection interface controller 1905, and a back-end connection interface controller 1906.
And data is transferred between the data buffers 1907.

The front-end connection interface controller 1905 is an interface to the SAN 4 and is, for example, a SAN host 3 according to a fiber channel protocol.
Send and receive data and control signals to and from

The back-end connection interface controller 1906 is connected interface 1
For example, data and control signals are transmitted to and received from the disk via an interface such as ATA, SAS, or fiber channel.

The data buffer 1907 includes a front end connection interface controller 19.
A cache is provided in which data transmitted and received between 05 and the back-end connection interface controller 1906 is temporarily stored.

That is, the data transfer controller 1904 transfers data read / written to / from the disk via the SAN 4 between the interfaces 1905 and 1906. Further, data read / written to / from these disks is transferred to the data buffer 1907.

The LAN connection interface controller 1908 is an interface to the management LAN 5 and can transmit and receive data and control signals to and from the management terminal device 2 by, for example, the Ethernet protocol.

  FIG. 6 is an explanatory diagram of the control program 1903 according to the embodiment of this invention.

RAID group setting program 1911, RAID group management program 1912
, A failure processing program 1913, and an I / O control program 1914. For the operation of these programs, a drive setting holding table 1915, a drive state management table 1916, a RAID group state management table 1917, and a background processing number management table 1918 are stored.

The RAID group setting program 1911 executes a process for setting a RAID by a drive mounted on the disk array device 1 (see, for example, FIG. 13).

The RAID group management program 1912 is a RAID group setting program 191.
The RAID set by 1 is maintained.

The failure processing program 1913 monitors a failure that occurs in a drive installed in the disk array device 1, and performs processing related to the drive in which the failure has occurred and a drive mounted in the same unit as the failed drive (for example, FIG. (Refer FIG. 15, FIG. 16).

The I / O control program 1914 performs processing related to reading data from drives mounted in the disk array device 1 and writing data to these drives.

  Next, various tables used during operation of the control program will be described.

FIG. 7 is an explanatory diagram of a configuration of a RAID generated by a disk mounted on the disk array device 1 according to the embodiment of this invention.

For example, RAID group 1 has a drive A0 (unit ID is A, drive number is 0.
No. drive), drive A1, drive B1, drive B2, and drive C2 constitute a RAID of level 5. In RAID group 4, RAID of level 6 is configured by six drives, drive A3, drive B3, drive C3, drive D3, drive E3, and drive F3.

FIG. 8 is an explanatory diagram of the drive setting content holding table 1915 according to the first embodiment of this invention.

In the drive setting content holding table 1915, the state set for the drive mounted on the disk blade is recorded for each drive. Specifically, the storage capacity of the drive, the use status of the drive (whether “used” or unused “spare”), and the RAID number to which the drive belongs are recorded. Specifically, the RAID number column of five drives, drive A0, drive A1, drive B1, drive B2, and drive C2, is “1”.
”Is recorded, these drives belong to the RAID group 1.

FIG. 9 is an explanatory diagram of the drive status management table 1916 according to the first embodiment of this invention.

The drive status management table 1916 records the current status of the drive mounted on the disk blade for each drive. Specifically, the storage capacity of the drive and the status of the drive (whether “in use” or “failure” has occurred, “rebuild” processing is being executed, “evacuation” processing is being executed, (Unused “spare”) and the RAID number to which the drive belongs. For example, drive B0 is a drive belonging to RAID group 2, but a failure has occurred in drive B0.

FIG. 10 is an explanatory diagram of the RAID group state management table 1917 according to the first embodiment of this invention.

In the RAID group status management table 1917, for each set RAID group,
The total storage capacity, RAID level (which type of RAID is configured), normal redundancy, current redundancy, the identification symbols of the drives constituting this RAID, and the RAID status are recorded.

For example, the drives configuring RAID group 1 are five drives A0, A1, B1, B2, and C2, and these drives are operating normally (that is, RAID group 1). On the other hand, the drives constituting RAID group 2 are B0, C0, D0, E0.
And F0, as shown in the drive status management table 1916 (FIG. 9), a failure has occurred in the drive B0, so this RAID group 2 is in the “failure” state. As a result, the normal redundancy of RAID group 2 is 1, whereas the current redundancy is reduced to 0.

FIG. 11 shows the background processing count management table 1918 according to the first embodiment of this invention.
It is explanatory drawing of.

The background processing count management table 1918 includes a disk blade (unit ID).
) And the number of background processes executed on the disk mounted on the disk blade. That is, as will be described later, the number of background processes is added at the start of the rebuild process when a failure occurs in the disk (S130 in FIG. 14) and subtracted at the end of the rebuild process (S167 in FIG. 15). ), Updated. Further, it is added at the start of the data copy process when a failure occurs in the disk (S141 in FIG. 14), and is subtracted at the end of the data copy process (S176 in FIG. 16) to be updated. In S145 of FIG. 14, the background processing number management table 1918 is also used to determine whether or not a unit may be removed (that is, a unit for which background processing is continuing cannot be removed).

FIG. 12 is an explanatory diagram of a RAID group management screen displayed on the management terminal device 2 according to the first embodiment of this invention.

This RAID group management screen is displayed by selecting a RAID group. The selected RAID group is displayed in a “selected group information” column provided at the lower left of the RAID group management screen. The RAID group management screen displays the status of each disk mounted on each disk blade.

The administrator then selects the drives that make up the RAID group and
ID groups can be configured. The selected drive can also be set as a “spare” that does not constitute a RAID group.

FIG. 13 is a flowchart of the RAID group creation processing according to the first embodiment of this invention, and is executed by the RAID group setting program 1911.

When the RAID group setting program 1911 receives a RAID group creation request from the management terminal device 2 (S101), there is a request for the drive setting content holding table 1915, the drive state management table 1916, and the RAID group state management table 1917. A RAID group is temporarily registered (S102).

After that, referring to the unit first described in the drive setting content holding table 1915 (S103), the number s of drives that are spares in the unit is acquired (
S104). Then, the RAID group n having the largest number of used drives a in the unit is searched (S105). Thereafter, the RAID group state management table 1917 is referenced to obtain the redundancy r of the searched RAID group n (S106).

  Thereafter, the number of drives required for the unit removal process (a−r + 1) is compared with the remaining number of spare drives that can be used in the unit removal process (total number of spare drives−s). It is determined whether or not the number of drives necessary for maintaining redundancy remains as a spare (S107). That is, when a failure occurs on a disk, it is determined whether or not the failure unit can be removed and the failed disk can be replaced.

  If the number of drives required for processing is greater than the remaining number of available spare drives, the spare drive required to maintain the minimum redundancy will be insufficient when the unit is removed. Is determined to be impossible, and a creation error due to a shortage of spare drives is notified (S111). Then, the temporary registration of each management table created in step S102 is deleted (S112).

  On the other hand, if the number of drives required for processing is less than the remaining number of available spare drives, there are enough spare drives to maintain the minimum redundancy even when the unit is removed. Determine that it is possible. Then, it is determined whether or not this investigation has been performed for all units (S108).

As a result, if all the units have not been checked, the process proceeds to step S113, the next unit to be checked is selected, and the influence when the unit is removed is checked (S104).
To S107).

On the other hand, if all the units have been examined, the temporary registration of each management table created in step S102 is formally registered in each table (S109). Then, the management terminal device 2 is notified of the successful creation of the RAID group (S110).

FIG. 14 is a flowchart of unit replacement processing when a drive failure occurs according to the first embodiment of this invention, and is executed by the failure processing program 1913.

When a drive failure is detected (S121), the drive status management table 1916 is referenced to obtain the unit ID (α), drive number k, and assigned RAID group number n of the failed drive (S122). Then, the state of the drive k recorded in the drive state management table 1916 is changed to “failure”, and the state of the RAID group n recorded in the RAID group state management table 1917 is changed to “failure” (S123). And RAI
1 is subtracted from the current redundancy of the RAID group of RAID group n in the D group state management table 1917 (S124).

Whether the current redundancy of the RAID group n is smaller than 1 (that is, 0)
It is determined whether or not (S125).

As a result, if the current redundancy of RAID group n is 1 or more, it is determined that the minimum required redundancy is maintained, and in order to investigate the drive mounted in the same unit as the failed drive, The drive number variable i is initialized (S132). On the other hand, if the current redundancy of RAID group n is less than 1, the process proceeds to step S126.

In step S126, whether the current redundancy of RAID group n is less than 0 (ie,
Whether or not it is a negative number).

As a result, if the current redundancy of RAID group n is smaller than 0, the user is notified that RAID is not configured and data may be lost (S151).

On the other hand, if the current redundancy of the RAID group n is 0 or more, a drive that is a spare from a unit whose unit ID is other than α is searched (S127). If no spare drive is found as a result of this search (S128), the user is notified to add a spare drive (S152). On the other hand, if a spare drive is found (S128), the status of this spare drive in the drive status management table 1916 is changed to “rebuild”. R
Instead of the failed drive, the drive is added to the configuration drive information in the AID group status management table 1917 (S129). And the background processing number management table 191
1 is added to the number of background processes whose unit ID is recorded in 8 (S130).
. After that, data reconstruction of RAID group n is executed in the background process (S
131). In step S132, the drive number variable i is initialized.

When the variable i is initially set in step S132, the variable i is compared with the drive number k of the failed drive acquired in step S122 (S133). As a result, if i = k, it is determined that this drive is a failed drive, the process proceeds to step S131, 1 is added to the drive number variable i, and the process proceeds to the next drive. On the other hand, if i ≠ k, it is determined that this drive is not a failed drive, and drive αi (unit I
The assigned RAID group number m of D is α and the drive number is i is acquired (S1).
34). The RAID group number m in the RAID group status management table 1917
1 is subtracted from the current redundancy of the RAID group (S135).

Whether the current redundancy of RAID group m is less than 1 (that is, 0)
It is determined whether or not (S136).

As a result of the determination, if the current redundancy of the RAID group m is 1 or more, it is determined that the minimum required redundancy is maintained, and the drive number variable i is set to investigate the next drive. 1 is added (S143). On the other hand, if the current redundancy of the RAID group n is less than 1, the process proceeds to step S137.

In step S137, whether the current redundancy of the RAID group m is less than 0 (ie,
Whether or not it is a negative number).

By the processing of S136 and S137, it is possible to estimate the change in the RAID state due to the drive mounted on the unit whose unit ID is α. That is, the unit ID
RAID to which a drive other than the failed drive belongs when the unit with α is removed
Estimate the impact on the group.

As a result, if the current redundancy of the RAID group m is smaller than 0, the user is notified that RAID is not configured and data may be lost (S151).

On the other hand, if the current redundancy of the RAID group m is 0 or more, a spare drive is searched for from units other than the unit ID α (S138). If no spare drive is found as a result of this search (S139), the user is notified to add a spare drive (S152). On the other hand, if a spare drive is found as a result of this search (S13).
9) The status of the spare drive recorded in the drive status management table 1916 is changed to “save” (S140). And the background processing number management table 191
8 is added to the number of background processes whose unit ID is α recorded in 8 (S141).
. Thereafter, the data copy of the drive αi is executed in the background process (S142).
). In step S143, 1 is added to the variable i of the drive number.

When the variable i is updated in step S143, it is determined whether or not the variable i exceeds the maximum value of the drive number (S144). As a result, if the variable i is smaller than the maximum value of the drive number, the process returns to step S132 to investigate the next drive. On the other hand, if the variable i exceeds the maximum value of the drive number, the investigation for all the drives is completed, so the background processing number with unit ID α is 0 with reference to the background processing number management table 1918. It is determined whether or not there is (S145). As a result, if the number of background processes whose unit ID is α is not 0, the background process related to the unit whose unit ID is α is continuing, so that a predetermined time has elapsed (S146), and step S is performed again.
The determination of 145 is performed. On the other hand, if the number of background processes whose unit ID is α is 0,
Since the background process related to the unit having the unit ID α is not performed, the process proceeds to step S147.

In step S147, the drive status recorded in the drive status management table 1916 and having no failure with the unit ID α is changed to “unused” (S147).
. Then, the unit replacement instruction LED of the unit whose unit ID is α is turned on (S148).
), The drive replacement instruction LED of the drive whose drive ID is αk is turned on (S149).
. Thereafter, the user is instructed to replace the unit whose unit ID is α (S150).

FIG. 15 is a flow chart for RAID group data reconstruction processing according to the first embodiment of this invention, which is executed by the failure processing program 1913. Moreover, it starts in step S131 of FIG. 14 and is executed in the background.

Data is read from all normal drives constituting the RAID group n (S161).
). Then, by calculating the parity (exclusive OR) of these read data, the same data as that stored in the failed drive is calculated, and the data is restored (S162). Then, the restored data is written in the spare drive searched in step S127 of FIG. 14 (S163).

Then, it is determined whether all data has been written (S164). As a result, if writing of all data has not been completed, the process returns to step S161, and data is read from all normal drives constituting the RAID group n to reconstruct data. On the other hand, if all the data has been written, the drive status management table 1916
The status of the spare drive recorded in is changed to “in use”. And RAID
The drive configuration recorded in the group status management table 1917 is updated, and the status of the RAID group n is changed to “normal” (S165). Then, 1 is added to the current redundancy of the RAID group with the RAID group number n in the RAID group state management table 1917 (S166). Further, since the data reconstruction process in the background is completed, 1 is subtracted from the number of background processes whose unit ID is α recorded in the background process number management table 1918 (S167).

FIG. 16 is a flowchart of drive data copy processing according to the first embodiment of this invention, which is executed by the failure processing program 1913. Further, step S1 in FIG.
It starts at 42 and runs in the background.

First, data is read from the drive αN (drive whose unit ID is α and drive number is N) (S171). Then, the read data is written in the spare drive searched in step S138 of FIG. 14 (S172).

Then, it is determined whether all data has been written (S173). As a result, if writing of all data is not completed, the process returns to step S171, and further data is read and data is copied. On the other hand, if all the data has been written, the configuration drive recorded in the drive status management table 1916 is updated, and the status of the spare drive is changed to “in use” (S174). Then, 1 is added to the current redundancy of the RAID group with the RAID group number n in the RAID group state management table 1917 (
S175). Further, since the data copy process in the background is completed, 1 is subtracted from the background process number of the unit ID (α) recorded in the background process number management table 1918 (S176).

FIG. 17 is a flowchart of processing when the replacement unit is inserted according to the first embodiment of this invention.

If the insertion of the disk blade is confirmed (S181), it is determined whether the correct unit has been inserted based on whether the unit ID set in the inserted disk blade is correct (S182). Specifically, the unit ID set for the inserted disk blade is equal to the ID of the unit set for the removed disk blade, or the unit ID set for the inserted disk blade is already mounted. It can be confirmed by a method such as not overlapping with the IDs of other units.

If it is confirmed that the unit is correct, the unit replacement instruction for the inserted unit L
The ED is turned off (S183).

Then, the RAID group n to which the drive of the inserted unit belongs is selected with reference to the drive setting content holding table 1915 (S184).

Thereafter, referring to the RAID group state management table 1917, the RAID group n
The normal redundancy and the current redundancy are acquired (S185). Then, the normal redundancy and the current redundancy are compared (S186). As a result, if the current redundancy is smaller than the normal redundancy,
Since the number of drives constituting the RAID group n is less than normal, the drive to which the reconstructed data is to be written is selected, and the status of the drive recorded in the drive status management table 1916 is set to “Rebuild”. Change to After that, in background processing RA
Data reconstruction of ID group n is executed (S187).

Thereafter, it is determined whether or not the investigation for all RAID groups to which the drives of the inserted units belong has been completed (S188). As a result, if all the RAID groups have not been checked, the next RAID group is selected (S189), the process returns to step S185, and the redundancy of the RAID group is further checked. On the other hand, all RAI
If the investigation on the D group has been completed, the unused drive (ie, step S1)
In 87, the status of the drive that was not used for data reconstruction is changed to “spare” (S1).
90).

FIGS. 18 to 29 show how the drive B0 fails and unit B is replaced.
A change in the drive status management table 1916 is shown.

Before a failure occurs in drive B0 (drive with unit ID B and drive number 0), the state of drive B0 is “in use” (FIG. 18).

When a failure occurs in the drive B0, the state of the drive B0 becomes “failure” (S in FIG. 14).
123). Then, the state of the searched spare drive A2 is changed to “rebuild” (FIG. 14).
S129), the spare drive A2 starts to reconstruct the data of the drive B0 (FIG. 14).
S131). FIG. 19 shows a state during data reconstruction of the drive B0. When the data reconstruction is completed, the state of the drive A2 is changed to “in use” and the drive A2 is changed to R
Used as a drive constituting the AID group 2 (FIG. 20).

After data reconstruction processing of the failed drive (or in parallel with this processing), drive B1
Data save processing of .about.B3 is performed. Drive B1 belongs to RAID group 1 and has a RAI
Since the RAID level of D group 1 is RAID 5, its redundancy is 1. Therefore, since the redundancy becomes 0 when the drive B1 is removed, the copy process for saving the data in the drive B1 is started.

In other words, the state of the searched spare drive D2 is changed to “save” (S14 in FIG. 14).
0), copying of data of the drive B1 to the spare drive D2 is started (S14 in FIG. 14).
2). FIG. 21 shows the data save state of the drive B1. When the data copy is completed, the state of the drive D2 is changed to “in use”, and the drive D2 is used as a drive constituting the RAID group 1 (FIG. 22).

Next, the data saving process of the drive B2 is performed. The drive B2 belongs to the RAID group 1. Therefore, if drive B2 is removed, the redundancy will be 0, so drive B
The copy process for saving the second data is started. Specifically, the searched spare drive E
2 is changed to “evacuation” (S140 in FIG. 14), and the spare drive E2 is changed to the drive B2.
Starts copying the data (S142 in FIG. 14). FIG. 23 shows a state where data is being saved in the drive B2. When the data copy is completed, the state of the drive E2 is changed to “in use”, and the drive E2 is used as a drive constituting the RAID group 1 (FIG. 24).

Next, the data saving process of the drive B3 is performed. The drive B3 belongs to the RAID group 4, and since the RAID level of the RAID group 4 is RAID 6, its redundancy is 2. Therefore, even if the drive B3 is removed, the redundancy is maintained at 1, so it is determined that there is no need to save the data in the drive B3 (S136 in FIG. 14). FIG. 25 shows a state in which data is being saved in the drive B3.

If writing occurs during the save process, the write data is written to both the save source drive and the save destination drive. Therefore, even if data is written during the save process, the contents of the save source drive and the save destination drive do not differ.

With the above processing, since the processing of all the drives in unit B is completed, the unit replacement processing (FIG. 14) is completed. This state is shown in FIG.

Thereafter, when the unit B is removed and a new disk blade is attached, the state of all the drives mounted on the unit B becomes “unused” (FIG. 27). Thereafter, the state of the drive B3 to which the reconstructed data is written is changed to “reconstruction”, and the data reconstruction starts (FIG. 28). When the reconstruction is completed, the state of the drive B3 in which the reconstructed data is written is changed to “in use”, and the state of the unused drives (B0 to B2) is changed to “spare” (FIG. 29). .

As described above, in the first embodiment, when a unit (disk blade) is removed, data reconstruction and data copying are performed so that the minimum redundancy of the RAID group related to the unit is maintained. . Then, since replacement of the unit and the drive is instructed after completion of these processes, the unit can be easily replaced.

Next, a second embodiment of the present invention will be described. In the second embodiment, RAID
By setting guaranteed redundancy for each group, even if a drive failure occurs, the R
Control is performed so that the redundancy of the AID group does not fall below the guaranteed redundancy. In the second embodiment, only processes different from those of the first embodiment described above will be described, and description of processes common to the first embodiment will be omitted.

FIG. 30 illustrates a RAID group state management table 1917b according to the second embodiment of this invention.
It is explanatory drawing of.

The RAID group state management table 1917b includes the same data (RAID group number, total storage capacity, RAID) as the RAID group state management table 1917 of the first embodiment.
In addition to the ID level, normal redundancy, current redundancy, constituent drive identification symbol, and RAID status), guaranteed redundancy is recorded. The guaranteed redundancy is the minimum redundancy guaranteed in the RAID group, and control is performed so that the redundancy of the RAID group does not become smaller than the guaranteed redundancy even if a drive belonging to the RAID group fails. is doing. A description of the same data as the RAID group state management table 1917 of the first embodiment is omitted.

  FIG. 31 is an explanatory diagram of a RAID group management screen according to the second embodiment of this invention.

The RAID group management screen of the second embodiment is the same as the R of the first embodiment described above.
Unlike the AID group management screen (FIG. 12), a “redundancy policy” button is provided. When this “redundancy policy” button is operated, a redundancy policy setting screen shown in FIG. On the redundancy policy setting screen, for each RAID group, select whether to always maintain complete redundancy, to maintain the minimum redundancy, or to allow temporary lack of redundancy when replacing units can do.

FIG. 32 is a flowchart of guaranteed redundancy setting processing according to the second embodiment of this invention.
This is executed in the RAID group setting program 1911.

First, when “always maintain complete redundancy” is selected (S261), RA
The guaranteed redundancy of the ID group state management table 1917b is set to the same value as the normal redundancy (S262). If “maintain minimum redundancy” is selected (S263), the guaranteed redundancy of the RAID group state management table 1917b is set to 1 (S264). Further, when “to allow temporary loss of redundancy” is selected at the time of unit replacement (S265), 0 is set as the guaranteed redundancy in the RAID group state management table 1917b (S266). .

FIG. 33 is a flowchart of the RAID group creation processing according to the second embodiment of this invention, and is executed by the RAID group setting program 1911. The RAID group creation processing of the second embodiment is the same as the RAID group creation processing (FIG. 13) of the first embodiment except for S206, and the description of the same processing is omitted.

In step S105, the RAID with the largest number of used drives a in the unit
After searching for the group n, the guaranteed redundancy r of the searched RAID group n is obtained by referring to the RAID group state management table 1917 (S206).

  Then, the number of drives required for the unit removal process (a−r + 1) is compared with the remaining number of spare drives available for the unit removal process (total number of spare drives−s). It is determined whether the number of drives necessary for maintaining the limited redundancy remains as a spare (S107). That is, when a failure occurs on a disk, it is determined whether or not the failure unit can be removed and the failed disk can be replaced.

  As a result of the determination, if the number of drives required for processing is larger than the remaining number of available spare drives, it is determined that there are not enough spare drives required to maintain the guaranteed redundancy r set for RAID group n when the unit is removed. Then, a creation error due to a shortage of spare drives is notified (S111). Then, the temporary registration of each management table created in step S102 is deleted (S112).

  On the other hand, if the number of drives required for processing is less than or equal to the remaining number of available spare drives, it is determined that there are enough spare drives to maintain the guaranteed redundancy r set in RAID group n even when the unit is removed. To do.

FIG. 34 is a flowchart of unit replacement processing when a drive failure occurs according to the second embodiment of this invention, which is executed by the failure processing program 1913. The unit replacement process of the second embodiment is the same as the unit replacement process (FIG. 14) of the first embodiment except for S225 and S236, and the description of the same process is omitted.

In step S124, after subtracting 1 from the current redundancy of the RAID group of RAID group n in the RAID group state management table 1917, it is determined whether or not the current redundancy of RAID group n is smaller than the guaranteed redundancy (S225).

As a result of the determination, if the current redundancy of the RAID group n is equal to or greater than the guaranteed redundancy, the drive number variable i is initialized (S132). On the other hand, if the current redundancy of RAID group n is less than 1, the process proceeds to step S126. In step S126, it is determined whether the current redundancy of RAID group n is less than zero.

In step S135, the RAI of the RAID group state management table 1917 is also displayed.
After subtracting 1 from the current redundancy of the RAID group with D group number m, RAID group m
It is determined whether the current redundancy is smaller than the guaranteed redundancy (S236).

As a result of the determination, if the current redundancy of the RAID group m is equal to or greater than the guaranteed redundancy, 1 is added to the variable i of the drive number (S143). On the other hand, if the current redundancy of the RAID group n is less than 1, the process proceeds to step S137. In step S137, it is determined whether or not the current redundancy of the RAID group m is smaller than zero. By the processing in S236 and S137, it is possible to estimate the change in the RAID state due to the drive mounted on the unit whose unit ID is α. That is, the influence on the RAID group to which the drive other than the failed drive when the unit with the unit ID α is removed is estimated.

As described above, in the second embodiment, a guaranteed redundancy is set for each RAID group, and even if a drive failure occurs, control is performed so that the redundancy of the RAID group does not fall below the guaranteed redundancy. The redundancy can be managed according to the required performance (that is, the importance of stored data) for each RAID group.

Next, a third embodiment of the present invention will be described. In the third embodiment, drive maintenance information is provided and can be additionally applied to the first and second embodiments described above.

FIG. 35 is an explanatory diagram of a RAID group management screen according to the third embodiment of this invention. In the RAID group management screen of the third embodiment, a display field for drive maintenance information is provided at the bottom of the screen.

FIG. 36 is a flowchart of the maintenance urgent notification determination process according to the third embodiment of this invention, and selects drive maintenance information displayed on the RAID group management screen.

Thereafter, referring to the unit first described in the drive setting content holding table 1915 (S301), the number s of drives that are spares in the unit is acquired (
S302). Then, the RAID group n having the largest number of used drives a in the unit is searched (S303). Thereafter, the guaranteed redundancy r set for the searched RAID group n is obtained by referring to the RAID group state management table 1917 (S3).
04).

  Thereafter, the number of drives required for the unit removal process (a−r + 1) is compared with the remaining number of spare drives available for the unit removal process (total number of spare drives−s), and the minimum value is obtained when the unit is removed. It is determined whether the number of drives necessary to maintain the limited redundancy remains as a spare (S305). That is, when a failure occurs on a disk, it is determined whether or not the failure unit can be removed and the failed disk can be replaced.

  As a result of the determination, if the number of drives required for processing is larger than the remaining number of available spare drives, the guaranteed redundancy cannot be maintained if a drive failure occurs. Therefore, the guaranteed redundancy set for RAID group n even when a drive failure occurs It is determined that there are not enough spare drives to maintain r, and a notice that the maintenance urgency level is high (a failure drive needs to be replaced or a drive needs to be added quickly) is notified (S309).

  On the other hand, if the number of drives required for processing is less than or equal to the remaining number of available spare drives, it is determined that there are enough spare drives to maintain the guaranteed redundancy r set in RAID group n even when the unit is removed. To do. Then, it is determined whether or not all units have been examined (S306).

As a result, if all the units have not been checked, the process proceeds to step S307, the next unit to be checked is selected, the flow returns to step S302, and the next unit is continuously checked.

On the other hand, if all the units have been investigated, it is possible to maintain the guaranteed redundancy even if a failure occurs in one drive, so that the maintenance urgency is low (S308).

The maintenance urgency level notification determination process of the third embodiment described above is applicable to the second embodiment described above. However, in steps S304 and S305, not the guaranteed redundancy but the RAID group n. By using the current redundancy, it can be applied to the first embodiment.

As described above, in the third embodiment, since drive maintenance information for maintaining the redundancy of the RAID group is provided, it is possible to prevent the redundancy of the RAID group from lowering than expected. it can. In addition, since it is possible to omit maintenance with low urgency by urgency determination, maintenance that had to be performed multiple times in the past can be performed all at once. Time can be reduced.

It is a block diagram which shows the structure of the disk array apparatus of embodiment of this invention. It is an external view which shows the structure of the disk array apparatus of embodiment of this invention. 1 is a perspective view of a drive housing constituting a disk array device according to an embodiment of the present invention. It is a block diagram of the disk blade which comprises the disk array apparatus of embodiment of this invention. It is a block diagram which shows the structure of the disk array controller of embodiment of this invention. It is explanatory drawing of the control program of embodiment of this invention. It is explanatory drawing of the RAID structure of embodiment of this invention. It is explanatory drawing of the drive setting content holding table of the 1st Embodiment of this invention. It is explanatory drawing of the drive status management table of the 1st Embodiment of this invention. It is explanatory drawing of the RAID group state management table of the 1st Embodiment of this invention. It is explanatory drawing of the background processing number management table of the 1st Embodiment of this invention. It is explanatory drawing of the RAID group management screen displayed on the management terminal device of the 1st Embodiment of this invention. 5 is a flowchart of RAID group creation processing according to the first embodiment of this invention. It is a flowchart of the unit exchange process of the 1st Embodiment of this invention. 6 is a flowchart of a data reconstruction process for a RAID group according to the first embodiment of this invention. It is a flowchart of the data copy process of the drive of the 1st Embodiment of this invention. It is a flowchart of the process at the time of insertion of the exchange unit of the 1st Embodiment of this invention. It is explanatory drawing of the drive status management table (before failure occurrence) of embodiment of this invention. It is explanatory drawing of the drive state management table (during data reconstruction) of an embodiment of this invention. It is explanatory drawing of the drive status management table (after data reconstruction) of embodiment of this invention. It is explanatory drawing of the drive status management table (during data copy) of embodiment of this invention. It is explanatory drawing of the drive state management table (after data saving) of an embodiment of this invention. It is explanatory drawing of the drive status management table (during data copy) of embodiment of this invention. It is explanatory drawing of the drive state management table (after data saving) of an embodiment of this invention. It is explanatory drawing of the drive state management table (during the save process) of the embodiment of this invention. It is explanatory drawing of the drive status management table (after completion | finish of unit replacement | exchange process) of embodiment of this invention. It is explanatory drawing of the drive status management table (after unit replacement | exchange) of embodiment of this invention. It is explanatory drawing of the drive state management table (during data reconstruction) of an embodiment of this invention. It is explanatory drawing of the drive status management table (after data reconstruction) of embodiment of this invention. It is explanatory drawing of the RAID group state management table of the 2nd Embodiment of this invention. It is explanatory drawing of the RAID group management screen of the 2nd Embodiment of this invention. It is a flowchart of the guaranteed redundancy setting process of the 2nd Embodiment of this invention. 10 is a flowchart of RAID group creation processing according to the second embodiment of this invention. It is a flowchart of the unit exchange process of the 2nd Embodiment of this invention. It is explanatory drawing of the RAID group management screen of the 3rd Embodiment of this invention. It is a flowchart of the maintenance emergency level notification judgment process of the 3rd Embodiment of this invention.

Explanation of symbols

1 disk array device 2 management terminal device 3 SAN host 4 SAN
5 Management LAN
10a to 10i Drive housing 19 Controller housing 101 to 113 Drive blade 121 to 133 Unit replacement instruction LED
191 Disk array controller 1901 CPU
1902 Memory 1903 Control program 1904 Data transfer controller 1905 Front end connection interface controller 1906 Back end connection interface controller 1907 Data buffer 1908 LAN interface controller 1911 RAID group setting program 1912 RAID group management program 1913 Fault processing program 1914 I / O control program 1915 Drive Setting holding table 1916 Drive state management table 1917 RAID group state management table 1918 Background processing number management table 2010, 2020, 2030, 2040 Drive (canister)
2110, 2120, 2130, 2140 Connector 2210, 2220, 2230, 2240 Drive replacement instruction LED
2400 Connector 2500 Rotary switch (ID setting part)

Claims (10)

Includes a plurality of drive housings in which a plurality of units in which a plurality of drives are detachably mounted are detachably mounted, and a controller housing in which a disk control unit is provided. In a storage device in which a RAID is configured by the created disk,
The disk controller is
A drive status management table for recording the status of the drive, a RAID group status management table for recording the status of the RAID group, and a back indicating whether data reconstruction processing using RAID and data replication processing between drives are in progress A ground processing number management table, and
When a failure of the drive is detected, the state of the failed drive in the drive state management table is changed to “failure”, and the state of the RAID group to which the failed drive belongs in the RAID group state management table is changed to “failure”. A fault detection unit to be changed;
It is determined whether the redundancy of the RAID group to which the failed drive belongs is smaller than the set request value of 1 or more and 0 or more, and other drives mounted in the unit in which the failed drive is mounted A determination unit for determining whether or not the redundancy of the RAID group to which the group belongs is smaller than the set one or more required values and 0 or more;
If the redundancy of the RAID group to which the failed drive belongs is smaller than the set request value of 1 or more and 0 or more, the status of the spare drive retrieved from a unit other than the unit in which the failed drive is mounted Is changed to “rebuild”, the data of the failed drive is rebuilt in the spare drive, and during this rebuild process, a data rebuilding unit that reflects the status of the spare drive in the background processing number management table ; ,
If the redundancy of the RAID group to which the other drive belongs is smaller than the set request value of 1 or more and 0 or more, the spare drive retrieved from a unit other than the unit in which the other drive is installed The status is changed to “evacuation”, the data of the other drive is duplicated to the spare drive, and the data duplication unit that reflects the status of the spare drive in the background processing number management table during the data duplication processing,
Based on the background processing number management table , it is determined whether or not the processing by the data reconstruction unit and the data replication unit is completed, and when these processes are completed, the unit on which the failed drive is mounted and the fault A replacement instruction unit for notifying that the drive can be replaced.   The data reconstruction unit, when the unit is mounted, if the redundancy of the RAID group to which the drive mounted in the unit should belong is smaller than the normal redundancy, the drive mounted in the mounted unit The storage apparatus according to claim 1, wherein the data of the RAID group is reconstructed, and the exchange instructing unit stops exchangeable notification regarding the inserted unit in accordance with the data reconstruction processing. Before SL disk control unit includes a RAID group creation unit for creating a RAID by said multiple disks,
The RAID group creation unit
An information acquisition unit that acquires the number of unused spare drives s in the unit, the number of drives a in the unit belonging to the RAID group, and the redundancy r of the RAID group;
By comparing the number of drives required for the unit removal process (a−r + 1) with the remaining number of spare drives available for the unit removal process (total number of spare drives−s), the spare is removed when the unit is removed. A determination unit that determines whether or not a drive is insufficient, and determines whether or not a disk can be replaced when a failure occurs, before performing disk replacement;
The storage apparatus according to claim 2, wherein the disk control unit further includes a notification unit that notifies the fact when the determination unit determines that the spare drive is insufficient. Before Symbol disk control unit,
A RAID group creation unit for creating a RAID with the plurality of disks;
When replacing the unit, by reconstructing or duplicating the data of the disk mounted on the unit, a replacement instruction unit for notifying that the unit can be replaced;
The storage apparatus according to claim 2, comprising: The determination unit determines whether the redundancy of a RAID group to which a drive in the unit in which a failure is detected belongs maintains a predetermined value when replacing the unit ;
If the redundancy of the RAID group to which the failed drive belongs does not maintain a predetermined value , the data reconstruction unit stores the data of the failed drive in a free drive retrieved from a unit other than the unit in which the failed drive is mounted. the storage device according to 請 Motomeko 4 to rebuild the.   The data reconstruction unit, when the unit is mounted, if the redundancy of the RAID group to which the drive mounted in the unit should belong is smaller than the normal redundancy, the drive mounted in the mounted unit The storage apparatus according to claim 5, wherein the data of the RAID group is reconstructed. The determination unit determines whether or not the redundancy of a RAID group to which a drive in the unit in which no failure has been detected maintains a predetermined value when replacing the unit ;
If the redundancy of the RAID group to which the drive belongs does not maintain a predetermined value , the data duplicating unit stores the data of the other drive in an empty drive searched from a unit other than the unit in which the other drive is mounted. the storage device according to 請 Motomeko 4 replicate. The determination unit
When a RAID group is created, a spare drive is removed if the unit is removed based on the number of unused spare drives in the unit, the number of drives in the unit belonging to the RAID group, and the redundancy of the RAID group. There the storage device according to 請 Motomeko 4 you determine whether insufficient. Before SL based on the drive state management table and the RAID group state management table storage device according to claim 4 having an information output unit that outputs information for displaying the status of and the RAID group of the drive. The exchange instruction unit
A drive replacement instruction display unit for notifying that the drive mounted on the unit can be replaced;
A drive replacement instruction unit for controlling display of the drive replacement instruction display unit;
5. The storage apparatus according to claim 4, further comprising: a power supply unit that supplies power necessary for the operation of the drive replacement instruction display unit and the drive replacement instruction unit even when the unit is removed from the storage apparatus. .

Images