JPH11191037A - Data storage device - Google Patents

Abstract

(57) [Summary] PROBLEM TO BE SOLVED: To improve the use efficiency of a buffer in a drive and to improve read performance when performing pre-reading when a plurality of data read request processes are operating simultaneously. SOLUTION: A recording medium 350 controlled by a disk controller 340 and data read ahead from the recording medium 350 are stored in a disk drive 300 constituting a disk storage device, and a plurality of segments 360 to 360 are stored.
Drive buffer 330 divided and managed in 376
And a disk I / F 310 for controlling data input / output to / from the disk controller, and a microprocessor 320 for controlling the whole, and the microprocessor 320 transmits data to each of a plurality of data sets coming from the disk controller. A drive buffer management unit 323 that dynamically performs an individual allocation / release operation for each of the plurality of segments 360 to 376 to a data read request corresponding to an individual data set in accordance with an instruction accompanying the read request; Was.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data storage technique, and more particularly to a technique effective when applied to, for example, the efficiency of continuous reading processing of stored data.

[0002]

2. Description of the Related Art In general, a disk control device accesses a central processing unit when continuous data is instructed from the central processing unit and when the access pattern from the central processing unit is recognized as being continuous. Data is read in advance on the cache by prefetching. As a result, the data in the cache may be transferred when the central processing unit accesses, and the speed is high because the operation time of the disk drive is not waited.

However, if the access speed of the central processing unit> the drive read speed (the transfer speed of the entire ECC group in the case of RAID5), the read-ahead is not completed before the access of the central processing unit. Is reduced. Therefore, it is important to increase the reading speed of the disk drive.

As a method of improving the drive reading speed, for example, a method disclosed in Japanese Patent Application Laid-Open No. 08-152973 is to use a disk drive with a buffer and perform the pre-reading by itself. Here, the prefetch performed by the drive is referred to as prefetch. However, when the disk controller performs prefetch, by performing the prefetch at the same time, the next range of the prefetch range can be read into the buffer of the drive. At this time, since the prefetch operation is performed subsequent to the reading of the requested data by the control device, there is no need for a mechanical overhead such as a head positioning time and a rotation wait. Therefore, the use efficiency of the disk drive can be improved when prefetching is not performed as compared with the case where prefetching is not performed, and the transfer speed to the central processing unit during continuous data access can be improved accordingly.

[0005]

SUMMARY OF THE INVENTION David A. Patterson et al., University of California Report No. UCB / CSD / 887391 (1.
987), a plurality of disk drives are combined, and input / output data from the central processing unit is divided and recorded / reproduced into / from each group (herein referred to as an ECC group). However, when part of the ECC group becomes unavailable due to a temporary or permanent failure, redundant data may be stored in a part of the ECC group so that the failed data can be recovered from other normal disk drives. It reports on the storage method. Among them, they changed the redundant data storage method from RAID 1 to RAID
It is classified into five levels called five. Of these five levels, RAID 5 in particular has a feature that a load is not concentrated on a specific storage device by not fixing a disk drive that stores redundant data.

For this reason, when RAID5 is used, redundant data may exist in the range of the prefetch target depending on the data length to be prefetched. Since redundant data is used when a disk drive fails and is not normally used, pre-reading the range of this redundant data also causes a reduction in the use efficiency of the memory medium constituting the buffer and the drive.

Further, when a plurality of prefetches are operating simultaneously, a plurality of different ranges viewed from the disk drive are continuously accessed. Therefore, if the prefetch is operated under this condition, the data prefetched and stored in the drive buffer may be overwritten by the prefetch that operates in association with another prefetch, and may be wasted.

In RAID 5, when a disk drive read failure occurs, the following correction read is executed to recover the failed data. That is, data is read from all disk drives in the ECC group except for the failed disk drive. At this time, redundant data is also read. The fault data is calculated from the read data and the redundant data and recovered. For this reason, if the disk drive is faulty and cannot be read, and pre-reading is performed while recovering data by performing a correction read, the performance is lower than in the normal state because redundant data must also be read. descend.

The following drive copy and correction copy processes are available for continuous access to the drive in addition to read-ahead. That is, the correction copy is a means for recovering all data on the failed drive from the remaining drives in the ECC group and copying the data to the spare disk drive when the disk drive fails and becomes inoperable. Drive copy is
This is a means for copying all data on one disk drive to a spare drive. When a drive copy or a correction copy is performed while performing I / O processing from the central processing unit, the drive copy or the correction copy reads the copy source data, recovers the data during the correction copy, and copies the data, ie, a spare disk drive. The process of writing to is repeatedly executed to copy.

At this time, although the read from each disk drive is continuous, if an input / output request is issued from the central processing unit on the way, the head needs to be positioned again. This delays the read time in drive copy and correction copy, and increases the time required for copying. In this case, by prefetching the copy source data of the drive copy and the correction copy using the prefetch means, the reading speed of the copy source data can be increased, and the time required for the copy can be reduced.

[0011] However, when a plurality of pre-reads are operating simultaneously, a plurality of different ranges are continuously accessed as viewed from the disk drive. Therefore, if the prefetch is operated under this condition, the data prefetched and stored in the drive buffer may be overwritten by the prefetch that operates in association with another prefetch, and may be wasted.

A first object of the present invention is to improve the use efficiency of a buffer in a drive and improve the read performance of prefetching when prefetching is performed while a plurality of data read request processes are operating simultaneously. It is in.

It is a second object of the present invention to efficiently perform pre-reading and improve read-ahead read performance when redundant data is included in a pre-read target range in a data storage device having a redundant storage configuration.

A third object of the present invention is to improve the read performance of the pre-read in the correction read in a data storage device having a redundant storage configuration.

A fourth object of the present invention is to efficiently perform drive copy and correction copy in a data storage device having a redundant storage configuration, and to reduce the time required for copying.

[0016]

According to the present invention, there is provided a drive provided with a rotary storage medium in which data is stored, interposed between the drive and a host device, and data transfer between the drive and the host device. In a data storage device including a storage control device that controls a buffer provided in the drive and temporarily storing data read from the rotary storage medium, and managing the buffer by dividing the buffer into a plurality of segments, For each of a plurality of different read requests generated from a higher-level device, a buffer management control logic for executing pre-reading of data from the rotary storage medium to the buffer regarding the read request using different segments.

More specifically, as an example, in order to achieve the first object, a data storage device according to the present invention comprises a buffer dividing means for dividing a drive buffer into several areas (segments); When read-ahead operations are performed concurrently, the storage controller determines the read-ahead priority according to the access frequency of the higher-level device to the read-ahead data, allocates segments from those having higher priority, and allocates segments. It is possible to adopt a configuration having a segment allocating means for causing the drive to prefetch only read ahead.

In order to achieve the second object, the data storage device according to the present invention, when reading a range to be read from a recording medium and then prefetching subsequent data, when the range to be prefetched is redundant data, The redundant data skipping means for skipping redundant data and prefetching from the next address of the redundant data can be adopted.

In order to achieve the third object, a data storage device according to the present invention comprises a failure detecting means for detecting a failure in a drive, and a redundant storage for the drive when prefetching when a failure occurs in the drive. A configuration may be provided having redundant data prefetch means for indicating that data is to be prefetched and prefetching redundant data.

In order to achieve the fourth object, a data storage device according to the present invention includes a redundant data prefetch means for prefetching redundant data as a prefetch target during drive copy and correction copy, and a prefetch, drive copy and correction copy execution. In some cases, a segment allocation means may be provided which preferentially allocates a segment to a drive copy or a collection copy, and causes the drive to prefetch only the assigned read-ahead, drive copy, or correction copy.

When a read request is received from a higher-level device, the presence or absence of a continuous data instruction or the access pattern of the higher-level device is determined. Causes the next range to be prefetched. At this time, by using the redundant data skip means, the prefetch execution efficiency is improved, the drive occupancy time is reduced, the drive wait time at the time of prefetch is reduced, the drive response time is reduced, and the continuous data It is possible to improve the transfer speed to the host device at the time of access.

At the time of execution of prefetching, the prefetched data is prevented from being overwritten by another prefetching by the segment allocating means for judging the access frequency of the host device to the prefetched data and allocating segments according to the access frequency. In addition, it is possible to improve the segment hit rate, increase the read-ahead read speed, and improve the data transfer speed to the host device as a whole system.

When read-ahead is performed when a drive cannot be read due to a failure, by using redundant data prefetch means, redundant data required to recover data of a failed drive can be read at high speed. This makes it possible to perform high-speed recovery of faulty data and improve the transfer speed of read data to a higher-level device during a collection read.

When a drive copy or a collection copy is executed, segments are allocated according to the presence / absence of prefetching, the access frequency, and the presence / absence of the drive copy / collection copy. It is possible to prevent overwriting by another prefetch, drive copy, or collection copy, improve the segment hit rate, efficiently perform drive copy, collection copy, and reduce the time required for copying.

[0025]

Embodiments of the present invention will be described below in detail with reference to the drawings.

(Embodiment 1) FIG. 1 is a conceptual diagram showing an example of the configuration of an information processing system including a data storage device according to a first embodiment of the present invention. In the present embodiment, a case where the present invention is applied to a disk storage device having a redundant storage configuration will be described as an example of a data storage device.

The information processing system according to the present embodiment comprises a central processing unit 10, a disk control unit 20, and eight disk drives 300 which can operate independently as storage devices.
And a disk storage device composed of

The disk storage device of the present embodiment has seven disk drives 300 (disk drive 300a).
ECC group (a unit of data recovery in the event of a failure), and the remaining one is a spare disk drive 3
00h. Central processing unit 10 and disk control unit 2
0 is connected via a channel path 60, and the disk controller 20 and the disk drive 300 are connected by a drive path 70 that can operate independently.

The disk controller 20 has a host I
/ F40, a data divider / reproducer 50, a microprocessor 30, a cache memory 110, a shared memory 140 provided with a cache directory 120,
Redundant data generator 130 and drive controller 80
It is composed of The host I / F 40 and the data divider / reproducer 50 are connected to the channel path 60 and the cache memory 1.
10 and the microprocessor 30 and the signal line.

In the present embodiment, a look-ahead table 150 and a copy table 160 having the following configurations are set in a part of the shared memory 140.

The microprocessor 30 is connected to the cache memory 110, the cache directory 120, the redundant data generator 130, and the drive controller 80 by signal lines, and each is realized by a microprogram in the microprocessor 30. Cache memory reference means 31, cache directory reference means 32, parity generator control means 34, difference data generator control means 35, drive controller control means 3
6 is controlled. Also, the microprocessor 30
A sequential access mode determining unit 38 checks whether a sequential access is designated from the central processing unit 10 by a microprogram in the microprocessor 30. The read / write data from the central processing unit 10 is used to determine the actual disk drive 300. It has a mapping operation means 39 for calculating the above storage position.

The cache memory 110, the cache directory 120, and the redundant data generator 130 are connected by signal lines. The cache memory 110 is connected to the drive controller 80 by a signal line, and can mutually transfer data. The cache memory 110 is divided into a write surface 111 for storing write data from the central processing unit 10 and a read surface 112 for storing data read from the disk drive 300. , Each of which is a slot 113 which is a unit divided into sector lengths.
~ 118.

FIG. 2 is a conceptual diagram showing an example of the configuration of the disk drive 300 constituting the disk storage device of the present embodiment. In the present embodiment, the disk drive 30
0 is, for example, a magnetic disk drive using a magnetic disk as a storage medium.

The disk I in the disk drive 300
/ F 310, a drive buffer 330, and a disk controller 340 for controlling the operation of a recording medium 350 composed of a rotary storage medium such as a magnetic disk are connected to the microprocessor 320 by signal lines, respectively. 320.

In the case of this embodiment, the drive buffer 3
A plurality of segments 360 to 376 are arranged in 30. The segment 360 is a segment used when reading the requested range of the disk controller 20 regardless of the presence / absence of the prefetch instruction.
Sixteen of 76 are prefetch segments to which dynamic allocation is performed in response to a prefetch instruction or the like.

The microprocessor 320 operates according to a microprogram in the microprocessor 320.
The disk controller 310 controls the disk I / F 310 and
I / F control means 321 for receiving a request from
And a drive buffer reference means 322 for referring to each of the segments 360 to 376 in the drive buffer 330.
And a drive buffer management unit 323 for managing segments in the order of access from the disk control device 20.
And a disk controller control means 324 for controlling the disk controller 340.

FIG. 3 is a conceptual diagram showing an example of a method of mapping input / output data to / from a central processing unit to a disk drive in the disk storage device of the present embodiment having a redundant storage configuration.

First, the redundant data P0 is located on the rightmost side of the first column.
00, and the second and subsequent columns are located one position to the left of the parity position of the previous column. If the redundant data storage position of the previous column is the leftmost position, it is the rightmost position. Redundant data P001 to P004 are arranged as described above. The division D000 to D029 of the write data from the central processing unit 10 is as follows.
From the right side of the redundant data, however, when the redundant data is on the rightmost side, the mapping is performed in order from the leftmost side. The redundant data in each column is the data in each column, for example, D000 to D0.
05 is generated and stored so as to be equal to the exclusive OR of each of the data in each column, and if one of the data in each column fails, the exclusive data of the remaining data in the column and the redundant data are excluded. Failure data can be recovered by OR.

FIG. 4 is a conceptual diagram showing an example of a more detailed configuration of the information processing system including the disk storage device of the redundant storage configuration according to the present embodiment illustrated in FIG. In the disk control device 20 having the configuration of FIG. 4, the channel path 60 of FIG.
1 to 260-8, the host I / F 40 and the data dividing / reproducing device 50 of FIG.
3-2, the microprocessor 30, the drive controller 80, and the redundant data generator 130 in FIG. 1 correspond to the disk adapters 233-1 to 233-4, and the cache memory 110 in FIG.
1 corresponds to the shared memories 234-1 to 234-2, respectively.

These components of the disk controller 20 are connected to each other via data transfer paths 237-1 to 237-2. Further, these components are connected to a service processor via a communication bus 236 within the controller. 235. The service processor 235 includes the maintenance terminal 2
It is operated by a maintenance person via 50.

The drive path 70 in FIG. 1 includes drive paths 270-1 to 270-4 and drive path 270-
5-270-8, drive path 270-9-270-1
2, corresponding to the drive paths 270-13 to 270-16.

The disk drive 300 shown in FIG. 1 includes disk drives 242-1 to 242-32 and disk drives 242-242 included in a plurality of disk drive boxes 241-1 to 241-2 constituting the storage unit 240.
33-242-64.

In FIG. 1, there is one ECC group, but in FIG. 4, there are a total of eight ECC groups. Disk drives 242-1 to 242-7, 242
-9 to 242-15, 242-17 to 241-23, 2
42-25 to 242-31, respectively, are ECC groups. The disk drives 242-8, 242-
16, 242-24 and 242-32 are spare disk drives. Disk drives 242-33 to 242-
The same applies to 64.

FIG. 5 is a conceptual diagram showing an example of the configuration of the look-ahead table 150 used in the disk storage device of the present embodiment. The look-ahead table of this embodiment is 15
0 is the previous host access position 15 for each logical device number 150a and host-specified data set range 150b.
0c, host access count 150d, last access time 150e, average arrival frequency (average access interval 150
f), prefetching order 150 g, prefetching execution flag 15
It stores various control information such as 0h, the target segment number 150i, and the like, and manages the execution state of the prefetch for each data set based on the control information.

FIG. 6 is a conceptual diagram showing an example of the configuration of the copy table 160 used in the disk storage device of the present embodiment. In the present embodiment, a drive number is assigned to each disk drive, and a copy pointer 160b indicating the progress of copying and a segment number 160c designated for the disk drive are managed for each drive number (copy target drive number 160a).

Now, a host (central processing unit 10) is provided for the disk storage device of this embodiment having such a configuration.
Control unit 2 when a read request is issued from
An example of the operation of the microprocessor 30 in 0 will be described with reference to the flowcharts shown in FIGS.

When a read request is issued from the host in step 1000, the flow advances to step 1100 to determine whether or not the requested data exists in the cache memory 110. This is determined by the cache directory reference means 32 referring to the cache directory 120. If the data does not exist in the cache memory 110, the flow advances to step 1200 to transfer the host request data from the disk drive to the cache memory 110.
Then, the data on the cache memory 110 is transferred to the host. When the host request data exists in the cache memory 110, the process proceeds to step 1300, and the data in the cache memory 110 is transferred to the host. Next, in step 1400, the data set range specified at the time of the read request from the host is checked, and it is determined whether the data set length is equal to or greater than a predetermined value. This is because only a data set having a certain length or more is to be prefetched. If it is less than the specified value, the process proceeds to step 1500 and ends. If the value is equal to or larger than the predetermined value, the process proceeds to step 1600, and the look-ahead table 15
It is determined whether the same logical device number 150a and data set range 150b have already been registered on 0. If not registered, the process proceeds to step 1700 to register the logical device number 150a and the data set range 150b. Then step 18
In step 00, the average access interval 150f is calculated using the look-ahead table 150.

That is, the average access interval 150f read from the prefetch table 150 is Hm, the host access count 150d is N, and the last access time 150e is T.
L, and assuming that the current time is TN, the current average access interval 150f (Hm) is calculated as follows: Hm = {Hm × N + (TN−TL)} / (N + 1). It is stored in the interval 150f. Next, the process proceeds to step 1900, where the previous access time 150
The current time is stored in e. Next, the process proceeds to step 2000, where it is determined whether or not there is a host continuous data instruction. If there is no host continuous data instruction, the process proceeds to step 2100.

Steps 2100 to 2500
Is to determine whether the host access is continuous or not. The description will be made in the following order. First, step 21
00, the number of accesses 150 in the look-ahead table 150
Add 1 to d. Next, the process proceeds to step 2200, where it is determined from the look-ahead table 150 whether the previous access position (150c) +1 is the current access position. If the previous access position (150c) +1 is not the current access position, the process proceeds to step 2400, and the look-ahead table 150
The current access position is stored in the previous access position 150c, and the process proceeds to step 2700 to end the process. If the previous access position (150c) +1 is the current access position, the process proceeds to step 2300, where the current access position is stored in the previous access position 150c of the look-ahead table 150. Next, the process proceeds to step 2500, where the number of accesses 15
It is determined whether 0d is greater than or equal to a predetermined value. If it is less than or equal to the predetermined value, the process proceeds to step 2700 and ends.

When the number of accesses 150d is larger than the predetermined value or when there is a continuous data instruction from the host, the flow advances to step 2600 to refer to the pre-reading flag 150h of the pre-reading table 150 and execute pre-reading in the same data set. Is determined. If the prefetch has already been performed, the process proceeds to step 2700, and the process ends. If the prefetch has not been performed, the process proceeds to step 3000,
In the entire look-ahead table 150, the priority of the look-ahead of the data set is determined. This is determined based on how many data sets have an average access interval 150f shorter than the average access interval 150f of the data set. Next, the process proceeds to step 3100, where it is determined whether the drive copy or the correction copy is being performed by using the drive state reference unit 33. Drive copy,
If the collection copy is being executed, the process proceeds to step 3200, and the number of usable segments is set to the number of mounted segments− (the number of drive copies and the number of collection copies executed). This is because some of the segments 361 to 376 are preferentially allocated for drive copy and collection copy, and the use of read-ahead segments is restricted accordingly. Next, the process proceeds to step 3300, where it is determined whether the priority of the prefetch is within the number of usable segments. If the number of segments is equal to or more than the number of usable segments, the process proceeds to step 3400, and it is determined that no prefetch instruction is given at the time of prefetching. Next, the process proceeds to step 3500, where it is determined whether the prefetch has executed the prefetch last time. If prefetch has not been executed, a segment number at the time of prefetch is newly assigned in step 3600. The new segment number was used by the previous look-ahead with the 16th priority,
Assign a segment number. In the case of executing the previous prefetch, the process proceeds to step 3700, and the previously used segment number is assigned as the designated segment number. Next, the process proceeds to step 3800, where the drive status is referred to by using the drive status reference means 33. Next, the process proceeds to step 3900, and in the case of the collection access state, step 40
Then, the process proceeds to 00 and the redundant data is also subjected to prefetching and prefetching. Next, the process proceeds to step 4100 to calculate a pre-read target drive and a read target range of each drive. Next, the process proceeds to step 4200, where the read start, read data length, presence / absence of prefetch instruction, segment number when prefetch instruction is present, and presence / absence of skip of redundant data are designated for each prefetch target drive. A read request is issued to 300. Next, the data requested in step 4300 is stored in the cache memory 110.
To be stored.

Next, the operation of the disk drive 300 when a read request is received from the disk controller 20 will be described with reference to FIG.
This will be described with reference to the flowchart of FIG.

First, in step 5000, the disk I / F control means 321 receives a read request from the disk controller 20 using the disk I / F 310. At this time, information such as a read request range (read start address and transfer data length), presence / absence of a prefetch instruction, presence / absence of a segment number designation, presence / absence of skip of redundant data, and the like are received. Next, proceed to step 5100,
If there is no prefetch instruction, the process proceeds to step 5500. If there is a prefetch instruction, step 5200.
Proceed to. In step 5200, a prefetch target range is calculated. In the prefetch target range, the start address is the next address of the read request range of the disk control device 20, and the data length is the segment size. However, when there is redundant data in this range and a redundant data skip is instructed, the redundant data is not taken as the target range. Next, the process proceeds to step 5300, and if there is no instruction of the segment number, the process proceeds to step 5400. In step 5400,
As a segment for prefetch, the disk controller 20
The most recently referenced segment among the segments 361 to 376 is assigned, and the flow advances to step 5500. If there is a segment designation in step 5300, the process proceeds to step 5500. In step 5500, the disk controller 340 positions the head at the start address of the required range of the disk controller 20 and transfers the required range of the disk controller 20 to the segment 360. In synchronization with this, the disk I / F 310 transfers the data transferred to the segment 360 to the disk controller 20 through the disk I / F 310. In step 5600, the request range of the disk controller 20 is changed to all segments 360.
Execute until stored in.

Then, the process proceeds to a step 5700, wherein it is determined whether or not there is a prefetch instruction. If there is no prefetch instruction, the process proceeds to step 5800 and ends. If there is a prefetch instruction, the process proceeds to step 5900, where the prefetch is executed following the reading of the requested data by the disk controller 20. At this time, the data is stored in the segment specified by the disk control device 20 or the segment calculated in step 5400. During the execution of the prefetch, the determination from step 6000 to step 6300 is performed. First, step 6000 determines whether the prefetch target range has been transferred to the segment or not. If the transfer has been completed, the process proceeds to step 6100 and ends. Step 62
00 determines whether the address that is about to be read corresponds to redundant data. If the data corresponds to redundant data, the flow advances to step 6300 to determine whether there is an instruction to skip redundant data. If there is an instruction to skip redundant data, in step 6400, the range of redundant data is skipped. That is, the disk controller control means 324 uses the disk controller 340 to move the head to the next read position of the recording medium 350. Thereafter, the process proceeds to step 6500 to restart the prefetch. If there is no redundant data skip instruction, the range of redundant data is also prefetched continuously. This processing is executed until the target range read is completed in the determination of step 6000.

At the time of drive copy and correction copy, the microprocessor 30 in the disk control device 20
An example of the operation will be described with reference to a flowchart shown in FIG.

First, in step 7000, it is determined whether the prefetch was executed last time, that is, whether segments have been allocated for drive copy and correction copy. This is determined based on whether the segment number 160c of the copy table 160 has been determined. If a segment has not been allocated, the flow advances to step 7100 to refer to the look-ahead table 150 and allocate a segment number used by the look-ahead of the 16th priority. If the segment has already been allocated, the flow advances to step 7200 to prefetch using the previously used segment.
Next, the process proceeds to step 7300 to refer to the drive to be read and the read range of each drive. At the time of drive copying, a range of a predetermined length from one copy source disk drive and the address indicated by the copy pointer 160b is targeted. At the time of collection copy, all disk drives 300 other than the failed disk drive in the ECC and the copy pointer 160b of each disk drive 300
A range of a predetermined length from the address shown in FIG. Next, the process proceeds to step 7400, where a read request is issued to the disk drive 300 to be copied. At this time, for each disk drive 300, the number previously obtained as the read target address, data length, segment number,
Specifies no skipping of redundant data. Then step 7
Proceeding to 500, the data read from the disk drive 300 is stored in the cache memory 110. Next, the process proceeds to step 7600, where it is determined whether it is a collection copy, and in the case of a collection copy, the process proceeds to step 7700. In step 7700, data on the failed disk drive is recovered from the data stored in the cache memory 110 in step 7500 by using the redundant data generator 130. Next, the process proceeds to step 7800 and proceeds to step 7700.
In the case of the data recovered by the above or the drive copy, the disk drive 300 (300a to
The data read from 300g) is written to the spare disk drive 300h. Next, the process proceeds to step 7900, where it is determined whether the copy is completed. This is copy table 1
60, the copy pointer 160b is
It is determined whether the maximum address of 00 has been reached. When copying is completed, the process proceeds to step 8000 and ends. If copying has not been completed, the process proceeds to step 8100. In step 8100, 1 is added to the copy pointer 160b, and the process proceeds to step 7000, where the next copy pointer 16b is added.
Copying from the address indicated by 0b is started.

As described above, according to the disk storage device of the present embodiment, the write data coming from the central processing unit 10 is distributed to the plurality of disk drives 300 together with the redundant data generated from the write data. When a plurality of data read request processes from the central processing unit 10 are simultaneously operating in a disk storage device having a redundant storage configuration for
This has the effect of increasing the use efficiency of the drive buffer 330 in the disk drive 300 and improving the read performance of pre-reading.

Further, when redundant data is included in the pre-read target range, pre-read can be performed efficiently, and the read performance of pre-read can be improved.

Also, the disk drives 300a to 300
The effect that the read performance of the pre-read in the collection read at the time of occurrence of a failure in one of the g can be improved.

The disk drives 300a-300
g and the spare disk drive 300h, the drive copy and the collection copy can be efficiently performed, and the time required for the copy can be reduced.

(Embodiment 2) Next, an example of a second embodiment of the data storage device of the present invention will be described with reference to FIG.

The device configuration of FIG. 11 in the second embodiment is such that the parity generator control means 34, the redundant data generator 130, and the copy table 160 are removed from FIG. The input / output address from the central processing unit 10 and the address of the disk drive 300 correspond one-to-one, and the mapping operation unit 3 illustrated in FIG.
9 and the mapping illustrated in FIG. 3 are unnecessary. The disk control device 20 when a read request is issued from the host to the disk storage device having such a configuration.
The operation of the microprocessor 30 is as shown in the flowcharts illustrated in FIGS. 7, 8, and 9 described above. However, the redundant data skip processing in step 4200 in FIG. 8 is not required. Steps 6200 to 6500 in FIG. 9 are unnecessary because redundant data does not exist. The detailed operation is the same as in the first embodiment.

According to the disk storage device of the present embodiment, the address of input / output data exchanged with the central processing unit 10 and the storage position on the disk drive 300 have a one-to-one correspondence. In a disk storage device having a non-redundant storage configuration, when read-ahead is performed while a plurality of data read request processes are operating at the same time, it is possible to improve the use efficiency of the buffer in the drive and improve the read performance of the read-ahead. Is obtained.

Next, an example of a method for judging whether various functions of the present invention as exemplified in the above-described embodiments are implemented by observing the operation of the data storage device from the outside, for example. explain.

For example, in the case of the device having the configuration shown in FIG. 1, when the drive path 70 is a SCSI bus,
A SCSI monitor 400 is connected to the drive path 70 of one drive constituting the ECC group.
A plurality of data sets 1,. . .
Create n.

Then, each operation under the following confirmation conditions # 1 to # 6 is intentionally performed.
Via 00, the behavior of the device is observed. However, in the following description, the area of the data set m (1 ≦ m ≦ n) on the drive 300a to which the SCSI monitor 400 is connected is
Called area m.

<Confirmation Condition # 1> A data storage device having the redundant storage configuration illustrated in FIG. 1 or FIG. 4 or the non-redundant storage configuration illustrated in FIG. Check whether the management is performed to divide into segments.

. The central processing unit 10 issues a read command to the area m with a sequential instruction.

[0068] After a sufficient time,
Issue read command to other area.

[0069] Read command is issued to the next address.

In the case of such a check condition # 1, if the above-described function of the present invention is implemented, the following observation result is obtained via the SCSI monitor 400.

That is, in the disk drive 300,
If the prefetch from the recording medium 350 to the drive buffer 330 is performed, the disk drive 300 has no overhead for head positioning, and starts data transfer after the disk drive 300 receives the read command. There is no time interval (read overhead) until the data is read.

<Confirmation Condition # 2> When a buffer on a disk drive is divided into a plurality of segments and managed, the access order of read requests from a higher-level device is managed for each segment, and the most recently accessed It is confirmed whether or not an operation (LRU management) for allocating the read segment to the prefetch corresponding to another read request is performed.

[0073] From the central processing unit 10, from the area 1 to m
Issue read commands with sequential instructions. However, a sufficient time is required between the issuance of each read command.

[0074] Issue read command for area 1.

[0075] Issue read command for area m.

In the case of this confirmation condition # 2, when LRU management is performed in a plurality of segments of the drive buffer 330, since the prefetch data for the area 1 has already been overwritten, the read overhead corresponding to the head positioning time Occurs.

In addition, there is prefetch data for the area m, and no read overhead occurs.

<Confirmation Condition # 3> When a plurality of prefetches corresponding to a plurality of read requests are operating in parallel, the priority of the prefetch is determined based on the access frequency of the higher-level device to the prefetched data. Check whether the operation of assigning a segment from a high look-ahead is performed.

[0079] Issue a read command with sequential instructions to data sets 1 to n. At this time,
The number of accesses for each data set is expressed as data set 1 <
Data set 2. . . <Data set n.

[0080] The read command is issued once with the sequential instruction in the order of 1 from the area n.

[0081] Issue read command for area 1.

[0082] Issue read command for area n.

At this time, when segments are allocated in accordance with the access frequency for each data set as in the present invention, prefetching to the area 1 is not performed, so that the readout corresponding to the head positioning time Overhead occurs.

Since the prefetch for the area n is performed, the read overhead corresponding to the head positioning time does not occur.

Normally, since the data set is not conscious in SCSI, it is possible to confirm by this method.

<Confirmation Condition # 4> In the data storage device having the redundant storage configuration, it is confirmed whether or not the skip operation of the redundant data is performed at the time of pre-reading by the disk drive.

A read command is sequentially issued to consecutive addresses with a sequential instruction for the data set 1 in the ECC group having the RAID 5 configuration.

At this time, in the case of a RAID 5 configuration, each area is divided by redundant data. Therefore, when the redundant data skip function is not provided as in the present invention, a read overhead corresponding to the head positioning time occurs in reading the area 1. In other words, when the redundant data skip function according to the present invention is provided, the read overhead corresponding to the head positioning time does not occur when reading the area 1.

<Confirmation Condition # 5> In the data storage device having the redundant storage configuration, it is confirmed whether or not the pre-read operation of the redundant data in the disk drive is performed at the time of executing the collection read.

. Disk drives 300 b to 30 b other than the disk drive 300 a to which the SCSI monitor 400 is mounted in the RAID 5 configuration ECC group
One of the 0g is removed to intentionally set an error (collection read execution) state.

[0091] A read command is issued sequentially to consecutive addresses with a sequential instruction for data set 1.

At this time, when the prefetch of the redundant data is performed at the time of the collection read as in the present invention, the read overhead does not occur in the read of the area 1.

<Confirmation Condition # 6> In the data storage device having the redundant storage configuration, it is confirmed whether the read-ahead of the disk drive at the time of executing the drive copy is executed.

Drive copy is executed using the disk drive 300a on which the SCSI monitor 400 is mounted as a copy source.

At this time, when prefetching is performed at the time of drive copying as in the present invention, no reading overhead occurs in the disk drive 300a on which the SCSI monitor 400 is mounted.

As described above, whether or not various functions of the present invention are implemented can be confirmed through the SCSI monitor 400.

Although the invention made by the inventor has been specifically described based on the embodiments, the present invention is not limited to the above-described embodiments, and can be variously modified without departing from the gist thereof. Needless to say, there is.

For example, the disk drive is not limited to a magnetic disk drive or the like having a magnetic disk as a rotary storage medium.
A disk drive having a general rotary storage medium such as a VD-RAM can be widely used.

[0099]

According to the data storage device of the present invention, when read-ahead is performed while a plurality of data read request processes are operating simultaneously, the use efficiency of the buffer in the drive is improved and the read performance of the read-ahead is improved. Can be improved,
The effect is obtained.

Further, in the data storage device having the redundant storage configuration, when redundant data is included in the prefetch target range, the prefetch can be efficiently performed and the prefetch read performance can be improved.

Further, there is an effect that the read performance of the pre-read in the collection read in the data storage device having the redundant storage configuration can be improved.

In the data storage device having the redundant storage configuration, the drive copy and the correction copy can be performed efficiently, and the time required for the copy can be shortened.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram illustrating an example of a configuration of an information processing system including a data storage device according to a first embodiment of the present invention.

FIG. 2 is a conceptual diagram showing an example of a configuration of a disk drive constituting a disk storage device according to an embodiment of the present invention.

FIG. 3 is a conceptual diagram showing an example of a method of mapping input / output data with a central processing unit to a disk drive in a disk storage device having a redundant storage configuration according to an embodiment of the present invention.

FIG. 4 is a conceptual diagram showing an example of a more detailed configuration of the information processing system according to the embodiment illustrated in FIG. 1;

FIG. 5 is a conceptual diagram showing an example of a configuration of a look-ahead table used in a disk storage device according to an embodiment of the present invention.

FIG. 6 is a conceptual diagram showing an example of the configuration of a copy table used in a disk storage device according to an embodiment of the present invention.

FIG. 7 is a flowchart showing an example of the operation of the storage control device constituting the disk storage device according to the embodiment of the present invention, together with FIG. 8;

FIG. 8 is a flowchart showing an example of the operation of the storage control device constituting the disk storage device according to the embodiment of the present invention, together with FIG. 7;

FIG. 9 is a flowchart illustrating an example of an operation of a drive constituting the disk storage device according to the embodiment of the present invention;

FIG. 10 is a flowchart showing an example of an operation of a drive constituting the disk storage device according to the embodiment of the present invention.

FIG. 11 is a conceptual diagram illustrating an example of a configuration of a data storage device according to a second embodiment of the present invention.

[Explanation of symbols]

10 central processing unit, 20 disk control unit, 30
Microprocessor, 31: cache memory reference means, 32: cache directory reference means, 33: drive status reference means, 34: parity generator control means,
35 ... difference data generator control means, 36 ... drive controller control means, 38 ... sequential access mode determination means, 39 ... mapping operation means, 40 ... host I / F,
50: data division / reproducer, 60: channel path, 70
... drive path, 80 ... drive controller, 110
... cache memory, 111 ... write surface, 112 ... read surface, 113-118 ... slot, 120 ... cache directory, 130 ... redundant data generator, 140 ... shared memory, 150 ... prefetch table, 150a ... logical device number, 150b ... Data set range, 150c ...
Previous host access position, 150d: Host access count, 150e: Previous access time, 150f: Average access interval, 150g: Prefetch order, 150h: Prefetch execution flag, 150i: Target segment number, 160
... copy table, 160a ... copy target drive number, 160b ... copy pointer, 160c ... segment number, 231-1 to 231-2 ... host adapter, 23
2-1 to 232-2 ... cache memory, 233-1 to
233-4: Disk adapter, 234-1 to 234-
2. Shared memory 235 Service processor 236
... Communication bus in the control unit, 237-1 to 237-2 ... data transfer path, 240 ... storage unit, 242-1 to 242-
32 ... Disk drive, 242-33 to 242-64
... disk drive, 250 ... maintenance terminal, 260-1
260-8: Channel path, 270-1 to 270-4 ...
Drive path, 270-5 to 270-8 ... drive path, 270-9 to 270-12 ... drive path, 270
-13 to 270-16 ... drive path, 300 (300
a to 300 g): Disk drive, 300 (300
h) spare disk drive, 310 disk I /
F, 320: microprocessor, 321: disk I
/ F control means, 322... Drive buffer reference means, 3
23: drive buffer management means, 324: disk controller control means, 330: drive buffer, 340: disk controller, 350: recording medium, 360 to 376
... Segment, 400 ... SCSI monitor.

Claims (3)

[Claims] 1. A drive provided with a rotary storage medium for storing data, and a storage interposed between the drive and a host device for controlling transmission and reception of the data between the drive and the host device. A data storage device including a control device, a buffer provided in the drive and temporarily storing the data read from the rotary storage medium, and managing the buffer by dividing the buffer into a plurality of segments. And a buffer management control logic for executing, for each of a plurality of different read requests generated from the higher-level device, pre-reading of the data from the rotary storage medium to the buffer according to the read request by using the different segments. A data storage device comprising: 2. The data storage device according to claim 1, wherein the storage control device divides the write data received from the host device into a plurality of data units and generates at least one redundant data from the data units. A plurality of data units and the redundant data are distributed and stored in different drives, and when a failure occurs in any of the drives during reading of the data, the data units read from the remaining drives and The buffer management control logic performs a correction read for restoring a failed data unit from the redundant data, and the buffer management control logic executes a prefetch to the segment of the redundant data when the correction read is performed. Otherwise, the prefetch target range is the redundant data A first operation for pre-reading the data from the next address of the redundant data, a first drive for storing data, and a spare second drive other than the first drive. Recovering data on the failed first drive from the remaining first drive when the device becomes inoperable due to a failure, and storing the recovered data on a spare second drive; Performing at least one of a drive copy in which all data on one of the first drives is copied to a spare second drive; and assigning the segment preferentially to the correction copy and the drive copy. 2. A data storage device, which performs at least one of the operations of 2. 3. The data storage device according to claim 1, wherein the buffer management control logic manages an access order of the read request from the higher-level device for each of the segments, and accesses the oldest one. And assigning the read segment to the prefetch corresponding to another read request.
When a plurality of prefetches corresponding to a plurality of the read requests are operating in parallel, the priority of the prefetch is determined by the access frequency of the higher-level device to the prefetched data, and the higher the priority, the higher the priority. A fourth operation of allocating the segment from prefetching, wherein at least one of the following operations is executed.