JP2009070489A - Storage device, its control method and disk device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To minimize performance degradation of a response due to speed change when attaining power-saving by changing rotational speed in a storage device which comprises disk devices capable of read/write processing at a plurality of rotational speeds. <P>SOLUTION: In a controller part 101, disk devices which change speed simultaneously are limited so that data corresponding to a read request can be recovered from the data stored in the disk device of unchanging speed, and rotational speeds of a plurality of disk devices are sequentially made to change speed. A RAID (Redundant Array of expensive Disk) group comprises the plurality of disk devices and disk devices which change speed simultaneously are limited based on a configuration of the RAID group. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a storage apparatus, a control method therefor, and a disk apparatus. For example, the present invention relates to a technique for suppressing performance deterioration due to a shift when power saving is achieved by changing the rotation speed of the disk apparatus.

  In general, in a sub-storage system or server, a system is constructed in accordance with the maximum load (maximum I / O processing request amount). In actual system operation, the maximum load is rare, so there is a problem that excessive operation is often performed and excessive power is consumed.

  In order to solve this problem, three methods have been studied as a method of controlling the disk device in accordance with the operation status of the system. 1) The first method is a method in which the disk device is stopped during an idle period to save power. This method is suitable for applications with low access frequency and long idle time. 2) The second method is a disk device having a low-revolution speed power-saving mode (read / write is not possible), and a transition to the low-revolution speed power-saving mode during the idle period is intended to save power. It is. This method is suitable for applications with a relatively high access frequency and a short idle time. 3) The third method comprises a disk device that can perform read / write processing at a plurality of rotational speeds, monitors the processing status of requests to the disk device, predicts the excess or deficiency of the performance of the disk device relative to the load, and According to the result, the disk rotation speed is changed to save power. This method is suitable for applications where the access frequency is high and the load varies greatly.

  In order to realize the above power saving (third method), for example, as in Non-Patent Document 1, there is a method of executing rotation speed control. In Non-Patent Document 1, an average time of request processing is obtained for each fixed number of requests (hereinafter referred to as average request processing time), and the obtained average request processing time is compared with the average request processing time examined immediately before. Then, the excess or deficiency of the performance of the disk device is predicted based on the difference with the average request processing time, and the rotational speed is changed based on the result, thereby realizing power saving.

S. Gurumurthi, A. Sivasubramaniam, M. Kandemir, H. Franke, “DRPM: dynamic speed control for power management in server class disks”, Annual International Symposium on Computer Architecture, (USA), IEEE Computer Society, 2003, VOL 30 , p.169-181

  However, in the rotation control method of Non-Patent Document 1, although the rotation speed is controlled in accordance with the load, the disk cannot be accessed during the shift, and there is a problem that the response time to the read request becomes longer due to the shift. is there.

  The present invention has been made in view of such circumstances, and provides a technique that enables processing of a read request even when the rotational speed is changed, and minimizes deterioration in response performance of the read processing due to a shift. Is.

  In order to solve the above-described problem, in the present invention, in a redundant storage device such as a RAID system, read request data for a disk device being shifted can be restored by data read from the disk device being not shifted. Thus, the disk device to be changed is limited, and the rotational speed is changed sequentially (after confirming the end of the previous shift process). Then, when it becomes necessary to read the data of the disk device being changed, the data is read from the disk device that has not changed speed, the target data is restored, and the read request can be processed.

  In the present invention, a large-capacity buffer memory is provided, and when a write request is received while the rotation speed is changed, the data is buffered in the memory, and the data is stored on the disk until the shift in the RAID group is completed. By preventing writing, a delay in read request processing after write request processing is prevented.

  Further features of the present invention will become apparent from the best mode for carrying out the present invention and the accompanying drawings.

  According to the present invention, when power saving is achieved by changing the rotational speed of the disk device, request processing can be performed even while the rotational speed is being changed, and performance degradation due to shifting can be minimized. This is particularly effective in systems with high performance requirements.

  The present invention relates to a storage device capable of presenting data recorded on a disk device being changed as a system even when the rotational speed of the disk device is being changed.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be noted that this embodiment is merely an example for realizing the present invention, and does not limit the technical scope of the present invention. In each drawing, the same reference numerals are assigned to common components.

<First Embodiment>
(1) Configuration of Storage Device FIG. 1 is a diagram showing a schematic configuration of a storage device (100) according to the first embodiment of the present invention. The storage device (100) includes a controller unit (101) for controlling the rotation speed, and a storage unit (103) configured by a group (102) of Redundant Arrays of Independent Disks (RAID) configured by a plurality of disk devices. It is equipped with. Each (RG1 to RGN) of the RAID group (102) represents a control unit on the system. Each RAID group is configured, for example, such that RG1 corresponds to RAID1 and RG2 and RG3 correspond to RAID5.

  The controller unit (101) includes a rotation speed management table (104), a speed step registration table (105), a rotation speed control unit (106), a buffer memory (107), and a buffer data management table (108). And. FIG. 1 shows only the configuration of the storage apparatus (100). For example, the storage apparatus (100) is connected to a server (not shown) and is connected to a client terminal apparatus (not shown) via a network (not shown). You may do it.

  FIG. 2 shows a configuration example of the RAID group (102). The RAID group (102) is composed of a plurality of disk devices (21). The storage device of the present invention is configured by a disk device that can perform read / write processing at a plurality of rotational speeds.

  The RAID group is composed of one or a plurality of logical spaces (Logical Units, hereinafter referred to as LU (22)). In the example of FIG. 2, the RAID group (102) is composed of two LUs (LU1 and LU2).

  FIG. 3 shows a configuration example of the LU (31) when configured with RAID5. When configured with RAID 5, the logical space on the disk is configured with data units called stripes (32). An LU (31) is created in the RAID group (102) with this stripe as a unit. In this stripe, not only data to be stored but also data for restoring data when the HDD fails (this data is referred to as parity data (33)) is recorded. In FIG. 3, stripes with numerical values indicate write data from the host computer, and stripes with “P” indicate parity data.

  FIG. 4 shows a setting example (speed step registration table (105)) of the rotational speed of the disk device that can perform read / write processing at a plurality of rotational speeds. The speed of each disk device is set within the range shown in the table of FIG. A speed step registration table is provided for each RAID group. In addition, although the example at the time of setting 5 steps | paragraphs of rotational speed for every 2500 rpm is shown in FIG. 4 by 2500 rpm as a minimum, it is not restricted to this.

  In the storage device (100), the rotation speed that can be set by the disk device is registered in advance in the speed step registration table (105) when the system is constructed.

  Further, when processing the request, the storage apparatus (100) sets the rotation speed to one of settable rotation speeds and processes the request. Specifically, the storage apparatus (100) registers the rotational speed actually executing the request processing among the rotational speeds shown in FIG. 4 in the rotational speed management table (104) of FIG. Process.

(2) Rotational speed change processing When there is a read request to the disk device being changed, the storage device (100) utilizes the data redundancy of the system and reads the data from the non-shifted disk device, The disk device to be shifted is limited so that the read request data to the disk device being shifted can be restored, and the rotational speed of the disk device is changed.

FIG. 5 is a flowchart for explaining the rotation speed changing process.
In the storage device (100), the controller unit (101) limits the disk device to be shifted so that the read request data to the disk device being shifted can be restored by reading data from the disk device that is not shifting. Then, the rotational speed of the limited disk device is changed (step S52). The disk device is limited according to the RAID configuration as will be described later.

  Next, the controller unit (101) checks whether all the shifts of the disk devices in the RAID group have been completed (step S53), and if all the shifts have been completed, the process ends. If all the gear shifts of the disk devices in the RAID group have not yet been completed, the controller unit (101) returns to step S51 and continues the gear shift process.

  Here, an example of the ATA command used for changing the rotation speed will be described. FIG. 6 shows an example of a Set Features command for setting the operation of the disk device. With the ATA command, for example, the rotation speed can be changed by inputting a subcommand code for changing the rotation speed into the Features register of the Set Features command. FIG. 6 shows an example in which 61h is input as a subcommand code indicating a change in rotation speed. It should be noted that numerical values unique to the subcommand can be set in the Sector Count register, LBA (Logical Block Address) Low register, LBA Mid register, and LBA High register. In the rotation speed setting command, for example, a numerical value indicating the rotation speed to be set may be input to the register of the Sector Count.

  FIG. 7 shows an example of the value of Sector Count indicating the rotation speed. For example, by inputting the value shown in FIG. 7 into the Sector Count register of the Set Features command, it is possible to change to a desired rotation speed.

  Further, FIG. 8 shows an example of the execution result of the Set Features command. When the Set Features command is completed, the disk device sets 0 to the BSY bit (81) of the Status register. Therefore, by referring to the BSY bit (81), the controller unit (101) can know the end of the rotation speed changing process.

(3) Read Request Processing FIG. 9 is a flowchart for explaining read request processing to the disk device during shifting. In the storage device (100), the controller unit (101) checks whether or not a read request for the disk device being shifted has been received (step S91). If it is a request to the disk device during the shift, the controller unit (101) reads data for restoration from the disk device that is not shifting in the RAID group (step S92), and read request data for the disk device during the shift. Is restored (step S93).

  Here, with reference to FIG. 10, an example of read data restoration processing when a read request is made to the disk device whose rotational speed is being changed in the RAID5 storage device (100) will be described.

  In FIG. 10, gear shifting is sequentially performed in the order of HDD-A, HDD-B, HDD-C, and HDD-D. In the case of RAID5, the rotation speed is changed one by one, and control is performed so that only one disk device changes the rotation speed at the same time. In this way, only one disk device cannot perform request processing. Therefore, even in RAID 5, even if there is a read request during gear shifting processing, data is read from the non-shifting disk device, so that a read request can be obtained. It is possible to restore data for.

  In the example of FIG. 10, when there is a read request to the HDD-B that is being shifted, the “data a” of the HDD-A, the “data c” of the HDD-C, and the “data d” of the HDD-D are read out. “Data b” is restored by XORing these data.

(4) Write Request Process FIG. 11 is a flowchart for explaining the process (write request process) when there is a write request while changing the rotation speed of the disk devices constituting the RAID group.

  In the storage device (100), when there is a write request while any of the disk devices constituting the RAID group is changing the rotation speed (step S111), the controller unit (101) receives the buffer memory (107). ) Is stored (step S112). Next, the controller unit (101) newly generates parity data corresponding to the write data (step S113), and stores the parity data in addition to the write data in the buffer memory (107) (step S114). The controller unit (101) registers the buffered write data and parity data management data in the buffer data management table (108) (step S115).

  Further, the controller unit (101) checks whether or not the shifts of the disk devices in the RAID group have been completed (step S116), and writes the data stored in the buffer (107) to the disk when the shifts are completed (step S117). If the shift of all the disk devices in the RAID group has not been completed (step S116), the process returns to step S112 and the subsequent processes are executed.

  As described above, the storage device (100) stores the write data and the corresponding parity data in the buffer (107) until the shift of all the disk devices in the RAID group is completed. 107) is written to the disk. As a result, even when there is a write request during the change of the rotation speed of the disk device of the RAID group and there is a read request thereafter, the read request can be processed quickly.

(5) Buffer Data Management Table FIG. 12 is a diagram showing a configuration example of the buffer data management table (108) for managing data on the buffer (107). The buffer data management table (108) is composed of a logical unit (LU) number, a start address in the LU, a stripe number / parity number, and a start address on the buffer (107).

  When the shift of all the disk devices in the RAID group is completed, the storage device (100) uses the information in the buffer data management table (108) to write the write data and the corresponding parity data to the disk.

  Further, when there is a read request during the rotation speed change, the storage apparatus (100) searches whether the data of the request is on the buffer (107), and the corresponding data is stored in the buffer data management table (108). If there is, the data is read from the buffer (107) and transferred to the host computer (controller unit).

  Further, when the write data being buffered is updated, the storage apparatus (100) registers only the latest write data management information in the buffer data management table (108) in order to maintain data consistency.

(6) Specific Processing in Each RAID Configuration In the following, as an example of a redundant storage device, RAID 1, RAID 5, and distributed RAID 5 (hereinafter referred to as distributed RAID 5) in which groups configured with RAID 5 are distributed are taken up. A method for changing the rotation speed and a method for processing a request in each storage apparatus will be described. Note that the scope of the present invention is not limited by the following description.

  FIG. 13 is a diagram illustrating a specific example of the rotation speed changing process in the RAID 1 storage apparatus. In FIG. 13, RAID1 is configured by two HDDs 1 (1301) and 2 (1302).

  For example, if there is a read request to the stripe 1 (1303) of the HDD1 (1301) during the shift of the HDD1 (1301), the controller unit (101) reads the data from the stripe 1 (1304) of the HDD2 (1302). When the change of the rotational speed of the HDD 1 (1301) is completed, the rotational speed of the HDD 2 (1302) is changed. When a write request is received during the change of the rotation speed of the disk device, the controller unit (101) stores the write data in the memory until the change of the rotation speed of the HDD1 (1301) and the HDD2 (1302) is completed. When the buffering is completed and the rotation speed of all the disk devices constituting the storage device (100) is changed, the buffered write data is written to the disk.

FIG. 14 is a diagram illustrating a specific example of rotation speed change processing in a RAID 5 storage device.
In FIG. 14, RAID 5 is configured by four disk devices.

  For example, if there is a read request for the stripe 2 (1402) during the shift of the HDD2 (1401), the controller unit (101) causes the stripe 1 (1404) of the HDD1 (1403) and the stripe 3 (1405) of the HDD3 (1405). 1406) and the stripe P (1408) of the HDD 4 (1407) are read, XOR of these data is taken to restore the data of the stripe 2 (1402), and the read request data is transferred.

  In RAID5, when there is a write request, the controller unit (101) generates corresponding parody data and also stores the data in the buffer (107). Then, after changing the rotation speed of the disk devices in the RAID group, write data and corresponding parity data are written to the disk. Further, when it is necessary to restore the read data using the data of the immediately preceding write request and the corresponding parity data in the read request subsequent to the write request, the read is performed using the data on the buffer (107). Restore data.

FIG. 15 is a diagram showing an example in the case where there is a write request in the stripe of the disk device being shifted in the RAID 5 storage device.
FIG. 16 is a diagram showing an example of a method for generating parity data for the writing of stripe 1 (1502) in FIG.

  In FIG. 15, when the stripe 1 (1502) is written, the controller unit (101) reads the stripe 2 (1503) and the stripe 3 (1504). Then, the controller unit (101) newly generates parity data P (1) ′ corresponding to the write data using the method of FIG. 16, and generates the stripe 1 ′ and parity P (1) ′ in the buffer data management table. Register at (108). In the buffer data management table (108), the stripe number or parity number, and the start address of write data or parity data on the buffer (107) are registered.

  FIG. 17 is a diagram illustrating an example of another method of generating parity data corresponding to the write of stripe 1 in the example of FIG. In FIG. 17, the parity data P (1) -1 (1703) is generated by XORing the stripe 1 (1701) and the parity data P (1) (1702). Next, the parity data P (1) ′ (1705) is generated by XORing the generated parity data P (1) -1 (1703) and the newly written stripe 1 ′ (1704).

  FIG. 18 shows that the write processing of the stripe 1 (1502) shown in the example of FIG. 15 is executed while the rotational speed of the disk device in the RAID group is changed, and then a read request is made to the stripe 3 (1504). Shows a method of restoring read data in the case where the disk device 3 (1505) is changing the rotation speed. In FIG. 18, the controller unit (101) of the storage apparatus (100) reads the stripe 2 (1503) from the HDD 2. Further, the controller unit (101) reads out the stripe 1 ′ and the stripe P (1) ′ from the buffer data management table (108), and takes the XOR of them as shown in FIG. 18 to obtain the stripe 3 (1504). Restore the data.

  FIG. 19 is a diagram illustrating a configuration example in the case of distributed RAID 5 in which groups configured with RAID 5 are distributed and arranged. In the case of distributed RAID 5, the speed change process can be performed simultaneously in the RAID 5 group (1901, 1902) constituting RAID 5. However, one disk device (HDD) in each RAID5 group shifts one by one.

  In FIG. 19, when there is a read request during the shift of any HDD, the stripe data of the HDD being shifted is restored using the stripe data of the non-shifted disk device in the RAID5 group. Since this method is the same as the data restoration method in RAID5 described above, description thereof is omitted here.

  The method of processing the write request during the rotation speed change and the read request after the write is the same as in RAID 5.

<Second Embodiment>
(1) Configuration of Storage Device FIG. 20 is a diagram showing a schematic configuration of a storage device (2000) according to the second embodiment of the present invention. The storage device (2000) includes a controller unit (2001) that controls the rotation speed, and a storage unit (2003) that includes a RAID group (2002) that includes a plurality of disk devices. The RAID group (2002) represents a control unit on the system.

  Each RAID group (2002) includes one or a plurality of disk devices (2006) on which a large-capacity buffer memory (2004) and a buffer data management table (2005) for managing write data on the buffer are mounted.

  The controller unit (2001) has a rotation speed management table (2007) for managing the rotation speeds of the disk devices in the RAID group, and a speed step registration table (2008) for registering the speed steps of the disk devices constituting the storage device (2000). ) And a rotation speed control unit (2009). The rotation control unit (2009) limits the disk device that shifts the data so that the data of the read request for the disk device that is being shifted can be restored by the data read from the disk device that is not being shifted. The rotational speed is changed sequentially (after confirming the end of the previous shift process). In addition, the rotation control unit (2009) performs a process of writing the write request data to the large-capacity nonvolatile memory mounted in the disk device during the rotational speed shift of the disk devices constituting the RAID group. Note that the disk device limiting method for changing the rotational speed depends on the adopted RAID configuration (RAID 1 to RAID 5), as in the first embodiment.

(2) Buffer Data Management Table FIG. 21 is a diagram showing a configuration example of the buffer data management table (2005) installed in the disk device (2006) constituting the storage device (2000).
The buffer data management table (2005) includes information on the LBA of the data stored in the buffer (2004), the address on the buffer (2004), and the size of the data. Using this buffer data management table (2005), data writing and reading are managed. That is, when the shift of all the disk devices in the RAID group is completed, the controller unit (2001) of the storage device (2000) uses the information in the buffer data management table (2005) to write data and corresponding parity data. To the disc. In addition, when there is a read request during the rotation speed change, the controller unit (2001) searches whether the data of the request is on the buffer (2004), and the corresponding data is stored in the buffer data management table (2005). If there is, the data is read from the buffer (2004) and transferred to the host computer. Further, when the buffered write data is updated, the controller unit (2001) registers only the latest write data management information in the buffer data management table (2005) in order to maintain data consistency.

(3) Write request processing (processing on the controller side)
FIG. 22 is a flowchart for explaining write request processing by the controller unit (2001).

  The controller unit (2001) of the storage device (2000) initializes the large-capacity memory (2004) in the disk device before changing the rotation speed (step S2201). Next, the controller unit (2001) stores the write data in the large-capacity buffer memory (2004) of the disk device (2006) that is the target of the write request (step S2202). Subsequently, the controller unit (2001) checks whether or not a buffer full occurs due to the writing of the write data (step S2203), and if a buffer full occurs, determines whether the disk device (2006) in which the buffer full has occurred is shifting. It investigates (step S2204). Thus, the controller unit (2001) manages the remaining buffer capacity of the buffer memory (2004) in the disk device. If the disk device (2006) that caused the buffer full is not shifting, the controller unit (2001) writes the write data in the memory (2004) to the disk until the buffer full is eliminated (step S2205).

  Subsequently, the controller unit (2001) checks whether or not the shift processing of the disk devices in the RAID group (2002) has been completed (step 2206), and if the shift processing has been completed for all the disk devices, the buffer memory (2004). The disk device is instructed to write the above data onto the disk (step S2207).

  In step S2206, if the shift has not been completed in all the disk devices, the process returns to step S2202.

  In step S2204, if the disk device (2006) in which the buffer full has occurred is shifting, the controller unit (2001), after completing the shifting process, until the buffer full is eliminated, Data is written to the disk (step S2208). Thereafter, the processing after step 2206 is executed.

  Returning to step S2203, if there is no buffer full in the buffer memory (2004) of the disk device (2006) that is the target of the write request, the process proceeds to step S2206, and the subsequent processes are executed.

(4) Write request processing (processing on the disk device side)
FIG. 23 is a flowchart for explaining write request processing on the disk device (2006) side during rotation speed change.

  The disk device (2006) constituting the storage device (2000) receives an initialization request from the controller unit, and initializes the large-capacity buffer memory (2004) in the disk device (step S2301). When the disk device (2006) receives the write request, the disk device (2006) stores the write data in the large-capacity buffer memory (2004) (step S2302).

  Further, the disk device (2006) checks whether there is data of the same logical address on the buffer memory (2004) (step S2303). If there is no data of the same logical address, information on the write data is stored in the buffer data management table. (2005) is registered (step S2304). If there is a request from the controller unit to write the data in the large-capacity buffer memory (2004) to the disk (step S2305), the disk device (2006) writes the data in the buffer memory (2004) to the disk (step S2306). ). If it is determined in step S2305 that the controller unit does not request writing of data in the large-capacity buffer memory (2004) to the disk, the process returns to step S2302.

  In step S2303, if there is data of the same logical address on the buffer memory (2004), the disk device (2006) updates the buffer data management table (2005) so that only the latest write data is registered (step S2303). S2307). After the update, the processing after step 2305 is executed.

  Also in the storage device of the second embodiment, as performed in the first embodiment, the read request data to the disk device whose rotational speed is being changed is restored using data that has not been written to the disk and the corresponding parity data. It is possible. In that case, it is necessary to describe the addresses on the buffer in the table of FIG. 12 in Embodiment 1 by setting the disk device number and the logical address on the device as a set.

<Summary>
As described above, the storage device according to the present embodiment is configured by a disk device that can perform read / write processing at a plurality of rotational speeds. It can be used to minimize the deterioration of the read processing response performance.

  The storage device according to the present embodiment is characterized by processing when there is a read request for a disk device during a shift when any of a plurality of disk devices that can be driven at a plurality of rotational speeds is shifting. Specifically, in the controller unit, a disk device that shifts at a time is limited so that data corresponding to the read request can be restored from the data stored in the non-shifting disk device. The rotation speed of the device is sequentially changed. Then, a RAID group is configured by a plurality of disk devices, and the disk devices that are shifted at a time are limited based on the configuration of the RAID group. For example, when a RAID group is composed of RAID 1 or RAID 5, a plurality of disk devices can be shifted one by one, and when a RAID group is composed of distributed RAID 5, gear shifting can be performed simultaneously for each group. Within each group, a plurality of disk devices are shifted one by one. In this way, even if there is a read request during a shift, the read request can be processed without being affected by the shift process, so that the response time of the entire system is extremely slow. Can be prevented.

  More specifically, if the request is a read request for a disk device being shifted, the RAID group data redundancy is used to respond to the read request based on the data stored in the non-shifted disk device. Get the data. That is, for example, when a plurality of disk devices are configured with RAID 1, data is read from a mirrored disk device that is not being shifted. When a plurality of disk devices are configured with RAID 5 or distributed RAID 5, data corresponding to the read request is restored based on data and parity data read from the non-shifting disk device. By doing so, even if the rotational speed changing process is in progress, the data in the disk device being shifted can be restored and presented, so that the response performance as a system can be guaranteed. .

  If the request is a write request made during the shift of any of the disk devices, the write data write processing corresponding to the write request is executed after the shift processing is completed for all of the plurality of disk devices. When the plurality of disk devices are configured with RAID 5 or distributed RAID 5, the write data and parity data corresponding to the write data are stored in the buffer, and the system waits for the end of the shift process. If a read request is received after the write request is received and before the processing (before writing the data to the disk), the write data is required to restore the data corresponding to the read request. Read the write data and parity data from the buffer memory to restore the data corresponding to the read request. By doing so, even if there is a write request during the shift of the disk device and this read request can be processed even if there is a read request before the processing is completed, the response performance of the read processing is deteriorated. There is nothing to do. Note that when restoring data corresponding to the read request, write data and parity data read from the buffer and data acquired from the non-shifting disk device are used.

  According to the second embodiment, each of the disk devices constituting the RAID group stores a buffer memory for storing write data to be written, and a buffer for managing the correspondence between the address of the buffer memory and the logical address of the disk device And a data management table. Again, the disk devices that change gears at a time are limited according to the RAID configuration. As described above, when there is a read request for a disk device during shifting, data corresponding to the read request can be restored from data stored in the disk device during non-shifting. If the request is a write request, write data is written into the buffer memory in the disk device. Writing to the disk is executed by the disk device when a write data write request is received from the controller unit. Further, here, it is determined whether or not the buffer memory is full, and whether or not the disk device having the buffer memory in the full state is shifting, and the write data stored in the buffer memory according to the determination result The timing of the writing process is controlled. In this way, even if a read request occurs after a write request, delay in read request processing can be prevented and the response performance of the storage apparatus can be ensured.

It is a figure which shows schematic structure of the storage apparatus by 1st Embodiment. It is a figure which shows the structural example of a RAID group. It is a figure which shows the example of the data format in a RAID group (RAID 5). It is a figure which shows the example of the rotational speed which can be set in a disc apparatus. It is a flowchart for demonstrating the change process of the rotational speed of a storage apparatus. It is a figure which shows the example of the command for a rotational speed change. It is a figure which shows the example of the subcommand which shows rotation speed. It is a figure which shows the example of the end register | resistor of the command for changing a rotational speed. It is a flowchart for demonstrating the process with respect to the read request to the disk apparatus in the rotation speed change. It is a figure which shows the example of the read request process in rotation speed change in the storage apparatus of RAID5. It is a flowchart for demonstrating the write processing in rotation speed change. It is a figure which shows the example of the buffer data management table which manages the write request data in rotation speed change. FIG. 3 is a diagram showing an example of changing the rotational speed of a disk device in a RAID 1 storage device. It is a figure which shows the example of the speed change of the storage apparatus of RAID5. It is a figure which shows the example of the write request process in the rotation speed change in the storage apparatus of RAID5. It is a figure for demonstrating the parity data generation (example 1) corresponding to write data. It is a figure for demonstrating the parity data generation (example 2) corresponding to write data. It is a figure which shows the example at the time of decompress | restoring the read request data with respect to the disk apparatus which is changing the rotational speed using the last write request data. It is a figure which shows the example of the storage apparatus of distributed RAID5 which distributed and arrange | positioned the group comprised by RAID5. It is a figure which shows schematic structure of the storage apparatus by 2nd Embodiment. It is a figure which shows the structural example of the buffer data management table which manages the write data of a mass memory. It is a flowchart for demonstrating the write control with respect to the RAID group in which the rotational speed change by a controller part is carried out. It is a flowchart for demonstrating the write control during the rotation speed change by the disk apparatus (controller which is not illustrated).

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Controller part 102 RAID group 103 Storage part 104 Rotational speed management table 105 Speed step registration table 106 Rotational speed control part 107 Buffer memory 108 Buffer data management table 2001 Controller part 2002 RAID group 2003 Storage part 2004 Large capacity memory 2005 Buffer data management table 2006 Mass storage disk device 2007 Rotational speed management table 2008 Speed step registration table 2009 Rotational speed control unit

Claims (27)

A storage device that accesses a disk device in response to a request and processes the request,
A plurality of disk devices each capable of being driven at a plurality of rotational speeds;
When there is a read request for the disk device being shifted, the disk device that shifts at a time is limited so that the data corresponding to the read request can be restored from the data stored in the non-shifted disk device, A rotational speed control unit for sequentially shifting the rotational speeds of the plurality of disk devices;
A storage apparatus comprising: The plurality of disk devices constitute a RAID group,
The storage apparatus according to claim 1, wherein the rotation speed control unit limits the disk device that performs a shift at a time based on a configuration of the RAID group. When the RAID group is composed of RAID1 or RAID5, the rotational speed control unit performs a shift process on the plurality of disk devices one by one,
When the RAID group is configured by distributed RAID 5 in which a plurality of groups each configured by RAID 5 are distributed and arranged, the rotational speed control unit can perform shift processing simultaneously for each group. The storage apparatus according to claim 2, wherein the plurality of disk devices are subjected to a shift process one by one.   The storage apparatus according to claim 1, further comprising a request processing unit that processes requests to the plurality of disk apparatuses. The plurality of disk devices constitute a RAID group,
When the request is a read request for a disk device during shifting, the request processing unit uses the data redundancy of the RAID group and uses the data stored in the disk device during non-shifting to read the read request. The storage apparatus according to claim 4, wherein data corresponding to is acquired. When the plurality of disk devices are configured with RAID1,
The request processing unit reads data from a non-shifted mirrored disk device,
When the plurality of disk devices are configured with RAID 5 or distributed RAID 5 in which a plurality of groups each configured with RAID 5 are distributed and arranged, the request processing unit reads data and parity read from the non-shifting disk device 6. The storage apparatus according to claim 5, wherein the data corresponding to the read request is restored based on the data.   When the request is a write request made during a shift of any of the disk devices, the request processing unit, after completing the shift processing for all of the plurality of disk devices, writes the write data corresponding to the write request. The storage apparatus according to claim 4, wherein write processing is executed.   When the plurality of disk devices are configured with RAID 5 or distributed RAID 5 in which a plurality of groups each configured with RAID 5 are distributed and arranged, the request processing unit has the write data and parity data corresponding to the write data. The storage apparatus according to claim 7, wherein the storage device is stored in a buffer memory and waits for the shift process to end.   When a read request is received after the write request is received and before the processing, the request processing unit needs the write data to restore the data corresponding to the read request. The storage apparatus according to claim 8, wherein the write data and the parity data are read from the buffer memory, and the data corresponding to the read request is restored. A storage device that accesses a disk device in response to a request and processes the request,
A plurality of disk devices constituting a RAID group, each of which can be driven at a plurality of rotational speeds;
When there is a read request for the disk device being shifted, the disk device that shifts at a time is limited so that the data corresponding to the read request can be restored from the data stored in the non-shifted disk device, A rotational speed control unit for sequentially shifting the rotational speeds of the plurality of disk devices;
A request processing unit that processes requests to the plurality of disk devices,
Each of the plurality of disk devices has a buffer memory for storing write data to be written, and a buffer data management table for managing the correspondence between the address of the buffer memory and the logical address of the disk device,
The request processing unit, when the request is a write request, executes a write processing of write data corresponding to the write request in the buffer memory.   When the request is a write request made during a shift of any of the disk devices, the request processing unit stores the write data in the buffer memory and performs a shift process for all of the plurality of disk devices. 11. The storage apparatus according to claim 10, wherein a write process of the write data is executed after completion.   The request processing unit determines whether or not the buffer memory is full, and whether or not the disk device having the buffer memory in the full state is shifting, and is stored in the buffer memory according to the determination result. 12. The storage apparatus according to claim 11, wherein the timing of the write processing of the write data is controlled. The request processing unit transmits a write request for write data stored in the buffer memory to the disk device,
The storage apparatus according to any one of claims 10 to 12, wherein the disk apparatus executes writing of the write data when the write request is received. A method of controlling a storage device that accesses a disk device in response to a request and processes the request,
The storage device comprises a plurality of disk devices each capable of being driven at a plurality of rotational speeds,
When the rotational speed control unit has a read request for the disk device being changed, the data corresponding to the read request can be restored from the data stored in the non-shifted disk device at a time. A control method characterized by limiting disk devices to be shifted and sequentially shifting the rotational speeds of the plurality of disk devices. The plurality of disk devices constitute a RAID group,
The control method according to claim 14, wherein the rotation speed control unit limits the disk device that shifts at a time based on a configuration of the RAID group. When the RAID group is composed of RAID1 or RAID5, the rotational speed control unit performs a shift process on the plurality of disk devices one by one,
When the RAID group is configured by distributed RAID 5 in which a plurality of groups each configured by RAID 5 are distributed and arranged, the rotational speed control unit can perform shift processing simultaneously for each group. The control method according to claim 15, wherein a shift process is performed for each of the plurality of disk devices.   The control method according to any one of claims 14 to 16, wherein the storage device further includes a request processing unit, and the request processing unit processes requests to the plurality of disk devices. The plurality of disk devices constitute a RAID group,
When the request is a read request for a disk device during shifting, the request processing unit uses the data redundancy of the RAID group and uses the data stored in the non-shifting disk device to read the read request. The control method according to claim 17, wherein data corresponding to is acquired. When the plurality of disk devices are configured with RAID1, the request processing unit reads data from a non-shifted mirrored disk device,
When the plurality of disk devices are configured with RAID 5 or distributed RAID 5 in which a plurality of groups each configured with RAID 5 are distributed and arranged, the request processing unit reads data and parity read from the non-shifting disk device The control method according to claim 18, wherein the data corresponding to the read request is restored based on the data.   When the request is a write request made during a shift of any of the disk devices, the request processing unit, after completing the shift processing for all of the plurality of disk devices, writes the write data corresponding to the write request. The control method according to claim 17, wherein a writing process is executed.   When the plurality of disk devices are configured with RAID 5 or distributed RAID 5 in which a plurality of groups each configured with RAID 5 are distributed and arranged, the request processing unit has the write data and parity data corresponding to the write data. 21. The control method according to claim 20, further comprising: waiting in a buffer memory until the shift process is completed.   When a read request is received after the write request is received and before the processing, the request processing unit needs the write data to restore the data corresponding to the read request. The control method according to claim 21, wherein the write data and the parity data are read from the buffer memory, and data corresponding to the read request is restored. A method of controlling a storage device that accesses a disk device in response to a request and processes the request,
The storage device comprises a plurality of disk devices that constitute a RAID group, each of which can be driven at a plurality of rotational speeds,
Each of the plurality of disk devices has a buffer memory for storing write data to be written, and a buffer data management table for managing the correspondence between the address of the buffer memory and the logical address of the disk device,
When there is a read request for the disk device being changed, the rotation speed control unit changes the speed at a time so that data corresponding to the read request can be restored from the data stored in the non-shifted disk device. Limiting disk devices, sequentially changing the rotational speed of the plurality of disk devices,
A control method, wherein a request processing unit that processes requests to the plurality of disk devices executes write processing of write data corresponding to a write request in the buffer memory when the request is a write request.   When the request is a write request made during a shift of any of the disk devices, the request processing unit stores the write data in the buffer memory and performs a shift process for all of the plurality of disk devices. 24. The control method according to claim 23, wherein a write process of the write data is executed after completion.   The request processing unit determines whether or not the buffer memory is full, and whether or not the disk device having the buffer memory in the full state is shifting, and is stored in the buffer memory according to the determination result. 25. The control method according to claim 24, further comprising controlling a timing of the write processing of the write data. The request processing unit transmits a write request for write data stored in the buffer memory to the disk device,
The control method according to any one of claims 23 to 25, wherein the disk device executes the writing of the write data when the write request is received. A plurality of disk devices capable of being driven at a plurality of rotational speeds, each having a large-capacity memory and a buffer data management table for managing the data,
The write data is buffered in the large-capacity memory according to an instruction from the host, and the write data in the large-capacity memory is written to the disk when an instruction to write the large-capacity memory data to the disk is received from the host side. Disk unit to be used.

Images