WO2018042608A1

WO2018042608A1 - Storage unit and control method therefor

Info

Publication number: WO2018042608A1
Application number: PCT/JP2016/075734
Authority: WO
Inventors: 英美子藤松; 開道相浦; 純二岩崎
Original assignee: 株式会社日立製作所
Priority date: 2016-09-01
Filing date: 2016-09-01
Publication date: 2018-03-08

Abstract

[Problem] To propose a highly reliable storage unit, and a control method therefor, capable of readily and optimally improving performance in accordance with conditions of the storage unit. [Solution] A storage unit in which pools are formed by one or a plurality of logical volumes provided by one or more storage device groups, and which, in response to a request from a host device to write to a virtual volume, allocates a storage region to the virtual volume, from a pool associated with the virtual volume, wherein: the read cache hit rate for each storage device group and the frequencies of read access and write access to each storage device group are obtained; and if the read cache hit rate for a storage device group is equal to or less than a first threshold, the capacity of the cache region allocated to the pool used by the storage device group to provide logical volumes is increased, and if the frequency of read access or write access to the storage device group exceeds a second threshold, a new logical volume is added to the pool.

Description

Storage apparatus and control method thereof

The present invention relates to a storage apparatus and a control method thereof, and is particularly suitable when applied to a storage apparatus equipped with a virtualization function.

Conventionally, there is a virtualization function as one of the functions of a storage device. The virtualization function provides a virtual logical volume (hereinafter referred to as a virtual volume) to the host device, and when a write request is given from the host device to an unused storage area in the virtual volume. This is a function for dynamically allocating a physical storage area to the storage area.

In a storage device equipped with such a virtualization function, a predetermined number of hard disk devices are managed as a RAID (Redundant Arrays of Inexpensive Disks) group, and logical volumes defined on storage areas provided by the RAID group (hereinafter referred to as this). Is called a pool volume) or a plurality of them are managed as one pool. The physical storage area is assigned to the virtual volume by cutting out a storage area of a predetermined size from this pool.

Such a virtualization function has the advantage that it is not necessary to prepare in advance a hard disk device for the capacity of the virtual volume provided to the host device, and the introduction cost of the storage device can be reduced.

Japanese Patent No. 4473175 JP 2008-217575 A

By the way, in a storage device equipped with such a virtualization function, when all the RAID groups that provide pool volumes to the pool become highly loaded, the response speed of the RAID group to read access and write access from the host device decreases. However, there is a problem that the performance of the entire storage apparatus is degraded.

As a method for solving such a problem, for example, in Patent Document 1, the performance of a plurality of disk devices is periodically measured while the device is operating, and the performance degradation of any one of the disk devices is detected based on the measured performance value. In this case, it is disclosed that a disk device is added and data of an existing disk device is rearranged in the added disk device.

Further, Patent Document 2 monitors the access load on a logical disk belonging to a storage pool, and when a logical disk with a high access load is detected, a storage pool having a necessary capacity is secured to store the logical disk. A method for moving data is disclosed.

However, according to the performance improvement method disclosed in Patent Document 1, there is a problem that sufficient performance improvement cannot be performed by simply adding a disk device depending on the status of the storage device. According to the improvement method, there has been a problem that a corresponding load is generated in the storage apparatus in order to move the storage data of the logical disk to a new storage pool.

The present invention has been made in consideration of the above points, and an object of the present invention is to propose a highly reliable storage apparatus and its control method capable of easily performing the optimum performance improvement according to the situation of the storage apparatus.

In order to solve such a problem, in the present invention, one or a plurality of logical volumes are defined on a storage area provided by a storage device group composed of one or a plurality of storage devices, and one or a plurality of the storage device groups are defined. The pool is formed by one or more logical volumes to be provided, provides a virtual volume to a host device, and is associated with the virtual volume in response to a data write request from the host device to the virtual volume In a storage device that dynamically allocates storage areas from the virtual volume to the virtual volume, the storage area is logically divided into a plurality of areas, and the area is used as a cache area for temporarily storing data read from and written to the logical volume Cache memory allocated to the pool, and for each pool, A cache hit rate, which is a ratio in which read target data is stored in the cache area allocated to the pool at the time of read access in the device group, and the frequency of the read access and write access in each of the storage device groups, respectively When the cache hit rate at the time of the read access in the storage device group is equal to or less than a predetermined first threshold for each of the performance monitoring unit to be acquired and the pool, the storage device group provides the logical volume. The capacity of the cache area allocated to the pool is added, and if the frequency of the read access or the write access exceeds a predetermined second threshold, a new logical volume is added to the pool A performance improvement unit was provided.

In the present invention, one or more logical volumes are defined on a storage area provided by a storage device group composed of one or more storage devices, and one or more provided by one or more storage device groups. A pool is formed by the logical volume of the virtual volume, the virtual volume is provided to the host device, and the virtual volume is transferred from the pool associated with the virtual volume in response to a data write request from the host device to the virtual volume. In the storage apparatus control method for dynamically allocating a storage area, the storage apparatus is a cache in which the storage area is logically divided into a plurality of areas, and the area temporarily stores data read and written to the logical volume. A cache memory allocated to the corresponding pool as an area; For each pool, the cache hit rate, which is the ratio at which the read target data is stored in the cache area allocated to the pool at the time of read access in each storage device group, and each storage device group A first step of acquiring the frequency of the read access and the write access in the storage device, and the storage device has a predetermined first cache hit rate at the time of the read access in the storage device group for each pool. If the threshold is less than or equal to the threshold, the storage device group adds the capacity of the cache area allocated to the pool that provides the logical volume, and the frequency of the read access or the write access exceeds a predetermined second threshold. If it exceeds, add a new logical volume to the pool. Was provided and the steps.

According to this storage device and its control method, for each pool, the write performance is improved by adding a new logical volume, while the read performance depends on the pool status. Thus, the performance of the pool is improved by adding the capacity of the cache area of the pool and / or adding a new logical volume to the pool.

According to the present invention, it is possible to realize a highly reliable storage apparatus and its control method that can easily perform the optimum performance improvement according to the situation of the storage apparatus.

It is a block diagram which shows the whole structure of the computer system by this Embodiment. It is a conceptual diagram with which it uses for description of the program and various tables which were stored in the local memory. It is a conceptual diagram which shows the structural example of a RAID group performance information table. It is a conceptual diagram which shows the structural example of a RAID group information management table. It is a conceptual diagram which shows the structural example of a pool information management table. It is a conceptual diagram which shows the structural example of a CLPR management table. It is a conceptual diagram which shows the structural example of a batch schedule management table. It is a flowchart which shows the process sequence of a pool performance monitoring process. It is a conceptual diagram with which it uses for description of a pool performance monitoring process. It is a block diagram with which it uses for description of a pool performance monitoring process. (A) to (C) are charts for explaining the pool performance monitoring process. It is a flowchart which shows the process sequence of an access classification discrimination | determination process. It is a flowchart which shows the process sequence of a 1st read performance improvement process. It is a flowchart which shows the process sequence of a cache area | region addition process. It is a flowchart which shows the process sequence of a pool volume addition process. It is a flowchart which shows the process sequence of the allocated pool volume reallocation process. It is a flowchart which shows the process sequence of a 2nd write performance improvement process. It is a flowchart which shows the process sequence of a 2nd read performance improvement process. It is a flowchart which shows the process sequence of a read / write performance improvement process. It is a flowchart which shows the process sequence of a read / write performance improvement process. It is a flowchart which shows the process sequence of a batch improvement process. It is a flowchart which shows the process sequence of a batch post-process. It is a flowchart which shows the process sequence of a cash return process.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

(1) Configuration of Computer System According to this Embodiment In FIG. 1, reference numeral 1 denotes a computer system according to this embodiment as a whole. The computer system 1 includes a host device 2 that executes processing according to a user's job, and a storage device 3 that is equipped with a virtualization function. The host device 2 and the storage device 3 communicate with each other through a communication path 4. Connected through.

The host device 2 is a general-purpose server device provided with information processing resources such as a CPU 10, a memory 11, and an interface 12, and the CPU 10 executes application software 14 stored in the memory 11, whereby data to the storage device 3 is stored. Predetermined business processing including reading and writing is executed. The interface 12 is a hardware device having a function of performing protocol control when communicating with the storage apparatus 3 via the communication path 4.

The storage device 3 includes a data storage unit 21 including a plurality of storage devices 20 and a controller 22 that controls data input / output with respect to the data storage unit 21.

The storage device 20 of the data storage unit 21 is composed of a large-capacity nonvolatile storage device such as a hard disk device or an SSD (Solid State Drive). In the following description, it is assumed that the storage device 20 is composed of a hard disk device. A RAID group RG is formed by one or a plurality of storage devices 20, and one or a plurality of pool volumes PLVOL are created in a storage area provided by the storage devices 20 constituting one RAID group RG. A pool PL is formed by one or a plurality of pool volumes PLVOL provided by one or a plurality of RAID groups RG.

The virtual volume VVOL provided by the storage device 3 to the host device 2 is associated with one of the pools PL. When a write request for an unused storage area in the virtual volume VVOL is given from the host device 2, a predetermined size is generated from any pool volume PLVOL in the pool PL associated with the virtual volume VVOL. A storage area (hereinafter referred to as a real storage area) is cut out and assigned to the write destination storage area in the virtual volume VVOL, and the write data is stored in the real storage area.

The controller 22 includes information processing resources such as a host interface 23, a disk interface 24, a processor 25, a local memory 26, and a cache memory 27. The host interface 23 is a hardware device having a function of performing protocol control when communicating with the host device 2 via the communication path 4, and the disk interface 24 is communicating with the storage device 20 in the data storage unit 21. This is a hardware device having a function of performing protocol control.

The processor 25 is a hardware device having a function for controlling the operation of the entire storage apparatus 3, and the local memory 26 is mainly used to hold various programs and various data. Various processes of the entire storage apparatus 3 are executed by the processor 25 executing the program stored in the local memory 26.

The cache memory 27 is used to temporarily store data read / written from / to the data storage unit 21. In the case of the present embodiment, the storage area of the cache memory 27 is logically divided and managed in a plurality of areas called CLPR (CacheitionLogical 管理 Partition), and these CLPR are read / write target for the virtual volume VVOL. Each pool PL is allocated as a cache area for temporarily storing data.

(2) Performance improvement function and batch improvement function according to this embodiment Next, the performance improvement function and the batch improvement function installed in the storage apparatus 3 of this embodiment will be described.

In the storage device 3, the data written from the host device 2 to the virtual volume VVOL is temporarily stored in the cache area (CLPR) assigned to the pool PL associated with the virtual volume VVOL, and then stored in the storage device 3. The data is distributed and stored in each storage device 20 constituting the RAID group RG corresponding to when the load on the device 3 is low.

Thereafter, when a read request for the data is given from the host device 2 to the storage device 3, if the data remains in the cache area (cache hit), the data is read from the cache area. When the data does not remain in the cache area (cache miss), the data is read from the corresponding storage device 20 to the cache area (hereinafter referred to as staging). Thereafter, the data is transferred from the cache area to the host device 2.

In this case, in order to stage the data to be read, a corresponding time is required, and the response performance to the read request is reduced accordingly. Therefore, when the rate of cache hits during read processing (hereinafter referred to as the cache hit rate) is low, by increasing the capacity of the cache area allocated to the pool PL and increasing the amount of data that can be held in the cache area, The cache hit rate during read processing can be increased, and as a result, read performance can be improved.

On the other hand, when each RAID group RG that provides the pool volume PLVOL to the pool PL becomes heavily loaded, the read / write performance of the pool PL becomes worse because the speed of reading / writing data from / to the storage device 20 becomes slow.

In such a case, the pool volume PLVOL of the RAID group RG that does not provide the pool volume PLVOL is added to the pool PL, and each pool volume that has constituted the pool PL until then is added. By moving a part of the data respectively stored in the PLVOL to the added pool volume PLVOL, the load on each RAID group RG that provides the pool volume PLVOL to the pool PL can be reduced. As a result, the write performance and read performance of the pool PL can be improved.

Therefore, in the storage apparatus 3 of the present embodiment, for each pool PL defined in the apparatus, for each RAID group RG that provides the pool volume PLVOL to the pool PL, at least the unit time during the read process and the write process Acquire the number of accesses per second (hereinafter referred to as the access frequency) and the cache hit rate during read processing, and improve the performance of the pool PL based on these information The performance improvement function which performs is carried.

In practice, the storage apparatus 3 determines the write performance and read performance of the pool PL based on the information acquired as described above. If it is determined that the write performance has decreased, a new pool volume is added to the pool PL. While improving the performance of the pool PL by adding a PLVOL, if it is determined that the read performance has deteriorated, the capacity of the cache area allocated to the pool PL is determined according to the status of the pool PL. The performance of the pool PL is improved by increasing and / or adding a new pool volume PLVOL to the pool PL.

The storage device 3 also includes a pool PL that is used in the batch processing so that the batch processing scheduled in the host device 2 does not end within the scheduled time, so that the batch processing ends within the scheduled time. A batch improvement function for increasing the capacity of the cache area to be assigned to or adding a new pool volume PLVOL to the pool PL is also installed.

As a means for realizing such performance improvement function and batch improvement function, the local memory 26 (FIG. 1) of the controller 22 (FIG. 1) of the storage apparatus 3 has a performance as a program as shown in FIG. A monitoring program 30, an access type determination program 31, a performance improvement processing program 32, and a batch improvement program 33 are stored. As a management database 34, a RAID group performance information table 35, a RAID group information management table 36, a pool information management table 37, a CLPR A management table 38 and a batch schedule management table 39 are stored.

The performance monitoring program 30 acquires performance information for each RAID group RG defined in the storage device 3, monitors the performance of each pool PL in the storage device 3 based on the acquired performance information, and if necessary This is a program having a function of executing processing for improving the performance of the pool PL.

In practice, the performance monitoring program 30 performs, for each RAID group RG, the access frequency for each of the four types of access such as random read, sequential read, random write, and sequential write for that RAID group RG, and the cache hit for each access type. Rate and the usage rate of the storage device 20 in the RAID group RG (the ratio of the used capacity to the total capacity of the RAID group RG, hereinafter referred to as the storage device usage rate) is a unit time (1 second) The acquired information is stored in the RAID group performance information table 35 corresponding to the RAID group RG and managed.

The performance monitoring program 30 monitors the performance of each pool PL based on the information stored in the RAID group performance information table 35, and determines that the performance of the pool PL has deteriorated. 31 and the performance improvement processing program 32 and / or the batch improvement program 33 are called to execute processing for improving the performance of the pool PL.

The access type determination program 31 stores the access frequency for each of the above four types of access types stored in each RAID group performance information table 35, and a threshold value for the access frequency preset for each access type (hereinafter referred to as these). Based on the first IOPS (Input Output Per Second) threshold), the access type whose access frequency exceeds the corresponding first IOPS threshold is specified, and based on the specified result, the performance to be executed at that time This is a program having a function of instructing the performance improvement processing program 32 for improvement processing.

The performance improvement processing program 32 is a program having a function of executing various performance improvement processes for improving the performance of the pool PL in response to an instruction from the access type determination program 31. Further, the batch improvement program 33 performs the performance of the pool PL used in the batch processing so that the batch processing is completed within the time when the batch processing executed by the host device 2 is not completed within the scheduled time. It is a program having a function to improve.

Meanwhile, the RAID group performance information table 35 is a table used for holding and managing the performance information of the RAID group RG acquired by the performance monitoring program 30, and is created for each RAID group RG. As shown in FIG. 3, the RAID group performance information table 35 includes a serial number column 35A, an information acquisition time column 35B, a read related information column 35C, a write related information column 35D, and a storage device usage rate column 35E. .

In the serial number column 35A, serial numbers assigned to registration information (hereinafter referred to as record information) for each row (hereinafter referred to as record) of the RAID group performance information table 35 are stored. In the information acquisition time column 35B, the date and time when the record information is acquired is stored. The storage device usage rate column 35E stores the storage device usage rate at the time when the record information in the corresponding RAID group RG is acquired.

The read related information column 35C includes a random read IOPS column 35CA, a random read cache hit rate column 35CB, a sequential read IOPS column 35CC, and a sequential read cache hit rate column 35CD.

In the random read IOPS column 35CA, the number of random read accesses made to the corresponding RAID group RG in the time zone from the time when the record immediately before that record was acquired to the time when the record was acquired. (That is, the random read access frequency) is stored, and the random read cache hit rate column 35CB stores the cache hit rate at the time of random read access in that time zone. The sequential read IOPS column 35CC stores the number of sequential read accesses made to the corresponding RAID group RG in that time zone (that is, the sequential read access frequency), and is stored in the sequential read cache hit rate column 35CD. Stores the cache hit rate at the time of sequential read access in that time zone.

Similarly, the write related information column 35D includes a random write IOPS column 35DA, a random write cache hit rate column 35DB, a sequential write IOPS column 35DC, and a sequential write cache hit rate column 35DD.

In the random write IOPS column 35DA, the number of random write accesses made to the corresponding RAID group RG in the time zone from the time when the record immediately before that record was acquired to the time when the record was acquired. (That is, the random write access frequency) is stored, and the random write cache hit rate column 35DB stores the cache hit rate at the time of random write access in that time zone. The sequential write IOPS column 35DC stores the number of sequential write accesses (that is, the sequential write access frequency) performed for the corresponding RAID group RG during the time period, and the sequential write cache hit rate column 35DD stores the number of sequential write accesses. Stores the cache hit rate at the time of sequential write access in that time zone.

Accordingly, in the example of FIG. 3, the corresponding RAID group RG has a random read access frequency of “2009/11/14「 0:00 ”to“ 2009/11/14 0:01 ”. 1900 ”, the cache hit rate is“ 33 ”%, the sequential read access frequency is“ 780 ”, the cache hit rate is“ 9 ”%, and the random write access frequency is“ 90 ”. The cache hit rate is “93”%, the sequential write access frequency is “199”, the cache hit rate at that time is “55”%, and the average usage rate of each storage device 20 constituting the RAID group RG Was 30%.

The RAID group information management table 36 is a table used for managing the configuration information of each RAID group RG created in the storage apparatus 3, and as shown in FIG. A configured logical volume number column 36B, a configuration pool number column 36C, a RAID level column 36D, a storage device rotation number column 36E, a storage device number column 36F, and a logical volume capacity column 36G. In FIG. 4, the RAID group RG is abbreviated as “RG”, and the logical volume is abbreviated as “LU”. The same applies to the other drawings.

The created logical volume number column 36B stores the identification numbers (logical volume numbers) of all the logical volumes created in the storage device 3, and the corresponding logical volume is created in the RAID group number column 36A. The identification number (RAID group number) of the RAID group RG is stored, and when the corresponding logical volume configures any pool PL as a pool volume, the identification number of the pool PL is stored in the configuration pool number column 36C. (Pool number) is stored.

Further, the RAID level column 36D stores the RAID level of the corresponding RAID group RG, and the storage device rotation speed column 36E stores each storage device 20 constituting the RAID group RG (in the present embodiment as described above). The number of rotations per unit time of the hard disk device) is stored. Further, the number of storage devices 20 constituting the RAID group RG is stored in the storage device number column 36F, and the capacity of the corresponding logical volume is stored in the logical volume capacity column 36G.

Therefore, in the example of FIG. 4, the RAID group RG “1-1” has a RAID level “RAID5” composed of “5” storage devices 20 whose rotation speed per unit time is “1500”. In the storage area provided by this RAID group RG, a logical volume with a logical volume number of “00:01:01” with a capacity of “1200” GB and a “volume” with a capacity of “800” GB 00:01:02 ”is created, and both of these logical volumes are assigned to a pool PL having a pool number of“ 1 ”(which constitutes the pool PL). It has been shown.

The pool information management table 37 is a table used for managing configuration information of each pool PL defined in the storage apparatus 3, and as shown in FIG. 5, a pool number column 37A and a RAID group number column 37B. The allocated capacity column 37C, the unallocated capacity column 37D, the threshold excess flag column 37E, and the CLPR number column 37F are provided.

The RAID group number column 37B stores the RAID group number of each RAID group RG defined in the storage apparatus 3, and the pool number column 37A stores the logical volume provided by the corresponding RAID group RG as a pool volume. The identifier (pool number) of the pool PL to be included is stored.

The allocated capacity column 37C stores the total capacity of the storage areas allocated to the virtual volume VVOL (FIG. 1) associated with the corresponding pool PL out of the total capacity of the corresponding RAID group RG. The allocated capacity column 37D stores the total capacity of the storage area that is not yet allocated to the virtual volume VVOL out of the total capacity of the RAID group RG.

Further, in the threshold excess flag column 37E, whether or not the storage device usage rate of the corresponding RAID group RG has exceeded a preset threshold value (hereinafter referred to as a storage device usage rate threshold value) in the performance improvement process described later. An over-threshold flag is stored. The threshold excess flag is initially set to “0” and is updated to “1” when the storage device usage rate of the corresponding RAID group RG exceeds the storage device usage rate threshold.

Further, the CLPR number column 37F stores an identifier (CLPR number) unique to the CLPR assigned to the CLPR in the cache memory 27 (FIG. 1) allocated as a cache area to the corresponding pool PL.

Therefore, in the case of the example of FIG. 5, for example, the pool PL to which the pool number “1” is assigned is provided by each RAID group RG to which the RAID group number “1-1” or “1-2” is assigned, respectively. Among these, the RAID group RG “1-1” has “1600” GB assigned to the corresponding virtual volume VVOL and “0” GB assigned to the unassigned capacity. In other words, the storage device usage rate has already exceeded the storage area usage rate threshold (the value of the threshold excess flag is “1”). Further, it is indicated that a CLPR assigned with a CLPR number “1” in the cache memory 27 is assigned to the pool PL as a cache area.

The CLPR management table 38 is a table used for managing the CLPR assigned to each pool PL. As shown in FIG. 6, the CLPR number column 38A, the cache management number column 38B, the cache capacity column 38C, A cache write wait rate field 38D, a temporary allocation flag field 38E, and an allocation destination pool field 38F are provided.

The CLPR number column 38A stores the CLPR number of the corresponding CLPR, and the cache management number column 38B stores the management number (cache management) assigned to each divided area when the corresponding CLPR is divided and managed. Number) is stored. The cache capacity column 38C stores the capacity of the corresponding divided area in the corresponding CLPR.

Further, in the cache write wait rate field 38D, the ratio of data waiting to be written to the storage device 20 out of the data temporarily stored in the corresponding CLPR (hereinafter referred to as the cache write wait rate). Is stored.

Further, in the temporary allocation flag field 38E, the corresponding CLPR or CLPR divided area (hereinafter referred to as CLPR or the like) is assigned to a pool PL other than the pool PL to which the CLPR is allocated in the performance improvement process described later. A temporary lending flag indicating whether or not the lending is temporarily lent or assigned is stored. The temporary lending flag is set to “0” when the corresponding CLPR or the like is not lent or assigned to another pool PL, and the CLPR or the like is temporarily lent or assigned to another pool PL. Is set to "1".

Further, the assigned pool field 38F stores the pool number of the destination pool PL that has been lent or assigned when the corresponding CLPR or the like is temporarily lent or assigned to another pool PL. Is done.

Therefore, in the example of FIG. 6, a part of the storage area of the cache memory 27 is divided into three CLPRs each assigned with a CLPR number of “CLPR1”, “CLPR2”, or “CLPR3”. The area called “CLPR2” is divided into two divided areas of “2-1”, “2-2”, or “2-3” with a capacity of “8” GB, and the entire CLPR is waiting for a cache read. The rate is shown to be “20%”. Of these divided areas, only the divided area “2-1” is lent to the pool PL with the pool number “1” (the value of the temporary allocation flag is “1”), and the other divided areas are It is also shown that it is not lent to another pool PL (the value of the temporary allocation flag is “0”).

In FIG. 6, the CLPR in which the character string “CLPR5 (reserved)” is stored in the CLPR number column 38A is a reserved CLPR (hereinafter referred to as “reserved CLPR”).

The batch schedule management table 39 is a table used for managing a schedule of batch processing periodically executed by the host device 2 using the storage device 3, and as shown in FIG. And one or a plurality of batch processing columns 39A provided in association with the respective batch processings set in FIG.

Each batch processing column 39A is divided into a time column 39B, a day-of-week column 39C, a used pool column 39D, a used RAID group column 39E, and an in-time processing completion flag column 39F. The time column 39B, the day of week column 39C, the used pool column 39D, and the used RAID group column 39E are used when the corresponding batch processing is scheduled to be executed, the day of the week, and the batch processing is executed. The pool number of the pool PL and the RAID group number of the RAID group RG used when executing the batch processing are stored. The in-time process completion flag column 39F stores a flag indicating whether or not the last executed batch process has been completed within the scheduled time (hereinafter referred to as an in-time process completion flag). . The in-time processing completion flag is set to “1” if the batch processing can be completed within the scheduled time, and the batch processing cannot be completed within the scheduled time. Is set to “0”.

Therefore, in the case of the example of FIG. 7, the batch process “batch process 1” is performed by the pool volume PLVOL to the pool PL to which the pool number “1” is assigned in the “weekday” “0:00 to 3 o'clock” time zone. The RAID group RG that is assigned the RAID group numbers “1-1” and “1-2” of the RAID groups RG that provide the data is executed within the scheduled time for the batch process that was executed last. (The value of the in-time processing completion flag is “1”).

(4) Various Processes Related to Performance Improvement Function According to this Embodiment Next, specific processing contents of various processes related to the performance improvement function according to this embodiment will be described. In the following description, the processing subject of various processes will be described as a “program”, but it goes without saying that the processor 25 of the storage apparatus 3 actually executes the process based on the program.

(4-1) Pool Performance Monitoring Processing FIG. 8 shows a processing procedure for pool performance monitoring processing that is executed regularly or irregularly by the performance monitoring program 30 (FIG. 2). The performance monitoring program 30 monitors the current performance of each pool PL defined in the storage apparatus 3 according to the processing procedure shown in FIG.

In practice, when the performance monitoring program 30 starts this pool performance monitoring process, first, referring to the pool information management table 37 (FIG. 5), the pool PL defined in the storage apparatus 3 is followed by step SP2 and subsequent steps. Selects one unprocessed pool PL (SP1).

Subsequently, the performance monitoring program 30 pools the RAID group numbers of all RAID groups RG that provide the pool volume PLVOL (FIG. 1) to the pool (hereinafter referred to as the target pool) PL selected in step SP1. Obtained from the information management table 37 (SP2).

Next, the performance monitoring program 30 acquires the performance information of each RAID group RG that has acquired the RAID group number in step SP2 from the RAID group performance information table 35 (FIG. 3) corresponding to the RAID group RG (SP3). The performance information acquired at this time is the record information of each record stored in the RAID group performance information table 35 from the start of the previous pool performance monitoring process to the start of the current pool performance monitoring process. It shall be.

In addition, the performance monitoring program 30 continuously uses the storage device usage rate for a certain period (for example, 5 minutes) in the RAID group RG that acquired the RAID group number in step SP2 based on the performance information acquired in step SP3. Then, it is determined whether or not there is a RAID group RG that exceeds the above-mentioned storage device usage rate threshold (SP4). If the performance monitoring program 30 obtains a negative result in this determination, it proceeds to step SP17.

In contrast, when the performance monitoring program 30 obtains a positive result in the determination at step SP4, each RAID group RG in the pool information management table 37 whose storage device usage rate exceeds the storage device usage rate threshold value for a certain period of time. Each of the threshold excess flags stored in the threshold excess flag column 37E corresponding to each is set to ON (“1”) (SP5).

Subsequently, the performance monitoring program 30 selects one RAID group RG that has not been processed after step SP7 from the RAID groups RG whose excess threshold flag is set to ON in the pool information management table 37 (SP6). ).

The performance monitoring program 30 extracts the performance information corresponding to the RAID group (hereinafter referred to as the target RAID group) RG selected in step SP6 from the performance information of each RAID group RG acquired in step SP3. Record information at all times when the storage device usage rate exceeds the storage device usage rate threshold value is acquired from the extracted performance information (SP7). Therefore, for example, in the example of FIG. 3, when the storage device usage threshold is “40”%, the performance monitoring program 30 has serial numbers “4” to “7” and “10” as shown in FIG. The record information of this record is acquired in this step SP7.

Subsequently, the performance monitoring program 30 refers to the record information (FIG. 9) of each record acquired in step SP7 on the premise that the batch processing is sequential access (sequential read or sequential write). An average value of random read access frequency and random write access frequency for the entire target pool PL at a time when the usage rate exceeds the storage device usage rate threshold value is calculated, and at least one of these calculated average values is individually It is determined whether or not the total value in the target pool PL of the first IOPS threshold value, which is the threshold value of the random read or random write access frequency defined according to the performance of each RAID group RG, is exceeded. (SP8).

For example, as shown in FIG. 10, the target pool PL includes a first pool volume PLVOL1 and a second pool volume PLVOL2 provided by the first RAID group RG1, and a third pool volume provided by the second RAID group RG2. PLVOL3, and performance limit values (processing per unit time) for random read and random write of the first RAID group RG1 calculated from the specifications of the storage devices 20 constituting the first RAID group RG1. The possible number) is “2110” or “379” as shown in FIG. 11A, the random read of the second RAID group RG2 calculated from the specifications of each storage device 20 constituting the second RAID group RG2, and The performance limit value for random write is “211” as shown in FIG. It shall be “0” or “379”. Further, the first IOPS threshold value for random read and random write is defined as 80% of the performance limit value for random read or random write, respectively. The numerical values described in the “actual average value” row in FIGS. 11A and 11B are stored as the storage device usage rate of the target RAID group RG (assuming that it is the first RAID group RG1). It is assumed that it is the average value of the access frequencies of the corresponding access types in the entire target pool PL at the time when the device usage rate threshold is exceeded.

In this case, the average value of the random read access frequency for the entire target pool PL may be the sum of the average values of the random read access frequencies of the RAID groups RG that respectively provide the pool volumes PLVOL constituting the target pool PL. 11 (A) and 11 (B), as shown in FIG. 11 (C), “2526”, which is the average value of random read access frequency in the first RAID group RG1, “3790” is obtained by adding together “1264” which is the average value of the random read access frequency in the second RAID group RG2. Similarly, the average random write access frequency for the entire target pool PL is “322”, which is the average random write access frequency in the first RAID group RG1, and the random write access in the second RAID group RG2. The sum of “157”, which is the average value of the access frequency, is “480”.

In addition, the value of the first IOPS threshold for random reads as seen in the entire target pool PL can be the sum of the first IOPS thresholds in each RAID group RG that provides the pool volume PLVOL that constitutes the target pool PL. In the example of FIGS. 11A and 11B, as shown in FIG. 11C, the first IOPS threshold “1688” for the random read in the first RAID group RG1, “3376” is obtained by adding up “1688” which is the first IOPS threshold for random reads in the RAID group 2 of the second RAID group. Similarly, the value of the first IOPS threshold value for the random write in the entire target pool PL is “303”, which is the first IOPS threshold value for the random write in the first RAID group RG1, and in the second RAID group RG2. “606” is obtained by adding up “303” which is the first IOPS threshold for random writing.

Therefore, in this step SP8, the performance monitoring program 30 sets “3790”, which is the average value of the random read access frequency as seen in the entire target pool PL, as the first IOPS threshold for random reads as seen in the entire target pool PL. 3480 ”or“ 480 ”that is the average value of the random write access frequency as seen in the entire target pool PL exceeds“ 606 ”that is the first IOPS threshold for random write as seen in the entire target pool PL It will be judged whether or not. In this example, since the average value of the random read access frequency seen in the entire target pool PL exceeds the first IOPS threshold for random reads seen in the entire target pool PL, the performance monitoring program 30 makes this determination. A negative result will be obtained. Thus, at this time, the performance monitoring program 30 proceeds to step SP13.

On the other hand, when the performance monitoring program 30 obtains a negative result in the determination at step SP8, it acquires all information related to each batch process scheduled in the host apparatus 2 from the batch schedule management table 39 (FIG. 7) ( (SP9) With reference to the in-time process management flag of each batch process, it is determined whether or not all scheduled batch processes are completed within the scheduled time when the batch process is executed for the last time (SP9). SP10).

When the performance monitoring program 30 obtains a positive result in this determination, it proceeds to step SP12. On the other hand, if the performance monitoring program 30 obtains a negative result in the determination at step SP10, the performance monitoring program 30 calls the batch improvement program 33 (FIG. 2) and pools it in the target pool PL so that all batch processing is completed within the scheduled time. A batch improvement process for adding the volume PLVOL or adding the capacity of the cache area allocated to the target pool PL is executed (SP11).

Subsequently, based on the performance information of the target RAID group RG acquired in step SP3, the performance monitoring program 30 calculates the average value of the storage device usage rate for each access type in the time zone when no batch processing is scheduled. It is calculated, and it is determined whether or not at least one of the calculated average values of the storage device usage rates for each access type exceeds the above-mentioned storage device usage rate threshold value (SP12). If the performance monitoring program 30 obtains a negative result in this determination, it proceeds to step SP14.

On the other hand, when the performance monitoring program 30 obtains a positive result in the determination at step SP12, the performance monitoring program 30 performs the performance improvement processing for improving the performance of the target RAID group RG as the access type determination program 31 (FIG. 2) and the performance improvement processing program 32. (FIG. 2) is executed (SP13).

Subsequently, the performance monitoring program 30 determines whether or not the processing of step SP7 to step SP13 has been executed for all RAID groups RG for which the threshold excess flag is set to ON in step SP5 (SP14).

If the performance monitoring program 30 obtains a negative result in this determination, it returns to step SP6, and then switches the RAID group RG selected in step SP6 to another unprocessed RAID group RG in sequence, step SP6 to step SP14. Repeat the process.

When the performance monitoring program 30 eventually obtains a positive result in step SP14 by completing the processing of step SP7 to step SP13 for all RAID groups RG for which the threshold excess flag is set to ON, the performance acquired in step SP3 Based on the information, the average value of the cache hit rate for the entire target pool PL is calculated, and the calculated average value of the cache hit rate is equal to or greater than a preset threshold value (hereinafter referred to as the cache hit rate threshold value). Is determined (SP15). If the performance monitoring program 30 obtains a negative result in this determination, it proceeds to step SP17.

In contrast, when the performance monitoring program 30 obtains a positive result in the determination at step SP15, the CLPR or the like is temporarily assigned to the target pool PL in the performance improvement process at step SP13. A cache return process is executed to release etc. (SP16).

Next, the performance monitoring program 30 determines whether or not the processing of step SP2 to step SP16 has been executed for all the pools PL in the storage apparatus 3 (SP17).

When the performance monitoring program 30 obtains a negative result in this determination, the performance monitoring program 30 returns to step SP1, and thereafter performs the processing of steps SP1 to SP17 while sequentially switching the pool PL selected in step SP1 to another unprocessed pool PL. repeat.

Then, when the performance monitoring program 30 finishes executing the processing of step SP2 to step SP16 for all the pools PL in the storage apparatus 3 and ends the positive result in step SP17, the performance monitoring program 30 ends this pool performance monitoring processing.

(4-2) Performance Improvement Processing (4-2-1) Access Type Determination Processing FIG. 12 shows the access executed by the access type determination program 31 (FIG. 2) at step SP13 of the pool performance monitoring processing described above with reference to FIG. The process procedure of a classification determination process is shown.

When the performance monitoring program 30 proceeds to step SP13 of the pool performance monitoring process, it calls the access type determination program 31 (FIG. 2). When called by the performance monitoring program 30, the access type determination program 31 uses the performance information of the target RAID group RG acquired from the start of the previous pool performance monitoring process to the start of the current pool performance monitoring process. Then, among the random read, sequential read, random write, and sequential write for the target RAID group RG, the access type discrimination that discriminates the access type exceeding the preset first IOPS threshold for each of these access types The processing is executed according to the processing procedure shown in FIG. At this time, the access type determination program 31 uses the pool performance monitoring process (the performance information of the target RAID group RG acquired from the start of the previous pool performance monitoring process to the start of the current pool performance monitoring process) ( The performance information of the target RAID group RG acquired in step SP3 of FIG. 8) is used.

In practice, the access type discrimination program 31 starts this access type discrimination process when called by the performance monitoring program 30 in step SP13 of the pool performance monitoring process, and is first acquired in step SP3 of the pool performance monitoring process (FIG. 8). Based on the performance information of the target RAID group RG, it is determined whether there is only one access type that exceeds the first IOPS threshold (SP20).

Specifically, the access type determination program 31 calculates the average value of the access frequency for each access type based on the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring process (FIG. 8). The calculated average access frequency for each access type is compared with the first IOPS threshold value for that access type. Then, the access type determination program 31 determines whether the access type whose average access frequency exceeds the corresponding first IOPS threshold is only one of random read, sequential read, random write, and sequential write. Judge whether or not.

If the access type determination program 31 obtains a positive result in this determination, it determines whether the access type exceeding the first IOPS threshold is a write-related access type (random read or sequential read) (SP21). ). When the access type determination program 31 obtains a positive result in this determination, it calls the performance improvement processing program 32 (FIG. 2) to execute the first read performance improvement processing for improving the read performance (SP22). Thereafter, the access type determination process is terminated and the process returns to the pool performance monitoring process (FIG. 8).

If the access type determination program 31 obtains a negative result in the determination at step SP21, it calls the performance improvement processing program 32 to execute the first write performance improvement processing for improving the write performance (SP23). Then, the access type determination process is terminated and the process returns to the pool performance monitoring process.

On the other hand, when the access type determination program 31 obtains a negative result in the determination at step SP20, the access type exceeding the first IOPS threshold is only the write-related access type (random write and sequential write). Is determined (SP24). If the access type determination program 31 obtains a positive result in this determination, it calls the performance improvement processing program 32 to execute the second write performance improvement process for improving the write performance (SP25). This access type determination process is terminated and the process returns to the pool performance monitoring process.

If the access type determination program 31 obtains a negative result in the determination at step SP24, it determines whether or not the access type exceeding the first IOPS threshold is only a read-related access type (random read and sequential read). Judgment is made (SP26). When the access type determination program 31 obtains a positive result in this determination, it calls the performance improvement processing program 32 to execute the second read performance improvement process for improving the read performance (SP27), and thereafter This access type determination process is terminated and the process returns to the pool performance monitoring process.

On the other hand, to obtain a negative result in the determination at step SP26 means that the access types that exceed the first IOPS threshold include both write-related and read-related access types. Thus, at this time, the access type determination program calls the performance improvement processing program 32 to execute the write / read performance improvement process for improving both the write performance and the read performance (SP28), and thereafter this access type. The discrimination process ends and the process returns to the pool performance monitoring process.

(4-2-2) First Read Performance Improvement Process FIG. 13 shows the first read performance improvement process executed by the performance improvement process program 32 called by the access type determination program 31 in step SP22 of the access type determination process described above with reference to FIG. The specific processing content of the read performance improvement processing 1 is shown.

When called in step SP22 of the access type determination process, the performance improvement processing program 32 starts the first read performance improvement process shown in FIG. 13, and is first acquired in step SP3 of the pool performance monitoring process (FIG. 8). Based on the performance information of the target RAID group RG, the average value of the cache hit rate at the time of random read access and the average value of the cache hit rate at the time of sequential read access are calculated, respectively, and at least one of these average values is It is determined whether or not the cache hit rate threshold is below (SP30).

When the performance improvement processing program 32 obtains a positive result in this determination, it executes a cache area addition process for assigning an additional cache area (CLPR or CLPR divided area) to the target pool PL (SP31), and thereafter The first read performance improvement process ends and the process returns to the access type determination process.

On the other hand, when the performance improvement processing program 32 obtains a negative result in the determination at step SP30, it executes a pool volume addition process for adding a new pool volume PLVOL to the target pool PL (SP32). The first read performance improvement process ends and the process returns to the access type determination process.

(4-2-3) Cache Area Addition Processing FIG. 14 shows a specific processing procedure of the cache area addition processing executed by the performance improvement processing program 32 in step SP31 of the first read performance improvement processing described above with reference to FIG. Indicates.

When the performance improvement processing program 32 proceeds to step SP31 of the first read performance improvement processing, the performance improvement processing program 32 starts this cache area addition processing, and first determines whether or not a cache area has already been added to the target pool PL. (SP40). When the performance improvement processing program 32 obtains a positive result in this determination, it ends this cache area addition processing and returns to the read performance improvement processing. This is because when the process after step SP6 of the pool performance monitoring process is executed for each RAID group RG for which the threshold excess flag is set to ON in step SP5 of the pool performance monitoring process (FIG. 8), the target pool PL is set. This is to prevent the cache area from being added cumulatively.

On the other hand, when the performance improvement processing program 32 obtains a negative result in the determination at step SP40, the CLPR that can be additionally allocated to the target pool PL out of any CLPR or the like (CLPR or CLPR divided region). Etc. are searched on the CLPR management table 38 (FIG. 6) (SP41). Specifically, the performance improvement processing program 32 is a CLPR other than the CLPR originally assigned to the target pool PL, and the cache write wait rate stored in the cache write wait rate column 38D of the CLPR management table 38 is A CLPR or the like below a preset threshold value (hereinafter referred to as a cache write wait rate threshold value) is searched.

Subsequently, the performance improvement processing program 32 determines whether or not the CLPR that can be allocated to the target pool PL has been detected by the search in step SP41 (SP42). When the performance improvement processing program 32 obtains an affirmative result in this determination, the CLPR detected at step SP41 (or one CLPR of the plurality of CLPR detected at step SP41) is stored in the target pool PL. Further allocation (SP43), and then the process proceeds to step SP47.

On the other hand, when the performance improvement processing program 32 obtains a negative result in the determination at step SP45, it searches the CLPR management table 38 (FIG. 6) for a spare CLPR that can be additionally assigned to the target pool PL (SP44). ). Specifically, the performance improvement processing program 32 stores a character string “reserved” in the CLPR number column 38A (FIG. 6) of the CLPR management table 38, and the cache write wait rate is less than the above-described cache write wait rate threshold. Search for.

Subsequently, the performance improvement processing program 32 determines whether or not a spare CLPR that can be additionally allocated to the target pool PL has been detected by the search in step SP44 (SP45). When the performance improvement processing program 32 obtains a negative result in this determination, it ends this cache area addition processing and returns to the read performance improvement processing.

On the other hand, if the performance improvement processing program 32 obtains a positive result in the determination at step SP45, the spare CLPR detected at step SP44 (if a plurality of spare CLPRs are detected at step SP44, one of the spare CLPRs is detected). ) Is divided as necessary, and the spare CLPR or one of the divided areas is allocated to the target pool PL (SP46).

Next, the performance improvement processing program 32 is stored in the CLPR management table 38 in the temporary assignment flag column 38E (FIG. 6) corresponding to the CLPR or the like assigned to the target pool PL in step SP43 or step SP46, or the spare CLPR or its divided area. And the pool number of the target pool PL is stored in the corresponding assigned pool field 38F (FIG. 6) (SP47). Thereafter, this cache area addition process To return to the lead performance improvement process.

(4-2-4) Pool Volume Addition Processing FIG. 15 shows a specific processing procedure of pool volume addition processing executed by the performance improvement processing program 32 in step SP32 of the first read performance improvement processing described above with reference to FIG. Indicates.

When the performance improvement processing program 32 proceeds to step SP32 of the first read performance improvement processing, the performance improvement processing program 32 starts this pool volume addition processing and first determines whether or not the pool volume PLVOL has already been added to the target pool PL. (SP50). When the performance improvement processing program 32 obtains a positive result in this determination, it ends this pool volume addition processing. This is because when the process after step SP6 of the pool performance monitoring process is executed for each RAID group RG for which the threshold excess flag is set to ON in step SP5 of the pool performance monitoring process (FIG. 8), the target pool PL is set. This is to prevent the pool volume PLVOL from being added cumulatively.

On the other hand, if the performance improvement processing program 32 obtains a negative result in the determination at step SP50, the average value of the access frequencies for the entire target pool PL of the access type (random read and sequential read here) targeted at that time Is the value obtained by subtracting the second IOPS threshold from the actual average value in the target pool PL (“actual average value” − “first” 2 ") (SP51).

The second IOPS threshold is a numerical value that is preset for each access type as a target value of the access frequency when the access frequency of each access type is reduced by the performance improvement process executed by the performance improvement processing program 32. . As shown in FIGS. 11A and 11B, each RAID group RG is defined based on the specifications of the storage devices 20 constituting the RAID group RG (for example, the performance limit value of the RAID group RG). 60%), and the second IOPS threshold value of the pool PL is the sum of the second IOPS threshold values of the RAID groups RG that respectively provide the pool volumes PL in the pool PL as shown in FIG. Calculated as a value.

Therefore, in the case of the example of FIG. 11C, the performance improvement processing program 32 determines the random read from “3790” that is the actual average value of the random read access frequency for the target pool PL in this step SP51. The number of accesses is calculated by subtracting the second IOPS threshold “2532” for the sequential read, and for sequential reads, “1570”, which is the actual average value of the sequential read access frequency for the target pool PL, is used for sequential reads. The number of accesses is calculated by subtracting “2348” which is the second IOPS threshold.

Subsequently, the performance improvement processing program 32 searches the pool information management table 37 for an unused or spare RAID group RG (SP52). For example, in the example of FIG. 5, the RAID group RG assigned the RAID group number “5-1” is not assigned to any pool PL (the value in the pool number column 37A is “−”). Since it is a group RG, it is detected as an unused or spare RAID group RG in this step SP52.

Next, the performance improvement processing program 32 refers to the RAID group information management table 36 (FIG. 4) corresponding to the RAID group RG detected in step SP52, and finds a logical volume that can be allocated from the RAID group RG to the target pool PL. It is determined whether or not it exists (SP53).

Here, the “logical volume that can be allocated from the RAID group RG to the target pool PL” means a pool volume PLVOL provided by a RAID group RG other than the RAID group RG that provides the pool volume PLVOL of the target pool PL at that time. The pool volume PLVOL may already be created on a storage area provided by the RAID group RG, or may be newly created on the storage area. However, in the case of a pool volume PLVOL that has already been created, the actual average values of all the access types (see FIGS. 11A and 11B) of the RAID group RG that provides the pool volume PLVOL are all the access types. The condition is that it is less than the second IOPS threshold.

If the performance improvement processing program 32 obtains a negative result in this determination, it proceeds to step SP56. In contrast, if the performance improvement processing program 32 obtains a positive result in the determination at step SP53, it allocates the logical volume detected at step SP53 to the target pool PL as an additional pool volume PLVOL (SP54).

Further, the performance improvement processing program 32 indicates that the actual average value of the target access type (in this case, random read and sequential read) in the target pool PL after adding the pool volume PLVOL to the target pool PL in step SP54 is the relevant pool. It is determined whether or not the target pool PL is equal to or lower than the second IOPS threshold after the volume PLVOL is added to the target pool PL (SP55).

In fact, the pool volume PLVOL related to the target pool PL is added to the actual average value of the access type targeted by the target pool PL, and the actual average value of the target access type in the pool volume PLVOL added at that time is added. Is the actual average value of the target access type in the target pool PL after this (hereinafter referred to as the actual average value after addition), and the second IOPS threshold value of the target access type of the target pool PL is A value obtained by adding the second IOPS threshold value of the target access type in the added pool volume PLVOL at that time is the second of the target access type in the target pool PL after adding the pool volume PLVOL related to the target pool PL. IOPS threshold (hereinafter referred to as the second IOPS threshold after the addition) Since the called), improved performance processing program 32, in step SP55, extra after actual average value, will determine whether it is less than extra after the second IOPS threshold.

If the performance improvement processing program 32 obtains a negative result in this determination, it returns to step SP53, and thereafter obtains a negative result at step SP53 or obtains a positive result at step SP55 until the processing of step SP53 to step SP55 is performed. repeat. By this repeated processing, pool volumes PLVOL are sequentially added to the target pool PL.

Then, the performance improvement processing program 32 eventually returns a positive result in step SP55 when the performance limit value of the target access type (in this case, random read and sequential read) in the target RAID group RG becomes equal to or less than the second IOPS threshold. Is obtained, the pool volume addition process is terminated and the process returns to the first read performance improvement process (FIG. 13).

On the other hand, the performance improvement processing program 32 can be assigned as a pool volume PLVOL to the target pool PL before the performance limit value of the target access type in the target RAID group RG falls below the second IOPS threshold. If a negative result is obtained in step SP53 due to the absence of a logical volume, the pool volume PLVOL is not provided to the target pool PL with reference to the pool information management table 37 (FIG. 5), and other than the target pool PL It is determined whether there is a RAID group RG that provides a pool volume PLVOL to the pool PL and the storage device usage rate is equal to or less than the storage device usage rate threshold value (SP56).

If the performance improvement processing program 32 obtains a negative result in this determination, it ends this pool volume addition processing and returns to the first read performance improvement processing (FIG. 13). Therefore, in this case, read performance is not improved by adding a new pool volume PLVOL to the target pool PL.

On the other hand, when the performance improvement processing program 32 obtains a positive result in the determination at step SP56, the unallocated capacity column 37D () corresponding to the RAID group RG detected at step SP56 in the pool information management table 37 (FIG. 5). Referring to FIG. 13), whether or not a new logical volume can be created in the storage area provided by the RAID group RG is determined from the viewpoint of the unallocated storage capacity of the RAID group RG (SP57). ).

When the performance improvement processing program 32 obtains a negative result in this determination, it deletes the pool volume PLVOL that the RAID group RG provides to the pool PL other than the target pool PL from the pool PL, and re-enters the target pool PL. The allocated pool volume reassignment process described later with reference to FIG. 16 is executed (SP58), and thereafter, the process proceeds to step SP61.

In contrast, when the performance improvement processing program 32 obtains a positive result in the determination at step SP57, the logical volume provided to the target pool PL as the pool volume PLVOL on the storage area provided by the RAID group RG detected at step SP56. Is created, it is determined whether or not the access frequency of each access type of the RAID group RG is equal to or less than the corresponding second IOPS threshold (SP59).

In contrast, when the performance improvement processing program 32 obtains a positive result in the determination at step SP59, it creates a new logical volume in the storage area provided by the RAID group RG detected at step SP56, and creates the created logical volume. Is added to the target pool PL as a new pool volume PLVOL (SP60).

Next, in the same manner as in step SP55, the performance improvement processing program 32 determines that the actual average value of the entire target pool PL after adding a new pool volume PLVOL in step SP60 is less than or equal to the second IOPS threshold of the entire target pool PL. It is determined whether or not (SP61). When the performance improvement processing program 32 obtains a negative result in this determination, it returns to step SP56, and thereafter obtains a negative result at any of step SP56, step SP57 or step SP59, or obtains a positive result at step SP61. The processing from step SP56 to step SP61 is repeated. Through this repeated process, new pool volumes PLVOL are sequentially added to the target pool PL.

The performance improvement processing program 32 eventually ends this pool volume addition processing when the measured average value of the entire target pool PL eventually becomes equal to or less than the second IOPS threshold value of the entire target pool PL and a positive result is obtained in step SP61. Then, the process returns to the first read performance improvement process (FIG. 13).

In addition, when a new pool volume PLVOL is added to the target pool PL as described above, after the end of this pool volume addition process, the data stored in each pool volume PLVOL that has previously constituted the target pool PL A part may be moved to the newly added pool volume PLVOL. By doing so, the load of each RAID group RG that has previously provided the pool volume PLVOL to the target pool PL can be distributed to the RAID group RG that newly provides the pool volume PLVOL to the target pool PL. As a result, the response performance of the target pool PL can be immediately improved.

On the other hand, FIG. 16 shows the specific processing contents of the allocated pool volume reassignment processing executed by the performance improvement processing program 32 in step SP58 of the above-described pool volume addition processing (FIG. 15).

When the performance improvement processing program 32 proceeds to step SP58 of the pool volume addition processing (FIG. 15), it starts the allocated pool volume reassignment processing shown in FIG. 16, and first, the pool information management table 37 (FIG. 5) and Referring to the corresponding RAID group performance information table 35 (FIG. 3), all RAID groups RG (here, target pools) that are pools PL other than the target pool PL and that provide the pool volume PLVOL to the pool PL. The RAID group RG that does not provide the pool volume PVOL to the PL (however, it may be a RAID group RG other than this) searches the pool PL whose storage device usage rate is equal to or less than the storage area usage rate threshold, and the pool PL Pool information is acquired from the pool information management table 37 (FIG. 5). (SP70).

Subsequently, the performance improvement processing program 32 determines whether there is a pool PL that can delete the pool volume PLVOL from the viewpoint of capacity in any of the pools PL detected in step SP70 (SP71). ). This determination is made by determining whether or not the corresponding pool PL has been detected by the search in step SP71. If such a pool PL can be detected, an affirmative result is obtained in step SP71. When such a pool PL cannot be detected, a negative result is obtained in this step SP71.

When the performance improvement processing program 32 obtains a negative result in the determination at step SP71, it ends this allocated pool volume reassignment processing and returns to the pool volume addition processing (FIG. 15). Accordingly, in this case, the pool volume PLVOL assigned to the pool PL other than the target pool PL cannot be reassigned to the target pool PL.

On the other hand, when the performance improvement processing program 32 obtains a positive result in the determination at step SP71, it refers to the required RAID group performance information table 35 (FIG. 3) and from the viewpoint of capacity detected at step SP70. It is determined whether there is a pool PL in which the pool volume PLVOL can be deleted from the viewpoint of performance among the pools in which the pool volume PLVOL can be deleted as viewed (SP72).

This determination is made for each pool PL other than the target pool PL, even when the pool volume PLVOL is deleted from the pool PL, the actual average value of the pool PL after the pool volume PLVOL is deleted. This is done by determining whether there is a pool volume PLVOL that is less than the overall second IOPS threshold. If such a pool PL can be detected, a positive result is obtained in this step SP72, and if such a pool PL cannot be detected, a negative result is obtained in this step SP72.

When the performance improvement processing program 32 obtains a negative result in the determination at step SP72, it ends this allocated pool volume reassignment processing and returns to the pool volume addition processing (FIG. 15). Therefore, also in this case, the pool volume PLVOL constituting the pool PL other than the target pool PL cannot be reassigned to the target pool PL.

On the other hand, when the performance improvement processing program 32 obtains a positive result in the determination at step SP72, the pool volume that can be reassigned to the target pool PL of the pool volumes PLVOL assigned to the pool PL detected at step SP72. (Hereinafter, this is referred to as a reallocation target pool volume) After data stored in a PLVOL is migrated to another pool volume PLVOL that constitutes the pool PL, the reallocation target pool volume PLVOL is transferred from the pool PL. Delete (release the allocation of the reallocation target pool volume PLVOL to the pool PL) (SP73).

Subsequently, the performance improvement processing program 32 adds the above-described reallocation target pool volume PLVOL to the target pool PL (SP74), and thereafter performs the same step as step SP55 of the pool volume addition processing described above with reference to FIG. The measured average value of the target access type (random read and sequential read here) in the entire target pool PL after adding the reallocation target pool volume PLVOL in SP74 is equal to or less than the second IOPS threshold of the target pool PL. It is determined whether or not (SP75). When the performance improvement processing program 32 obtains a negative result in this determination, it returns to step SP71, and thereafter repeats the processing of steps SP71 to SP75 until it obtains a negative result at step SP72 or obtains a positive result at step SP75. .

When the performance improvement processing program 32 eventually obtains an affirmative result in step SP75 because the measured average value of the target access type in the entire target pool PL becomes equal to or less than the second IOPS threshold of the target pool PL. Then, the allocated pool volume deletion process is terminated and the process returns to the pool volume addition process (FIG. 15).

(4-2-5) Second Write Performance Improvement Processing On the other hand, FIG. 17 is executed by the performance improvement processing program 32 called by the access type determination program 31 in step SP25 of the access type determination processing described above with reference to FIG. The specific processing content of the second write performance improvement processing will be described.

When called in step SP25 of the access type determination process, the performance improvement process program 32 starts the second write performance improvement process shown in FIG. 17, and first, in the target RAID group RG, the random write access load Is higher than the sequential write access load (SP80). This determination is based on whether or not the value obtained by subtracting the second IOPS threshold from the average value of the random write access frequency is greater than the value obtained by subtracting the second IOPS threshold from the average value of the sequential write access frequency. It is done by judging.

For example, as shown in FIG. 10, the average value of random write access frequency is “90”, the average value of sequential write access frequency is “200”, the second IOPS threshold for random write is “227”, and When the second IOPS threshold is “836”, the value obtained by subtracting the second IOPS threshold for random writes from the average value of random write access frequency is “−137” (“(actually measured average value) in FIG. 10). -(Second IOPS threshold) "), and the value obtained by subtracting the second IOPS threshold for sequential write from the average value of sequential write access frequency is" -636 "times (" (( In this example, the load due to the random write is the sequential write because the actual average value) − (the second IOPS threshold) line). It is determined that the load is greater than

If the performance improvement processing program 32 obtains an affirmative result in this determination, the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring processing (FIG. 8) indicates the random write access frequency of the target RAID group RG. (SP81). Further, the performance improvement processing program 32 executes the pool volume addition process described above with reference to FIG. 15 to improve the write performance by adding the pool volume PLVOL to the target pool PL (SP83). 2 is ended, and the process returns to the access type determination process (FIG. 12).

On the other hand, when the performance improvement processing program 32 obtains a negative result in the determination at step SP80, the sequential write access frequency of the target RAID group RG is acquired in step SP3 of the pool performance monitoring process (FIG. 8). Obtained from the performance information of the RAID group RG (SP82). Further, the performance improvement processing program 32 executes the pool volume addition process described above with reference to FIG. 15 to improve the write performance by adding the pool volume PLVOL to the target pool PL (SP83). 2 finishes the write performance improvement process and returns to the access type determination process.

(4-2-6) Second Read Performance Improvement Process FIG. 18 shows a second read performance improvement process executed by the performance improvement process program 32 called in step SP27 of the access type determination process described above with reference to FIG. The specific processing content of is shown.

When the performance improvement processing program 32 is called by the access type determination program 31 (FIG. 2) in step SP27 of the access type determination processing, the performance improvement processing program 32 starts the second read performance improvement processing shown in FIG. Based on the performance information of the target RAID group RG acquired in step SP3 of the process (FIG. 8), the average value of the cache hit rate at the time of random read access and sequential read access is calculated (SP80).

Subsequently, the performance improvement processing program 32 has both the average value of the cache hit rate at the time of random read access calculated at step SP80 and the average value of the cache hit rate at the time of sequential read access are equal to or less than the above-described cache hit rate threshold value. It is determined whether or not there is (SP81).

When the performance improvement processing program 32 obtains a positive result in this determination, it adds a cache area to be allocated to the target pool PL, thereby improving the cache hit rate at the time of random read access and sequential read access. After executing the cache area addition process described above (SP82), the second read performance improvement process is terminated and the process returns to the access type determination process (FIG. 12).

On the other hand, the performance improvement processing program 32 has both the average value of the cache hit rate at the time of random read access calculated at step SP80 and the average value of the cache hit rate at the time of sequential read access are not less than the above-described cache hit rate threshold value. It is determined whether or not (SP83).

Obtaining a negative result in this determination means that only one of the average value of the cache hit rate during random read access and the average value of the cache hit rate during sequential read access is equal to or greater than the cache hit rate threshold. means. Thus, at this time, the performance improvement processing program 32 specifies whether the access type in which the average value of the cache hit rate calculated in step SP80 is equal to or less than the cache hit rate threshold is random read or sequential read (SP84).

Subsequently, the performance improvement processing program 32 executes the cache area addition process described above with reference to FIG. 14 to improve the cache hit rate of random read and sequential read by adding a cache area to be allocated to the target pool PL (SP85). .

Then, the performance improvement processing program 32 thereafter accesses the access type (random read or sequential read) identified in step SP84 based on the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring process (FIG. 8). The average access frequency is calculated (SP86).

Next, the performance improvement processing program 32 executes the pool volume addition process described above with reference to FIG. 15 for adding the pool volume PLVOL to the target pool PL using the average value calculated in step SP86 (SP90), and thereafter The second read performance improvement process is terminated.

On the other hand, obtaining a positive result in the determination at step SP83 means that both the average value of the cache hit rate at the time of random read access and the average value of the cache hit rate at the time of sequential read access are equal to or greater than the cache hit rate threshold value. Means. Thus, at this time, the performance improvement processing program 32, in the same way as the step SP80 of the second write performance improvement processing described above with reference to FIG. 17, causes the load due to random write to be greater than the load due to sequential write in the target RAID group RG. It is determined whether or not the value is higher (SP87).

If the performance improvement processing program 32 obtains a positive result in this determination, the average value of the random read access frequency based on the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring processing (FIG. 8). Is calculated (SP88). If the performance improvement processing program 32 obtains a negative result in the determination at step SP87, the performance improvement processing program 32 determines the access frequency of sequential reads based on the performance information of the target RAID group RG acquired at step SP3 of the pool performance monitoring processing (FIG. 8). An average value is calculated (SP89).

The performance improvement processing program 32 executes the pool volume addition process described above with reference to FIG. 15 for adding the pool volume PLVOL to the target pool PL using the average value calculated in step SP88 or step SP89 (SP90). Thereafter, the second read performance improvement process is terminated.

(4-2-7) Read / Write Performance Improvement Processing FIGS. 19A and 19B show the read / write performance improvement executed by the performance improvement processing program 32 called in step SP28 of the access type determination processing described above with reference to FIG. Specific processing contents of the processing are shown.

When the performance improvement processing program 32 is called by the access type determination program 31 (FIG. 2) in step SP28 of the access type determination processing, the performance improvement processing program 32 starts the read / write performance improvement processing shown in FIGS. 19A and 19B. Based on the performance information of the target RAID group RG acquired in step SP3 of the performance monitoring process (FIG. 8), an average value of random read access frequency and an average value of sequential read access frequency per unit time are respectively calculated. Then, based on the calculation result, it is determined whether only one of the random read and sequential read has the average value exceeding the corresponding first IOPS threshold (SP100).

If the performance improvement processing program 32 obtains a positive result in this determination, the performance improvement processing program 32 corresponds to the average value corresponding to the average value of the random read and sequential read access frequencies of the target RAID group RG calculated in step SP100. It is specified whether random read or sequential read exceeds the IOPS threshold of 1, and the average value of the cache hit rate of the specified random read or sequential read is calculated (SP101).

Subsequently, the performance improvement processing program 32 determines whether or not the cache hit rate calculated in step SP101 is less than or equal to the cache hit rate threshold (SP102). When the performance improvement processing program 32 obtains an affirmative result in this determination, the CLPR management table 38 (see FIG. 14) assigns CLPR and the like that can be assigned to the target pool PL in the same manner as in step SP41 of the cache area addition processing described above with reference to FIG. 6) Search above (SP103).

Next, the performance improvement processing program 32 determines whether or not a CLPR that can be additionally assigned to the target pool PL has been detected by the search in step SP103 (SP104). When the performance improvement processing program 32 obtains an affirmative result in this determination, the CLPR detected at step SP103 or the like (if a plurality of CLPR or the like is detected at step SP103) is stored in the target pool PL. Allocation (SP105), and then the process proceeds to step SP109.

On the other hand, when the performance improvement processing program 32 obtains a negative result in the determination at step SP104, it can be additionally allocated to the target pool PL in the same manner as at step SP44 in the cache area addition processing (FIG. 14). The CLPR is searched on the CLPR management table 38 (FIG. 6) (SP106).

Subsequently, the performance improvement processing program 32 determines whether or not a spare CLPR that can be additionally allocated to the target pool PL has been detected by the search in step SP106 (SP107). If the performance improvement processing program 32 obtains a negative result in this determination, it proceeds to step SP110.

On the other hand, when the performance improvement processing program 32 obtains a positive result in the determination at step SP107, the spare CLPR detected at step SP106 (one spare CLPR among them when a plurality of spare CLPRs are detected at step SP106). Is assigned to the target pool PL (SP108).

Next, the performance improvement processing program 32 is stored in the temporary assignment flag column 38E (FIG. 6) corresponding to the CLPR or the like assigned to the target pool PL in step SP105 or step SP108 in the CLPR management table 38 (FIG. 6) or the backup CLPR. Is set to ON (“1”) (SP109).

Furthermore, the performance improvement processing program 32 executes the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring process (FIG. 8), the second IOPS threshold for random write, and the second IOPS threshold for sequential write. Based on the above, the higher one of the random write and sequential write is specified (SP110). Specifically, this processing is performed by subtracting the second IOPS threshold from the average value of the random write access frequency and the value obtained by subtracting the second IOPS threshold from the average value of the sequential write access frequency. This is done by specifying the larger value.

Further, the performance improvement processing program 32 acquires only the performance information related to the random write or sequential write specified in step SP110 from the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring process (FIG. 8). Then, the pool volume addition processing described above with reference to FIG. 15 is executed using the acquired performance information (SP124). Then, the performance improvement processing program 32 thereafter ends this read / write performance improvement processing and returns to the access type determination processing (FIG. 12).

On the other hand, if the performance improvement processing program 32 obtains a negative result in the determination at step SP102, each access type (random) is based on the performance information of the target RAID group RG acquired at step SP3 of the pool performance monitoring processing (FIG. 8). All access types in which the average access frequency exceeds the first IOPS threshold among the read, sequential read, random write, and sequential write) are detected (SP111).

Subsequently, the performance improvement processing program 32 specifies the access type having the highest load among the access types detected in step SP111 (SP112). This process is performed by specifying the access type having the largest value obtained by subtracting the second IOPS threshold from the average access frequency of each access type detected in step SP111.

Next, the performance improvement processing program 32 acquires performance information related to the access type specified in step SP112 from the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring process (FIG. 8) (SP113). Then, the pool volume addition processing described above with reference to FIG. 15 is executed using the acquired performance information (SP124). Then, the performance improvement processing program 32 thereafter ends this read / write performance improvement processing and returns to the access type determination processing (FIG. 12).

On the other hand, when the performance improvement processing program 32 obtains a negative result in the determination at step SP100, based on the performance information of the target RAID group RG acquired at step SP3 of the pool performance monitoring processing (FIG. 8), The average value of the cache hit rate at the time of random read access is calculated (SP114).

Subsequently, the performance improvement processing program 32 determines that both the average value of the cache hit rate at the sequential read access and the average value of the cache hit rate at the random read access calculated in step SP114 are equal to or greater than the cache hit rate threshold value. It is determined whether or not there is (SP115).

を得 Acquiring a positive result in this determination means that the capacity of the cache area allocated to the target pool PL is not insufficient. Thus, at this time, the performance improvement processing program 32 proceeds to step SP111, and thereafter executes step SP111 to step SP113 to step SP124 as described above. Then, the performance improvement processing program 32 thereafter ends this read / write performance improvement processing and returns to the access type determination processing (FIG. 12).

On the other hand, obtaining a negative result in the determination at step SP115 means that the capacity of the cache area allocated to the target pool PL is insufficient (SP116). Thus, at this time, the performance improvement processing program 32 determines whether or not the average value of the cache hit rate at the time of random read access and sequential read access calculated at step SP114 is less than or equal to the cache hit rate threshold value (SP119). ).

Then, when the performance improvement processing program 32 obtains a positive result in this determination, it additionally allocates a cache area to the target pool PL by executing the cache area addition process described above with reference to FIG. 14 (SP120).

Further, the performance improvement processing program 32 specifies the higher one of random read and sequential read based on the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring process (FIG. 12), The performance information is acquired (SP121). Specifically, the performance improvement processing program 32 subtracts the second IOPS threshold from the average value of the random read access frequency, and subtracts the second IOPS threshold from the average value of the sequential read access frequency. Of the target RAID group RG acquired at step SP3 of the pool performance monitoring process, only the performance information related to the specified random read or sequential read is acquired.

Then, the performance improvement processing program 32 executes the pool volume addition processing described above with reference to FIG. 15 using the performance information acquired in step SP121 (SP124), and then ends this read / write performance improvement processing to access type determination processing. Return to.

On the other hand, when the performance improvement processing program 32 obtains a negative result in the determination at step SP119, it additionally allocates a cache area to the target pool PL by executing the cache area addition process described above with reference to FIG. (SP122).

Subsequently, in the same manner as in step SP121, the performance improvement processing program 32 specifies the higher one of random read and random write, and acquires performance information regarding the random read or random write in the target RAID group RG. (SP123).

The performance improvement processing program 32 executes the pool volume addition processing described above with reference to FIG. 15 using the performance information acquired in step SP123 (SP124), and then ends this read / write performance improvement processing to access type determination processing. Return to.

(4-3) Batch Improvement Process FIG. 20 shows the details of the batch improvement process executed by the batch improvement program 33 (FIG. 2) called by the performance monitoring program 30 in step SP11 of the pool performance monitoring process described above with reference to FIG. The typical processing content is shown. In the following description, it is assumed that the read / write access during batch processing is sequential access as described above.

When the batch improvement program 33 is called by the performance monitoring program 30 (FIG. 2) in step SP11 of the pool performance monitoring process, it starts the batch improvement process shown in FIG. 20, and first, the batch schedule management table 39 (FIG. 7). 1 is selected from the batch processes registered in (1) (SP130).

Subsequently, the batch improvement program 33 refers to the batch schedule management table 39 and uses the target RAID group RG when executing the batch process selected in step SP130 (hereinafter referred to as the target batch process). It is determined whether or not (SP131).

If the batch improvement program 33 obtains a negative result in this determination, it returns to step SP130. On the other hand, if the batch improvement program 33 obtains a positive result in the determination at step SP131, the sequential read access frequency in the target RAID group RG is equal to or higher than the first IOPS threshold during the target batch processing execution time zone. Whether or not (SP132).

Specifically, in this step SP132, the batch improvement program 33 first acquires the execution time zone (time zone from the start time to the end time) of the target batch process from the batch schedule management table 39, and performs the pool performance monitoring process. Based on the performance information of the target RAID group RG acquired in step SP3, an average value of sequential read access frequencies in the execution time zone of the target batch process is calculated, and the calculated average value is a first IOPS threshold value for sequential read. It is determined whether this is the case.

When the batch improvement program 33 obtains a negative result in the determination at step SP132, the batch improvement program 33 is set to execute the performance improvement process described above with reference to FIGS. 12 to 19B as the batch pre-process before the next execution of the batch process (SP133 Thereafter, the process proceeds to step SP137. Thus, according to the setting in step SP133, the performance improvement process is executed before the start of the target batch process, whereby an additional cache area or pool volume is allocated to the target pool PL.

On the other hand, if the batch improvement program 33 obtains a negative result in the determination at step SP132, the cache hit rate at the time of sequential read access and sequential write access in the execution time zone of the target batch process is less than the cache hit rate threshold value. Whether or not (SP134).

Specifically, in step SP134, the batch improvement program 33 uses the performance information of the target RAID group RG acquired in step SP3 of the pool performance monitoring process to cache the sequential read access in the execution time zone of the target batch process. Calculate the average hit rate. Then, the batch improvement program 33 determines whether or not the average value of the cache hit rate at the time of sequential read access is equal to or smaller than the cache hit rate threshold value during the target batch processing.

If the batch improvement program 33 obtains a negative result in the determination at step SP134, the performance improvement processing program 32 executes the pool volume addition processing described above with reference to FIG. 15 as batch preprocessing before the next execution of the target batch processing. (SP135), and then the process proceeds to step SP137. Thus, according to the setting of step SP137, the pool volume addition processing is executed by the performance improvement processing program 32 before the start of the target batch processing, whereby the pool volume PLVOL is additionally allocated to the target pool PL.

On the other hand, when the batch improvement program 33 obtains a positive result in the determination at step SP134, the performance improvement processing program 32 performs the cache area addition processing described above with reference to FIG. 14 before executing the target batch processing next time before the batch. It is set to be executed as a process (SP136). Thus, according to the setting of step SP136, the cache area addition process is executed by the performance improvement processing program 32 before the start of the target batch process, and as a result, the cache capacity (cache area divided area or spare cache area) is stored in the target pool PL. Will be added.

Thereafter, the batch improvement program 33 determines whether or not the processing of step SP131 to step SP136 has been executed for all batch processing registered in the batch schedule management table 39 (FIG. 7) (SP137).

If the batch improvement program 33 obtains a negative result in this determination, it returns to step SP130, and thereafter, the batch processing selected in step SP130 is sequentially switched to other unprocessed batch processing, and the processing of steps SP130 to SP137 is performed. repeat.

When the batch improvement program 33 finally obtains a positive result in step SP137 by completing the execution of the processes in steps SP131 to SP136 for all the batch processes registered in the batch schedule management table 39, the batch improvement process ends. Then, the process returns to the pool performance improvement process (FIG. 8).

Incidentally, FIG. 21 shows a processing procedure of batch post-processing that is executed every time one batch processing is completed in relation to such batch improvement processing. This batch post-processing releases the CLPR and the like additionally assigned to the pool PL in the batch processing, and this time, based on whether or not the batch processing is completed within the scheduled time, This is processing for setting the capacity of an additional cache area when adding CLPR or the like.

In practice, when one batch process is completed, the batch improvement program 33 starts the post-batch process. When the batch process is executed, CLPR or the like is added to the pool PL used in the batch process. It is determined whether or not it has been assigned to (SP140). If the batch improvement program 33 obtains a negative result in this determination, it proceeds to step SP142.

On the other hand, when the batch improvement program 33 obtains a positive result in the determination at step SP140, it releases the CLPR added to the pool PL used in the batch processing (SP141), and thereafter the batch schedule management table 39. With reference to the in-time processing completion flag 39F (FIG. 7) stored in the in-time processing completion flag column 39F (FIG. 7) corresponding to the batch processing in (FIG. 7), has this batch processing been completed within the scheduled time? It is determined whether or not (SP142).

When the batch improvement program 33 obtains a positive result in this determination, the cache area to be added when the cache area is additionally allocated to the corresponding pool PL when the batch process is executed next time is set to the same capacity ( If the cache area is not added in the current process, the specified capacity is set (SP143), and then the post-batch process is terminated.

On the other hand, if the batch improvement program 33 obtains a negative result in the determination at step SP142, the capacity to be added when the cache area is additionally allocated to the corresponding pool PL when the batch process is executed next time is determined this time. Is set to a capacity larger than the predetermined amount (a prescribed capacity when no cache area is added in the current process) (SP144), and then the post-batch process is terminated.

(4-4) Cache Return Processing FIG. 22 shows specific processing contents of the cache return processing executed by the performance monitoring program 30 in step SP15 of the pool performance monitoring processing described above with reference to FIG.

When the performance monitoring program 30 proceeds to step SP15 of the pool performance monitoring process, the performance monitoring program 30 starts the cache return process shown in FIG. 22, and first adds the target pool PL with reference to the CLPR management table 38 (FIG. 6). It is determined whether CLPR or the like is assigned (SP150). When the performance monitoring program 30 obtains a negative result in this determination, it ends this cache return processing and returns to the pool performance monitoring processing.

On the other hand, when the performance monitoring program 30 obtains a positive result in the determination at step SP150, it refers to the CLPR management table 38 (FIG. 6), and sets a threshold (( Hereinafter, this is referred to as a cache write wait rate threshold) or not (SP151).

When the performance monitoring program 30 obtains an affirmative result in this determination, the CLPR that has been originally assigned to the target pool PL in the order of the oldest time stamps the data of the same capacity as the CLPR that is additionally assigned to the target pool PL. And the data stored in the CLPR and the like additionally assigned to the target pool PL, and sequentially erased from the CLPR and the like additionally assigned to the target pool PL, and not written in the storage device 20 The data is sequentially moved to the CLPR originally assigned to the target pool PL (SP152).

Subsequently, the performance monitoring program 30 converts the data stored in the CLPR or the like additionally assigned to the target pool PL by the process of step SP152 and not written in the storage device 20 to the target pool PL. It is determined whether or not all of the CLPRs originally assigned to can be moved (SP153). When the performance monitoring program 30 obtains a negative result in this determination, it ends this cache return processing and returns to the pool performance monitoring processing.

On the other hand, when the performance monitoring program 30 obtains a positive result in the determination at step SP153, it releases the CLPR etc. temporarily and additionally assigned to the target pool PL, and the CLPR etc. is a CLPR divided area. When the divided area is borrowed from another pool, the divided area is returned to the renting source pool PL (SP154).

Further, the performance monitoring program 30 updates the temporary allocation flag such as CLPR released in step SP154 in the CLPR management table 38 (FIG. 6) to OFF (“0”) and stores it in the corresponding allocation destination pool field 38F. Is cleared (SP155), and then the cache return process is terminated.

On the other hand, if the performance monitoring program 30 obtains a negative result in the determination at step SP151, it increases the number of repetitions by “1” (SP156), and this number of repetitions is set to a predetermined threshold (hereinafter referred to as the number of repetitions). It is determined whether or not the threshold is reached (SP157).

If the performance monitoring program 30 obtains a negative result in this determination, it waits for a certain period (for example, several minutes) (SP158), and then returns to step SP151. Then, the performance monitoring program 30 thereafter repeats the processing from step SP151 to step SP158 until a positive result is obtained in step SP151 or step SP157.

By repeating the processing from step SP151 to step SP158, for example, when the write load of the target pool PL is temporarily increased, it is possible to wait for the write load to decrease.

When the performance monitoring program 30 eventually obtains a positive result in step SP151, it executes step SP152 to step SP155 as described above. In contrast, if the performance monitoring program 30 obtains a negative result in the determination at step SP157, it ends this cache return processing.

(5) Effects of this Embodiment As described above, in the storage apparatus 3 of this embodiment, for each pool PL created in the apparatus, access of each RAID group RG that provides the pool volume PLVOL to the pool PL Acquire the access frequency and cache hit rate for each type, and if the cache hit rate during random read access and sequential read access is below the cache hit rate threshold, add the cache area capacity of the corresponding pool PL. If the access frequency of any access type exceeds the first IOPS threshold, a new pool volume PLVOL is added to the pool.

Therefore, in this storage apparatus 3, for each pool PL, the write performance is improved by adding a new pool volume PLVOL, while the read performance is the same as that of the storage apparatus 3. Depending on the situation, the performance of the pool PL is improved by adding the capacity of the cache area of the pool and / or adding a new pool volume PLVOL to the pool PL. Therefore, according to the present storage device 3, it is possible to easily perform the optimum performance improvement according to the status of the pool PL.

(6) Other Embodiments In the above-described embodiments, the case where the present invention is applied to the storage apparatus 3 configured as shown in FIG. 1 has been described. However, the present invention is not limited to this. Therefore, the present invention can be widely applied to various other storage apparatuses equipped with a virtualization function.

In the above-described embodiment, the access types are divided into four types of random read, sequential read, random write, and sequential write. For each RAID group RG, the access frequency and cache hit rate for each of these four types of access are set. Although the case where the performance is improved by using the acquired information has been described, the present invention is not limited to this, and the access type is divided into two types, read and write, for each RAID group RG. In addition, the access frequency and cache hit rate for each of these two types of access may be acquired, and performance improvement may be performed using the acquired information.

Furthermore, in the above-described embodiment, as a method of improving the read performance and the write performance, for the read performance, the capacity of the cache area allocated to the pool PL is added and / or a new pool volume PLVOL is added to the pool PL. As described above, the write performance is improved by adding a new pool volume PLVOL to the pool PL. However, the present invention is not limited to this, and the capacity of the cache area is added. Further, by adding a method other than adding a new pool volume PLVOL to the pool PL, more optimal performance improvement may be performed.

The present invention can be applied to a storage apparatus equipped with a virtualization function.

DESCRIPTION OF SYMBOLS 1 ... Computer system, 2 ... Host device, 3 ... Storage device, 20 ... Storage device, 25 ... Processor, 26 ... Local memory, 27 ... Cache memory, 30 ... Performance monitoring program, 31 ... ... Access type determination program, 32 ... Performance improvement processing program, 33 ... Batch improvement program, 35 ... RAID group performance information table, 36 ... RAID group information management table, 37 ... Pool information management table, 38 ... CLPR management table, 39... Batch schedule management table.

Claims

One or a plurality of logical volumes are defined on a storage area provided by a storage device group composed of one or a plurality of storage devices, and pooled by one or a plurality of logical volumes provided by one or a plurality of the storage device groups The virtual volume is provided to the host device, and a storage area is dynamically allocated from the pool associated with the virtual volume to the virtual volume in response to a data write request from the host device to the virtual volume. In the storage device to be allocated,
A cache memory allocated to the corresponding pool as a cache area in which the storage area is logically divided into a plurality of divided areas, and the divided areas temporarily store data read and written to the logical volume;
For each of the pools, a cache hit rate that is a ratio in which data to be read is stored in the cache area allocated to the pool at the time of read access in each storage device group, and the read access in each storage device group And a performance monitoring unit for acquiring the write access frequency, and
For each pool, when the cache hit rate at the time of the read access in the storage device group is equal to or less than a predetermined first threshold, the storage device group is assigned to the pool that provides the logical volume. A performance improvement unit for adding a capacity of a cache area and adding a new logical volume to the pool when the frequency of the read access or the write access exceeds a predetermined second threshold. A storage device.
The performance improvement unit is
The logical volume provided by the storage device group that does not provide the logical volume to the pool is added to the target pool, and is stored in the logical volume constituting the target pool. The storage apparatus according to claim 1, wherein a part of data is moved to the added logical volume.
The performance improvement unit is
When the logical volume that can be added to the target pool does not exist, the logical volume provided to a pool other than the target pool is deleted from the pool, and the logical volume is targeted The storage apparatus according to claim 2, wherein the storage apparatus is reallocated to a pool.
The performance monitoring unit
The additional cache area is allocated to the pool, and the ratio of data waiting to be written to the storage device group out of data temporarily stored in the cache area is a predetermined third threshold value. Data stored in the cache area that has been added to the pool is erased in order from the oldest data among the data stored in the cache area when the following is true: The cache area added to the pool is released after data not written in the storage area group is moved to the cache area originally assigned to the pool. Item 4. The storage device according to Item 3.
If the batch processing by the host device does not end within the scheduled time, before the next execution of the batch processing, the performance improving unit allocates the capacity of the cache area allocated to the pool used in the batch processing. The storage apparatus according to claim 1, further comprising a batch improvement unit configured to add or set to add the new logical volume to the pool.
The batch improvement unit
When the capacity of the cache area allocated to the pool used in the batch process is added, the cache area of the added capacity is released after the end of the batch process,
Based on whether or not the batch process is completed within the scheduled time, the capacity of the cache area for the addition when the cache area is added to the pool used in the batch process at the time of the next batch process. The storage device according to claim 5, wherein the storage device is set.
One or a plurality of logical volumes are defined on a storage area provided by a storage device group composed of one or a plurality of storage devices, and pooled by one or a plurality of logical volumes provided by one or a plurality of the storage device groups The virtual volume is provided to the host device, and a storage area is dynamically allocated from the pool associated with the virtual volume to the virtual volume in response to a data write request from the host device to the virtual volume. In the storage device control method to be allocated,
The storage device
The storage area is logically divided into a plurality of areas, and the area has a cache memory allocated to the corresponding pool as a cache area for temporarily storing data read from and written to the logical volume,
The storage device, for each pool, a cache hit rate that is a ratio in which the read target data is stored in the cache area allocated to the pool at the time of read access in each storage device group, and each of the storage devices A first step of obtaining the frequency of said read access and write access in a group respectively;
When the cache hit rate at the time of the read access in the storage device group is less than or equal to a predetermined first threshold for each pool, the storage device group provides the logical volume for the storage device group. The capacity of the cache area allocated to the second is added, and when the frequency of the read access or the write access exceeds a predetermined second threshold, a new logical volume is added to the pool. And a storage apparatus control method comprising the steps of:
In the second step, the storage device
The logical volume provided by the storage device group that does not provide the logical volume to the pool is added to the target pool, and is stored in the logical volume constituting the target pool. The storage apparatus control method according to claim 7, wherein a part of data is moved to the added logical volume.
In the second step, the storage device
When the logical volume that can be added to the target pool does not exist, the logical volume provided to a pool other than the target pool is deleted from the pool, and the logical volume is targeted The storage apparatus control method according to claim 8, wherein the storage apparatus is reallocated to a pool.
The additional cache area is allocated to the pool, and the ratio of data waiting to be written to the storage device group out of data temporarily stored in the cache area is equal to or less than a predetermined third threshold value In this case, the data stored in the cache area is erased in order of the capacity of the added cache area, and the data stored in the cache area added to the pool A third step of releasing the cache area added to the pool after moving data that has not been written to the storage area group to the cache area originally assigned to the pool. The method for controlling a storage apparatus according to claim 9, wherein:
In the second step, the storage device
If the batch processing by the host device does not end within the scheduled time, before the next execution of the batch processing, add the capacity of the cache area to be allocated to the pool used in the batch processing, or The storage apparatus control method according to claim 7, wherein a setting is made to add a new logical volume to the pool.
The storage device
When the capacity of the cache area allocated to the pool used in the batch process is added, the cache area of the added capacity is released after the end of the batch process,
Based on whether or not the batch process is completed within the scheduled time, the capacity of the cache area for the addition when the cache area is added to the pool used in the batch process at the time of the next batch process. The storage apparatus control method according to claim 11, wherein the storage apparatus control method is set.