WO2016194979A1 - Storage system, storage control device, storage control method, and program - Google Patents

Abstract

The present invention provides a storage system that contributes to improved performance of a storage comprising multiple storage media. The storage system includes: a storage device having at least two storage media; and a storage control device. The storage control device divides each of the two or more storage media to define a first storage area and a second storage area therein, and manages the first storage area as a buffer area and the second storage area as a storage area to be offered to a host. Furthermore, the storage control device is equipped with: an access execution unit for controlling access to the first and second storage areas; and a determination unit for determining, on the basis of the data size of write data from the host and/or on the basis of a response state of a storage medium that is specified by the host as a write destination of the write data, whether or not write data should be stored in the buffer area.

Description

Storage system, storage control device, storage control method and program

(Description of related applications)
The present invention is based on the priority claim of Japanese patent application: Japanese Patent Application No. 2015-112114 (filed on June 2, 2015), the entire contents of which are incorporated herein by reference. Shall.
The present invention relates to a storage system, a storage control device, a storage control method, and a program. In particular, the present invention relates to a storage system including a storage apparatus composed of a plurality of storage media, a storage control apparatus, a storage control method, and a program.

Recently, an SSD (Solid State Drive) using a NAND flash memory as a storage element is widely used as a storage medium in an information system. The SSD has a higher performance especially for random access than an HDD (Hard Disk Disk Drive) that has been used as a storage medium.

On the other hand, it is known that repeated random writing with a small data size for an SSD causes performance degradation. Here, the NAND flash memory used for the SSD needs to erase already stored data (old data) when updating the stored data. Data is erased in units called blocks. The block includes a plurality of pages which are the minimum units of reading (Read) and writing (Write). Therefore, when it is necessary to update a part of a block by random writing with a small data size, it is necessary to control saving and re-storing of data stored in a page not to be updated.

In SSD, the above save and re-store are performed as internal processing. Therefore, a phenomenon (write amplification) in which the amount of data actually written in the NAND flash memory inside the SSD increases as compared with the data amount of the write request (write request) to the SSD.

Furthermore, write amplification by random writing with a small data size for SSD may cause a decrease in reliability due to the lifespan of SSD. Specifically, the NAND flash memory constituting the SSD increases the occurrence rate of bit errors with respect to stored data by repeatedly updating data. For this reason, it is necessary to suppress the number of updates so that errors in the NAND flash memory are within a correctable range, and this is normally treated as the lifetime of the NAND flash memory. Furthermore, in recent years, for the purpose of increasing the capacity of the NAND flash memory, the lifetime of the NAND flash memory has been shortened due to the miniaturization of the manufacturing process and the increase in the amount of information stored in the memory cells.

Under the circumstances as described above, the reduction of the write data amount for the NAND flash memory incorporated in the SSD is an important factor in improving the reliability of the SSD. That is, from the viewpoint of ensuring reliability by improving the performance of the SSD and extending its life, it is required to reduce random writing with a small data size for the SSD.

In response to such a request, a buffer or a cache is often used to reduce a writing process with a small data size to the SSD. For example, taking the case of using a buffer as an example, write data for the SSD or NAND flash memory constituting the SSD is temporarily stored in the buffer. After that, for example, as shown in Patent Document 1 and Non-Patent Document 1, data writing (Write) is performed on the SSD when data having consecutive addresses is accumulated, thereby avoiding random writing with a small data size. Yes.

Also, the buffer for the SSD serves to conceal the occurrence of a response delay in the SSD and prevent the write performance from being degraded. That is, many SSDs use garbage collection (GC) for management of a storage area configured by an internal NAND flash memory. Garbage collection involves saving and re-storing stored data in the NAND flash memory, and can cause a delay in processing access requests from the outside. However, even in such a case, the SSD can avoid the write request when the response delay occurs in the SSD by temporarily storing the write data in the buffer and responding to the access request source.

A storage medium used as a buffer for the SSD or NAND flash memory is often a DRAM (Dynamic Random Access Memory) or the like, but other storage media may be used. For example, Non-Patent Document 2 discloses that a NAND flash memory having a short lifetime is used by using a NAND flash memory having superior performance and lifetime as a storage element of a buffer compared to a NAND flash memory having a short lifetime within an SSD. A technique for avoiding frequent updates to the memory (update by random write) is disclosed.

Further, Non-Patent Document 3 discloses a technique for using an HDD as a write-once buffer for an SSD.

JP 2010-009548 A

It should be noted that the disclosures of the above prior art documents are incorporated herein by reference. The following analysis was made by the present inventors.

As described above, in recent years, the amount of data handled by information systems has increased, and there has been an increasing demand for improved performance and increased capacity for storage (storage devices) for storing and storing data. As a result, the number of SSDs required to meet the performance and capacity requirements for storage tends to increase.

On the other hand, as the number of SSDs increases, the performance and capacity required for a buffer for reducing random writes also increase. Thus, the buffer is required to improve performance and capacity in accordance with the quantity of SSDs. However, improvement in the performance and capacity of the buffer leads to an increase in the number of storage media constituting the buffer, which causes an increase in cost (hardware cost).

Further, in a large-capacity storage constituted by a large number of storage media, striping using RAID (Redundant Array of Inexpensive Disks) or the like, or redundancy between SSDs is performed. In particular, RAID-5 and RAID-6 are widely used because both striping and data protection can be achieved. In the RAID-5 and RAID-6 data formats, redundancy is maintained by updating the parity corresponding to the update data simultaneously with the data update. In other words, in the RAID configuration as described above, since the parity is always updated regardless of the size of the updated data, a large-capacity buffer is required to prevent an increase in the amount of write data for the SSD, and the hardware cost is reduced. To increase.

In addition, as the capacity of the buffer increases, the amount of lost data (data stored only in the buffer) increases due to the occurrence of a failure. Therefore, protection of stored data is desired. First, in order to avoid data loss due to a failure of a storage medium (such as a DRAM) constituting a buffer, it is necessary to make data redundant among a plurality of SSDs. For this reason, a capacity for storing redundant data such as parity is newly required, and the hardware cost of the storage medium constituting the buffer increases.

In addition, in order to avoid data loss due to a failure in which the power supply to the storage device stops, it is necessary to configure a buffer using a non-volatile storage element or storage medium. However, for example, a high-performance and non-volatile storage element such as a MRAM (Magnetroresistive Random Access Memory) or the like is usually more expensive than a DRAM, leading to a large cost increase.

In addition to this, it is conceivable to add a battery that constantly supplies power to the DRAM to realize the non-volatility of the buffer. However, even in such a case, a new battery is required. It is difficult to avoid an increase in cost.

As described above, a buffer (particularly a non-volatile buffer with sufficient performance and capacity) necessary to prevent random writing with a small data size that affects the performance and life of the SSD is provided at low cost. It is difficult. In addition, when a buffer having sufficient performance and capacity as described above cannot be used (there is no buffer corresponding to the SDD), the performance deteriorates due to the bottleneck of the buffer and the increase in random write with a small data size for the SSD. The problem is that lifetime consumption occurs. Furthermore, when the buffer is not nonvolatile and the power supply to the buffer is stopped due to a failure, there is a problem that data stored in the buffer is lost.

In view of the above situation, an object of the present invention is to provide a storage system, a storage control device, a storage control method, and a program that contribute to improving the performance of a storage composed of a plurality of storage media.

According to a first aspect of the present invention, a storage apparatus having at least two or more storage media and a storage area of each of the two or more storage media are defined by being divided into a first storage area and a second storage area. And a storage control device that manages the first storage area as a buffer area and the second storage area as a storage area that is provided to a host, wherein the storage control apparatus includes the first and second storage areas. An access execution unit that controls access to the storage area of the host, and write data to the buffer based on at least one of a data size of the write data from the host and a response status of the storage medium designated to write the write data from the host And a determination unit that determines whether or not to store in a region.

According to a second aspect of the present invention, a storage device having at least two or more storage media is controlled, and each storage area of the two or more storage media is divided into a first storage area and a second storage area. A storage control device for managing the first storage area as a buffer area and managing the second storage area as a storage area for providing to a host, wherein the first and second storage areas are accessed. And whether to store the write data in the buffer area based on at least one of the data size of the write data from the host and the response status of the storage medium designated to write the write data from the host. A storage control device is provided.

According to a third aspect of the present invention, a storage device including at least two or more storage media is included, and each storage area of the two or more storage media is divided into a first storage area and a second storage area. And controlling access to the first and second storage areas in a storage system managed as a buffer area and the second storage area as a storage area provided to a host. Determining whether to store the write data in the buffer area based on at least one of a write data size from the host and a response status of the storage medium designated to write the write data from the host A storage control method is provided.

According to a fourth aspect of the present invention, a storage device having at least two or more storage media is controlled, and each storage area of the two or more storage media is divided into a first storage area and a second storage area. The first storage area is managed as a buffer area, and the second storage area is managed as a storage area provided to a host, and is executed by a computer installed in a storage control device, Write data based on at least one of a process for controlling access to the first and second storage areas, a data size of the write data from the host, and a response status of the storage medium designated to write the write data from the host. And a process for determining whether or not to store in the buffer area.
This program can be recorded on a computer-readable storage medium. The storage medium may be non-transient such as a semiconductor memory, a hard disk, a magnetic recording medium, an optical recording medium, or the like. The present invention can also be embodied as a computer program product.

According to each aspect of the present invention, there are provided a storage system, a storage control device, a storage control method, and a program that contribute to improving the performance of storage composed of a plurality of storage media.

It is a figure for demonstrating the outline | summary of one Embodiment. It is a figure which shows an example of the storage system structure which concerns on 1st Embodiment. 2 is a diagram illustrating an example of an internal configuration of a host and a storage apparatus according to the first embodiment. FIG. It is a figure which shows an example of a storage execution management table. It is a figure which shows an example of a storage medium state management table. It is a figure which shows an example of a mapping management table. It is a figure which shows an example of a buffer area management table. 3 is a flowchart illustrating an example of the operation of the storage system according to the first embodiment. 3 is a flowchart illustrating an example of the operation of the storage system according to the first embodiment. 3 is a flowchart illustrating an example of the operation of the storage system according to the first embodiment. 3 is a flowchart illustrating an example of the operation of the storage system according to the first embodiment. 3 is a flowchart illustrating an example of the operation of the storage system according to the first embodiment.

First, an outline of one embodiment will be described. Note that the reference numerals of the drawings attached to the outline are attached to the respective elements for convenience as an example for facilitating understanding, and the description of the outline is not intended to be any limitation.

The storage system according to an embodiment includes a storage device 201 having at least two or more storage media, and a storage control device 202. The storage control device 202 divides and defines the storage areas of the two or more storage media into a first storage area and a second storage area, and defines the first storage area as a buffer area and a second storage area. As a storage area to be provided to the host. Further, the storage control device 202 includes an access execution unit 211 that controls access to the first and second storage areas, the data size of the write data from the host, and the storage medium designated to write the write data from the host. A determination unit 212 that determines whether to store write data in the buffer area based on at least one of the response statuses.

In the above storage system, when the storage apparatus 201 has a plurality of storage media, a partial area of the storage medium is secured as a buffer. In addition, the storage control device 202 has a case where a random write with a small write data size and poor access efficiency occurs, or when the storage medium designated as the write data write destination is in a response delay state. The performance of the storage apparatus 201 is improved by storing the write data in a buffer area prepared as a shared buffer in a plurality of storage media. In addition, the storage control device 202 is sufficient to create a buffer necessary for preventing random writing with a small data size that affects the performance and life of the storage medium from a part of the storage medium constituting the storage device 201. A buffer having high performance and capacity can be provided at low cost.

Hereinafter, specific embodiments will be described in more detail with reference to the drawings. In addition, in each embodiment, the same code | symbol is attached | subjected to the same component and the description is abbreviate | omitted.

[First Embodiment]
The first embodiment will be described in more detail with reference to the drawings.

FIG. 2 is a diagram illustrating an example of a storage system configuration according to the first embodiment. Referring to FIG. 2, in the first embodiment, the system includes one or more hosts 1, a storage device 2 composed of at least two storage media, and a network 3 connecting the host 1 and the storage device 2. Including. Note that the system configuration shown in FIG. 2 is an example, and is not intended to limit the system configuration. For example, the host 1 and the storage device 2 may be directly connected without going through the network 3, or the host 1 may be configured by a plurality of devices or systems connected by the network 3.

In the system according to the first embodiment, the host 1 connected to the network 3 accesses the data held by the storage apparatus 2. The storage device 2 includes, for example, two or more storage media such as an SSD configured with a NAND flash memory.

In the first embodiment, a part of each storage medium included in the storage device 2 (within the storage area of the storage medium) includes a virtual medium 21 (corresponding to the second storage area described above) as a virtual storage medium. A virtual buffer 22 (corresponding to the first storage area) is defined in advance. That is, each storage area of the storage medium included in the storage device 2 is divided and defined as a storage area forming a virtual buffer and a storage area provided to the host as a storage area used by the host. The storage control device manages these areas (virtual buffers and virtual areas).

An OS (Operating System) excluding software (operating system) that operates on the host 1 and functions (modules) for accessing data held by the storage device 2 is defined on each storage medium included in the storage device 2 Each of the recorded virtual media 21 is handled as one (single) storage medium. That is, the virtual medium 21 is provided to the host as an original storage medium.

FIG. 3 is a diagram illustrating an example of an internal configuration of the host 1 and the storage apparatus 2 according to the first embodiment. Referring to FIG. 3, the host 1 includes an access execution unit 101, a work memory 102, a buffer target determination unit 103, a buffer storage destination determination unit 104, a response delay determination unit 105, and a buffer data movement instruction unit 106. A storage execution management table 107, a storage medium state management table 108, a mapping management table 109, and a buffer area management table 110.

The host 1 includes a module that requests the storage device 2 to read / write data, a module that processes a request from the module and controls the storage device 2 (a module corresponding to the storage control device). , Is included. FIG. 3 shows a module for controlling the storage apparatus 2 and does not show a module for requesting data read / write. That is, the internal configuration of the host 1 shown in FIG. 3 is substantially the internal configuration of the storage control device.

In the following description, three storage media 2A to 2C among the plurality of storage media included in the storage device 2 are illustrated, and the virtual medium 21 defined in the storage medium 2A is defined as the virtual medium 21A and the storage medium 2B. The virtual medium is represented as a virtual medium 21B, and the virtual medium defined as the storage medium 2C is represented as a virtual medium 21C. Similarly, the virtual buffer 22 defined in the storage medium 2A is a virtual buffer 22A, the virtual buffer 22 defined in the storage medium 2B is a virtual buffer 22B, and the virtual buffer 22 defined in the storage medium 2C is a virtual buffer 22C. write.

In FIG. 3, illustrations of modules included in the storage device 2 and controlling each storage medium are omitted.

The access execution unit 101 included in the host 1 illustrated in FIG. 3 receives an access request to the storage apparatus 2 from software (application program or the like) operating on the host 1 or the OS, and performs necessary access processing on the storage apparatus 2. Do. That is, the access execution unit 101 controls access to the storage area (the virtual medium 21 and the virtual buffer 22) of the storage medium that forms the storage device 2.

The access execution unit 101 refers to the mapping management table 109 when the acquired access request is a read request (read request). When it is determined by referring to the mapping management table 109 that the data requested to be read exists in the virtual buffer 22, the access execution unit 101 stores the storage medium that stores the data and the data. Get address information. Thereafter, the access execution unit 101 reads the data to be read stored in the virtual medium 21 or the virtual buffer 22 from the target storage medium, and returns the result to the requesting software or OS.

When the access execution unit 101 issues a read request to the storage device 2, the access execution unit 101 defines the address specified for the virtual medium 21 or the virtual buffer 22, and defines the virtual medium 21 and the virtual buffer 22 in the storage device 2. Considering this, it is converted in advance into an address in each storage medium included in the storage device 2.

Data read from the storage device 2 is not a direct request to the storage device 2, but a copy of data to be read is stored in the working memory 102 or a cache memory (not shown) corresponding to the storage device 2. If there is, the access execution unit 101 may read the copy and substitute it without requesting the storage apparatus 2 to read it.

When the received access request is a write (write request), the access execution unit 101 registers the details of the write access to each storage medium included in the storage apparatus 2 in the storage execution management table 107. Further, the access execution unit 101 instructs the buffer target determination unit 103 to determine the storage destination of the data related to the write request with respect to the received write request, and obtains the storage medium and address as the storage destination from the buffer target determination unit 103 To do.

Upon receiving the instruction, the buffer target determining unit 103 determines whether the data related to the write request should be stored in the original write address (write destination address) on the virtual medium 21 or stored in the virtual buffer 22, and the determination result To the access execution unit 101.

As in the case of issuing a read request to the storage apparatus 2, the access execution unit 101 assigns the addresses specified for the virtual medium 21 or the virtual buffer 22 to the virtual medium 21 and the virtual buffer 22 in the storage apparatus 2. In consideration of the definition, it is converted in advance to an address in each storage medium included in the storage apparatus 2.

The access execution unit 101 issues a write request to the determined storage medium and address. When the write request issued to the storage apparatus 2 is completed, the access execution unit 101 stores the corresponding data in the mapping management table 109 and the buffer area management table 110 when the write data storage destination is the virtual buffer 22. Update.

Further, when the data before being updated by the completed write access is stored in the virtual buffer 22, the access execution unit 101 sets the corresponding entry in the buffer area management table 110 to a state in which invalid data is stored. Update. Further, when the updated data is stored in the virtual medium 21, the access execution unit 101 deletes the corresponding entry in the mapping management table 109. Finally, the access execution unit 101 deletes the completed write access information from the storage execution management table 107 and notifies the requesting software or OS of the completion of data writing.

The work memory 102 temporarily stores software operating on the host 1 and write data accompanying a write request from the OS. The write data stored in the work memory 102 is stored until a write request to the storage apparatus 2 by the access execution unit 101 is completed, and is deleted at an arbitrary time (timing) after the write request is completed.

The working memory 102 may be a non-volatile memory device such as an MRAM, in addition to the DRAM often used as the memory of the host 1. Further, the working memory 102 may be realized by a cache memory (not shown) for the storage apparatus 2 or a device shared with a memory used by the software of the host 1, the OS, or the like.

The buffer target determination unit 103 determines whether to store write data in the virtual buffer 22 for the write request received from the software, OS, or the like by the access execution unit 101. Specifically, when the write data is a write request that greatly affects performance degradation and life consumption in the storage medium included in the storage device 2, the buffer target determination unit 103 determines whether the write data of the write request is related to the write request. The storage destination is determined (determined) as the virtual buffer 22. For example, when the storage medium has an SSD as a storage medium, a write having a data size equal to or smaller than a predetermined threshold (predetermined standard) that is considered to increase write amplification in the NAND flash memory inside the SSD. The write request related to the data is determined as a storage target for the virtual buffer 22. That is, the buffer target determination unit 103 determines whether or not to store the write data in the virtual buffer 22 (buffer area) based on the data size of the write data.

Also, the buffer target determination unit 103 refers to the storage medium state management table 108 and confirms whether a response delay has occurred in the storage medium that is the storage destination of the write data. At this time, when the occurrence of a response delay is confirmed in the storage medium, the buffer target determining unit 103 determines that the storage location of the write data related to the write request from the host or the like is the virtual buffer 22 regardless of the content of the write request. To do. That is, the buffer target determination unit 103 stores the write data in the virtual buffer 22 (buffer area) based on the response status (whether response delay has occurred) of the storage medium designated to write the write data. It is determined whether or not.

In this way, the buffer target determination unit 103 determines whether to store the write data in the virtual buffer 22 based on at least one of the data size of the write data and the response status of the storage medium designated as the write destination of the write data. Determine.

The in-buffer storage destination determination unit 104 determines a storage medium having the virtual buffer 22 that is the storage destination for the write data that has been subjected to the storage determination by the buffer target determination unit 103 as the storage destination. Specifically, the buffer storage location determination unit 104 refers to the storage medium state management table 108 and extracts a storage medium that is not in a response delay state. That is, when the buffer target determining unit 103 determines that the write data is stored in the virtual buffer 22, the in-buffer storage destination determining unit 104 defines the storage medium that provides the virtual buffer 22 as any storage medium. The determination as to whether the virtual buffer 22 is the write data storage destination is made based on the response status of each storage medium in which the virtual buffer 22 is defined.

Furthermore, the buffer storage location determination unit 104 refers to the buffer area management table 110, and from among a plurality of storage media, a storage medium that has a sufficient storage area for storing write data (a sufficient free space exists). One of the storage media having the virtual buffer 22 to be selected). That is, when there are a plurality of storage media that are not in the response delay state, the storage medium that provides the virtual buffer 22 is related to which storage medium the virtual buffer 22 defined in is set as the storage destination of the write data. The determination is made based on the free capacity of each virtual buffer 22 (size of an area in which invalid data is stored). The storage medium determining unit 104 in the buffer selects the storage medium having sufficient free space by, for example, selecting a storage medium having the largest free space from storage media that are not in a response delay state. Just do it.

Subsequently, the buffer storage destination determination unit 104 determines a storage destination address (address in the virtual buffer 22) in the storage medium serving as a storage destination with reference to the buffer area management table 110. For example, when storing new write data in the write-once format in the storage medium area constituting the virtual buffer 22, the in-buffer storage location determination unit 104 stores in advance the address at which the last write data was stored, The address next to the address where the write data is stored last is set as the storage address of the new write data.

The response delay determination unit 105 acquires the issue time from the storage execution management table 107 for the write access issued from the access execution unit 101 to each storage medium included in the storage device 2 at an arbitrary cycle (timing), The write access execution time is calculated by comparing with the current time. When the calculated execution time exceeds a predetermined reference time, the response delay determination unit 105 determines that the storage medium of the storage apparatus 2 is in a response delay state. That is, the response delay determination unit 105 performs threshold processing on the execution time of the write access performed for each storage medium, and determines whether the storage medium is in a response delay state according to the result of the threshold processing. Determine. When the storage medium is in a response delay state, the response delay determination unit 105 updates the state of the storage medium in the storage medium state management table 108 to the response delay state.

The response delay determination unit 105 checks the storage execution management table 107 for the write access being executed by the storage medium in the response delay state, and if a storage location change request has not been made, the buffer storage location determination unit 104 is instructed to re-determine the storage medium and address so that the virtual buffer 22 becomes the storage destination. That is, when the response delay determination unit 105 determines that the storage medium that is executing write access is in the response delay state, the response delay determination unit 105 executes the storage medium in the response delay state to the buffer storage location determination unit 104. The virtual buffer 22 is instructed to store the write data corresponding to the current write access.

The buffer data movement instructing unit 106 refers to the buffer area management table 110 at an arbitrary cycle (timing) and confirms the free capacity of the storage area constituting the virtual buffer 22 of each storage medium.

When the free space in the storage area corresponding to the virtual buffer 22 is below a predetermined reference capacity, the buffer data movement instruction unit 106 issues an instruction to secure free space in the storage area below the reference to the access execution unit. 101. Specifically, the buffer data movement instruction unit 106 reads the write data stored in the storage area constituting the virtual buffer 22 of each storage medium from the virtual buffer 22 to the work memory 102 to the access execution unit 101. The write data is instructed to be written in the original storage address of the write data in the virtual medium 21. That is, the buffer data movement instruction unit 106 determines whether or not the free capacity of each virtual buffer 22 is greater than or equal to a predetermined reference value (greater than or equal to the reference capacity), and determines that the free capacity is less than the predetermined reference value. The data stored in the virtual buffer 22 is moved to the virtual medium 21 of the storage medium designated as the writing destination from the OS or the like.

Note that writing to the original storage destination address of the virtual medium 21 is a part of the process of moving data from the virtual buffer 22 to the virtual medium 21, and therefore the storage target determination by the buffer target determination unit 103 is not performed ( (Unconditionally, the virtual medium 21 is a data storage destination).

The buffer data movement instruction unit 106 updates the mapping management table 109 and the buffer area management table 110 when the write process to the virtual medium 21 is completed.

The above-described access execution unit 101, buffer target determination unit 103, in-buffer storage destination determination unit 104, response delay determination unit 105, and buffer data movement instruction unit 106 function as a control module that controls the storage apparatus 2. The control module can also be realized by a computer program that causes a computer mounted on the host 1 (storage control device) to execute processing, which will be described in detail later, using the hardware.

The storage execution management table 107 stores and manages information related to write access that is being issued from the access execution unit 101 to each storage medium. More specifically, the storage execution management table 107 is stored in the write access issuance time, write access issuance destination storage medium, write data issuance destination address, write data size, and work memory 102 by the access execution unit 101. Stores and manages pointer information of write data.

The storage execution management table 107 stores and manages a storage medium and an address indicating the original storage destination of the write data related to the write access when the write access to be managed stores data in the virtual buffer 22. . It should be noted that the storage execution management table 107 may exist on the same device as the working memory 102, for example, a cache memory (not shown) for the storage apparatus 2 or a memory used by the host 1 software, OS, or the like. good.

FIG. 4 is a diagram illustrating an example of information relating to write access included in the storage execution management table 107. The storage execution management table 107 includes an issue time 1071, an issue destination storage medium 1072, an issue destination address 1073, an issue size 1074, a storage data pointer 1075, a request for each write access issued by the access execution unit 101. A storage medium 1076, a request address 1077, a storage location change flag 1078, and a storage location change prohibition flag 1079 are managed.

The stored data pointer 1075 indicates the storage destination address of the write data stored in the work memory 102. The request storage medium 1076 indicates the storage medium where the access execution unit 101 operates on the host 1 and the write data requested by the OS, and the request address 1077 indicates the storage destination address in the storage medium. .

Of the information included in the storage execution management table 107, the issue destination storage medium 1072 and the issue destination address 1073 indicate the storage medium actually accessed by the access execution unit 101 and the address on the storage medium, and the request storage medium 1076 and the request An address 1077 indicates an access destination (storage medium, address) received by the access execution unit 101 from the OS or the like. For example, in the entry on the first line in FIG. 4, the access execution unit 101 is requested to write data to the address 0x00100000 (request address 1077) of the storage medium 2C (request storage medium 1076). This indicates that data write is instructed to address 0x00000020 (issue destination address 1073) of storage medium 2A.

The storage destination change flag 1078 is displayed when the storage destination determination unit 104 in the buffer has already changed the write data storage destination in response to the occurrence of a response delay in the storage medium indicated by the issue destination storage medium 1072. The flag is “1”.

The storage location change prohibition flag 1079 is a flag that is set to “1” in the case of a write corresponding to data movement processing from the virtual buffer 22 to the virtual medium 21 by the buffer data movement instruction unit 106.

The format of the data included in the storage execution management table 107 is not limited to the format shown in FIG. 4. For example, the storage destination is changed from the virtual medium 21 to the virtual buffer 22 by the buffer storage destination determination unit 104. You may provide the flag which shows this explicitly.

FIG. 5 is a diagram illustrating an example of information regarding the state of the storage medium included in the storage medium state management table 108. The storage medium state management table 108 stores and manages information indicating the response delay state of each storage medium determined by the response delay determination unit 105.

The storage medium state management table 108 manages a storage medium identifier 1081 such as an ID (Identifier) for identifying each storage medium, and a response delay flag 1082 that becomes 1 when each storage medium is in a response delay state. Note that the format of the data included in the storage medium state management table 108 is not limited to the format shown in FIG. 5, and instead of the response delay flag 1082 for each storage medium, for example, a detailed load is added in addition to the response delay state. A data structure represented by a numerical value may be used.

FIG. 6 is information included in the mapping management table 109, which is data requested to be stored in the virtual medium 21 and stored in the virtual buffer 22, and the storage location of the write data in the virtual buffer 22 It is a figure which shows an example of the information showing.

The mapping management table 109 is a memory that indicates whether data requested to be stored in the virtual medium 21 by software or an OS operating on the host 1 is stored in the virtual buffer 22 and the storage location of the write data in the virtual buffer 22. Stores and manages information on the medium and the address on the storage medium. Specifically, the mapping management table 109 stores and manages a virtual medium identifier 1091, a virtual medium address 1092, a virtual buffer identifier 1093, and a virtual buffer address 1094.

The virtual medium identifier 1091 is an identifier corresponding to the storage area of the virtual medium 21 defined for each storage medium, and the virtual medium address 1092 is the address.

The virtual buffer identifier 1093 is an identifier representing the virtual buffer 22 (storage area) in which data corresponding to the virtual medium identifier 1091 and the virtual medium address 1092 is stored, and the virtual buffer address 1094 is the address. For example, in the entry in the second row in FIG. 6, the data to be stored at the address 0x00000001 (virtual medium address 1092) of the virtual medium 21A is actually stored at the address 0x00000002 (virtual buffer address 1094) of the virtual buffer 22B. Indicates that

Here, there may be a case where the data in the storage area represented by the virtual medium identifier 1091 and the virtual medium address 1092 of the virtual medium 21 does not exist in the virtual buffer 22. In such a case, the corresponding virtual buffer identifier 1093 and virtual buffer address 1094 are set to invalid values such as 0xFFFFFFFF (hexadecimal notation) (see the first line in FIG. 6).

The mapping management table 109 is not limited to the format shown in FIG. 6. For example, a flag indicating that data corresponding to the virtual medium identifier 1091 and the virtual medium address 1092 is not stored in the virtual buffer 22 is separately provided. It may be provided.

Further, the mapping management table 109 does not have the table format shown in FIG. 6. For example, the virtual medium identifier 1091 and the virtual medium address 1092 are used as keys, and the corresponding virtual buffer identifier 1093 and virtual buffer address 1094 are used as values. It may be a data structure (tree structure) like a tree to be extracted. In this case, the capacity of the mapping management table 109 can be reduced by omitting entries in which no data is stored in the virtual buffer 22.

FIG. 7 shows information included in the buffer area management table 110. Whether or not write data is stored for each storage area used in the virtual buffer 22 defined in each storage medium, and the free space in the virtual buffer 22 in each storage medium. It is a figure which shows the information showing a capacity | capacitance, and an example. The buffer area management table 110 stores information on whether write data has been stored or write data can be newly stored for each storage area used in the virtual buffer 22 defined in each storage medium. And manage.

The buffer area management table 110 stores and manages the free capacity for each storage area of the virtual buffer 22 defined in each storage medium. The buffer area management table 110 is configured for each storage medium, and the virtual buffer address 1101 representing the storage area of the virtual buffer 22 for each storage medium and the write data stored in the storage area should be originally stored. A virtual medium identifier 1102 representing a storage area of the virtual medium 21 and its virtual medium address 1103 are stored and managed. The buffer area management table 110 manages the free capacity 1104 for each storage area of the virtual buffer 22 divided for each storage medium.

When write data can be newly stored in the storage area of the virtual buffer 22 represented by the virtual buffer address 1101, for example, an invalid value such as 0xFFFFFFFF (hexadecimal notation) is stored to newly write data. Indicates that data can be stored. For example, the entry in the first row of the table for the storage medium 2A in FIG. 7 indicates that no data is stored at the address 0x00000000 of the virtual buffer 22A, and the entry in the second row is at the address 0x00000001 of the virtual buffer 22A. This indicates that data to be stored at the address 0x00000000 of the virtual medium 21C is stored.

The buffer area management table 110 is not limited to the format shown in FIG. 7. For example, when entries corresponding to the virtual buffer address 1101 exist in order from the beginning to the end of the storage area of the virtual buffer 22, the virtual buffer The address 1101 can be omitted.

The host 1 has a storage unit (not shown) in which various tables such as the storage execution management table 107, the storage medium state management table 108, the mapping management table 109, and the buffer area management table 110 are constructed.

Next, the operation of the system according to the first embodiment will be described with reference to the drawings.

[Operation when reading]
A procedure in which the access execution unit 101 receives a read request from software or OS running on the host 1 and responds to the result of the read request will be described with reference to FIGS.

8, the access execution unit 101 refers to the mapping management table 109 when receiving a read request from the software operating on the host 1 and the OS. The access execution unit 101 confirms whether or not the virtual buffer identifier 1093 and the virtual buffer address 1094 corresponding to the read access (the virtual medium identifier 1091 and the virtual medium address obtained from the read access) are registered in the table. Whether or not the read-requested data is stored in the virtual buffer 22 is confirmed (depending on whether or not the acquisition result regarding the virtual buffer address 1094 from the mapping management table 109 is a valid address).

When the data requested to be read is stored in the virtual buffer 22 (Yes in step S12), the access execution unit 101 stores the virtual buffer address 1094 acquired in step S11, and the address of the storage medium indicated by the virtual buffer identifier 1093 And a read request for the converted address is made to the storage medium (step S13). For example, referring to the mapping management table 109 shown in FIG. 6, when 0x00000002 is obtained as the virtual buffer address 1094 corresponding to read access from the OS or the like (entry in the second row of the table in FIG. 6), the access is executed. The unit 101 converts the acquired address into a real address of the storage medium 2B in consideration of the definitions of the virtual medium 21 and the virtual buffer 22, and makes a read request for the converted address. Thereafter, the access execution unit 101 executes the processing after step S15.

On the other hand, if the read requested data is not stored in the virtual buffer 22 (No determination in step S12), the access execution unit 101 sets the address for the read request destination virtual medium 21 to the address in the read request destination storage medium. Convert to Then, the access execution unit 101 uses the converted address to make a read request to the read request destination storage medium (step S14). For example, referring to the mapping management table 109 shown in FIG. 6, when 0xFFFFFFFF is obtained as the virtual buffer address 1094 corresponding to the address related to read access from the OS or the like (entry in the first row of the table in FIG. 6), The access execution unit 101 converts the address 0x00000000 of the virtual medium 21A into the real address of the storage medium 2A in consideration of the definitions of the virtual medium 21 and the virtual buffer 22, and makes a read request for the converted address.

When the operation related to the read request for the storage medium included in the storage device 2 is completed, the access execution unit 101 responds with the read data (read data) to the software and OS operating on the requesting host 1 as a result, The process related to the read request is terminated (step S15).

The access execution unit 101 performs the processing from the above steps S11 to S15, receives the read request from the software running on the host 1 and the OS, and completes the operation of responding to the result.

[Operation when writing]
2, 3, 4, 5, 7, and 9, the access execution unit 101 receives a write request from software or OS running on the host 1, and stores it in a storage medium included in the storage apparatus 2. A procedure for completing the issue of the write request will be described.

Referring to FIG. 9, the access execution unit 101 stores the write data in the work memory 102 when receiving the write request from the software operating on the host 1 and the OS. Thereafter, the access execution unit 101 registers the write request in the storage execution management table 107 and sets the storage data pointer 1075, the request storage medium 1076, the request address 1077, and the issue size 1074 (see FIG. 4). The access execution unit 101 also sets the write request storage destination information for the issue destination storage medium 1072 and the issue destination address 1073. Further, the access execution unit 101 sets the storage location change flag 1078 and the storage location change prohibition flag 1079 corresponding to the write request in the storage execution management table 107 to “0”. (Step S201)

Next, regarding the write request, the buffer target determination unit 103 acquires a determination instruction regarding whether to store the write request in either the virtual medium 21 or the virtual buffer 22 from the access execution unit 101, and the size of the write data is It is determined whether it is below a predetermined reference value (step S202).

When the size of the write data is equal to or smaller than a predetermined reference value (Yes determination in step S203), the buffer target determination unit 103 determines that the write data related to the write request is stored in the virtual buffer 22, The process after step S207 is executed.

When the size of the write data is larger than the predetermined reference value (No determination in step S203), the buffer target determination unit 103 refers to the storage medium state management table 108 shown in FIG. 5 and stores data related to the write request. It is confirmed whether or not the storage medium including the virtual medium 21 is in a response delay state (step S204).

When the storage medium including the virtual medium 21 is in a response delay state (Yes determination in step S205), the buffer target determination unit 103 determines to store the write data related to the write request in the virtual buffer 22, and the process from step S207 onward Execute the process.

When the storage medium including the virtual medium 21 is not in a response delay state (No determination in step S205), the buffer target determination unit 103 sends the original (original) write data related to the write request to the access execution unit 101. The virtual medium 21 that is the storage destination and the address for the virtual medium 21 are notified as the storage destination, the buffer target determination process is completed, and the processes after step S211 are executed (step S206).

In step S207, the buffer target determination unit 103 instructs the in-buffer storage location determination unit 104 to change the storage location of the write data related to the write request.

When receiving the instruction regarding the change of the storage location of the write data related to the write request, the buffer storage location determination unit 104 refers to the storage medium status management table 108 shown in FIG. 5 and extracts a storage medium that is not in the response delay state ( Step S208).

Subsequently, the in-buffer storage location determination unit 104 refers to the buffer area management table 110 shown in FIG. 7, and determines the free capacity 1104 of the virtual buffer 22 included in each storage medium that is not in the response delay state, extracted in step S208. Check. Thereafter, the in-buffer storage destination determination unit 104 determines, for example, the virtual buffer 22 having the largest free capacity 1104 as the storage destination (step S209).

The in-buffer storage destination determination unit 104 refers to the buffer area management table 110 to determine the address of the area where the write data is stored in the virtual buffer 22 determined as the storage destination in step S209, and the buffer area management table 110 update the virtual medium identifier 1102 and virtual medium address 1103 corresponding to the virtual buffer address 1101 corresponding to the determined address of 110 with the storage destination virtual medium and address before the change of the write data (virtual buffer address and virtual medium identifier) Update the fields related to). At this time, the in-buffer storage location determination unit 104 updates the free space 1104 of the buffer area management table 110 with a value obtained by subtracting the capacity of the area used for storing write data, and stores information on the storage destination virtual buffer 22. In addition, the access execution unit 101 is notified (step S210).

When step S206 or step S210 is completed, the access execution unit 101 executes the processing after step S211.

The access execution unit 101 converts the storage destination address notified from the buffer target determination unit 103 or the in-buffer storage destination determination unit 104 into an address corresponding to the storage medium (step S211). That is, the access execution unit 101 converts the notified address of the virtual medium 21 or the virtual buffer 22 into the address of the storage medium in consideration of the definitions of the virtual medium 21 and the virtual buffer 22.

The access execution unit 101 issues a write command to the address of the storage medium including the virtual medium 21 or the virtual buffer 22 determined as the storage destination and converted in step S211 (step S212).

The access execution unit 101 updates the issue time 1071, the issue destination storage medium 1072, and the issue destination address 1073 of the storage execution management table 107 corresponding to the write command issued in step S212, and ends the processing (step S213).

In the flowchart shown in FIG. 9, a plurality of steps (processes) are described in order, but the execution order of the steps to be executed is not limited to the description order. For example, in the flowchart shown in FIG. 9, the size of the write data is verified in steps S202 and S203, and the response status of the storage medium designated as the storage destination is verified according to the result (steps S204 and S205). The order of the two verification operations may be reversed. More specifically, when the write destination storage medium is in a response delay state, the buffer target determining unit 103 may set the write data storage destination as the virtual buffer 22 regardless of the size of the write data ( Step S207 and subsequent steps may be executed).

When it is determined that the write data is to be stored in the virtual buffer 22 due to the data size of the write data, the access execution unit 101 writes the write data to the virtual buffer 22 in a write-once format. Issue a write command. In other words, when it is determined that the write data is stored in the virtual buffer 22 according to the response status of the storage medium to which it is written, regardless of the size of the write data, the write data is not necessarily in the write-once format. Thus, it may not be written to the virtual buffer 22 (a predetermined area of the virtual buffer 22 may be updated).

By the processing from step S201 to step S213, the access execution unit 101 receives a write request from the software and OS operating on the host 1, and completes issuing the write request to the storage medium included in the storage device 2. Is completed.

[Write completion response action]
Next, referring to FIG. 2, FIG. 3, FIG. 4, FIG. 5, FIG. 6 and FIG. 10, the write request issued to the storage medium is completed, The procedure for sending a completion response will be described.

When the access execution unit 101 detects the completion of writing in the storage medium, the access execution unit 101 refers to the storage execution management table 107, and refers to the storage location change flag 1078 and the storage location change prohibition flag 1079 from the corresponding write request information (step S31). ).

When any of the storage location change flag 1078 and the storage location change prohibition flag 1079 referred to in step S31 is “1” (Yes in step S32), the access execution unit 101 receives the software and OS operating on the host 1. Since the write operation does not correspond to the write request, the processing from step S35 is executed.

When both the storage destination change flag 1078 and the storage destination change prohibition flag 1079 referred to in step S31 are “0” (No determination in step S32), the access execution unit 101 stores the storage execution management table 107 shown in FIG. By referring to the issue destination storage medium 1072, the issue destination address 1073, the request storage medium 1076, and the request address 1077, the virtual buffer identifier 1093 and the virtual buffer address 1094 corresponding to the virtual medium identifier 1091 and the virtual medium address 1092 of the mapping management table 109 are obtained. Is updated (step S33).

The access execution unit 101 refers to the request storage medium 1076 and the request address 1077 of the storage execution management table 107, identifies the write request from the software and OS operating on the host 1, and returns a write request completion response (step S34).

The access execution unit 101 refers to the storage medium state management table 108, acquires the response delay flag 1082 of the storage medium identifier 1081 that matches the storage medium for which writing has been completed, and determines whether the storage medium is in the response delay state. (Check whether the response delay flag 1082 is “1”; step S35).

When the storage medium for which writing has been completed is not in the response delay state (No determination at step S36), the access execution unit 101 executes the processing after step S38.

When the storage medium for which writing has been completed is in the response delay state (Yes determination in step S36), the access execution unit 101 responds to the storage medium for which writing has been completed with a response delay flag 1082 in the storage medium state management table 108. Is set to “0” (the state of the storage medium is updated to a state that is not a response delay; step S37).

The access execution unit 101 deletes the entry corresponding to the completed write and the write data stored in the work memory 102 from the storage execution management table 107, and ends the process (step S38).

By performing the above steps S31 to S38, the operation relating to the write request issued to the storage medium is completed, and the operation for responding to the corresponding write request from the software operating on the host 1 and the OS is completed.

[Change storage location to virtual buffer]
Next, referring to FIG. 2, FIG. 3, FIG. 4, FIG. 5, FIG. 9 and FIG. 11, the access execution unit 101 detects the response delay state in each storage medium and writes to the storage medium being issued A procedure for changing the storage destination to the virtual buffer 22 so that the write data is stored in a storage medium different from the storage medium regarding the request will be described.

The response delay determination unit 105 refers to the issuance time 1071 in each entry of the storage execution management table 107, and determines that the storage medium in which the elapsed time from the write request issuance based on the current time is in a response delay state. It is confirmed whether it is within the time to perform (step S41).

When the elapsed time from issuance is within a certain range (Yes determination in step S42), the response delay determination unit 105 executes the processing after step S47.

When the elapsed time from the issue has passed a certain time (No determination in step S42), the response delay determination unit 105 determines that the storage medium issuing the write request is in the response delay state, and stores execution management The issue destination storage medium 1072 is identified with reference to the issue destination storage medium 1072 of the table 107. Further, the response delay determination unit 105 updates the response delay flag 1082 corresponding to the storage medium identifier 1081 corresponding to the storage medium in the storage medium state management table 108 to “1”, that is, the response delay state (step S43). ).

The response delay determination unit 105 refers to the storage location change flag 1078 and the storage location change prohibition flag 1079 of the entry referred to in step S41 in the storage execution management table 107, and has already performed storage location change processing for write access. Whether the storage location is prohibited from being changed (step S44).

As a result of the confirmation in step S44, when the storage destination has been changed or the change of the storage destination is prohibited (Yes determination in step S45), the response delay determination unit 105 executes the processing from step S47.

As a result of the confirmation in step S44, when the storage destination has not been changed and the storage destination can be changed (No determination in step S45), the response delay determination unit 105 notifies the in-buffer storage destination determination unit 104 of the response delay state. The storage medium and address corresponding to the virtual buffer 22 serving as the storage destination for write access are instructed (instruction for re-determining the storage location of the write data corresponding to the entry of interest; step S46).

The response delay determination unit 105 changes the entry of interest in the storage execution management table 107 to the next entry (determines the response delay of the write request related to the next entry). If the last entry is referenced in step S41, the response delay determination unit 105 sets the target entry as the first entry. The response delay determination unit 105 re-executes the processing from step S41 after changing the entry of interest (step S47).

The processing from the above steps S41 to S47 is performed, and the response delay determination unit 105 detects the response delay state in each storage medium, and changes the storage destination in the virtual buffer 22 regarding the write request issued to the storage medium. (Operation of storing data in the virtual buffer 22 so that the write data is stored in a storage medium different from the storage medium in the response delay state).

The in-buffer storage destination determination unit 104 and the access execution unit 101, which are instructed to re-determine the write data storage destination in step S46, re-store the write data in the virtual buffer 22 in accordance with the processing from step S208 shown in FIG. Determine the destination and complete the issue of the write request after changing the storage destination.

Further, when the process from step S41 is re-executed after the process of step S47, for example, the response delay determination process may be performed at an appropriate frequency by pausing the process for an arbitrary time.

[Secure virtual buffer capacity]
Next, referring to FIGS. 2, 3, 4, 6, 7, 8, and 12, when the free space of each storage medium constituting the virtual buffer 22 is insufficient, the virtual A procedure for moving write data from the buffer 22 to the virtual medium 21 to secure free space will be described.

Referring to FIG. 12, the buffer data movement instructing unit 106 acquires the free capacity 1104 corresponding to the storage medium in question in the buffer area management table 110, and the free capacity of each virtual buffer 22 is insufficient. It is confirmed whether or not the capacity exceeds the reference capacity for determining (step S51).

When the free capacity is equal to or larger than the reference capacity for determining the lack of capacity (Yes determination in step S52), the buffer data movement instruction unit 106 changes the storage medium of interest. Specifically, when a storage medium is identified by a number, the buffer data movement instruction unit 106 focuses on the next storage medium. If the number of the target storage medium is the end, the buffer data movement instructing unit 106 re-executes the processing from step S51 after changing the focus to the storage medium with the first number (step S53). .

When the free capacity is less than the standard capacity for determining the shortage of capacity (No in step S52), the buffer data movement instruction unit 106 refers to the buffer area management table 110, stores the write data, and Then, one entry for moving the write data to the virtual medium 21 is selected from any entries whose data has not been moved to the virtual medium 21 (step S54).

The buffer data movement instruction unit 106 instructs the access execution unit 101 to read out the write data selected in step S54 from the virtual buffer 22 and store it in the work memory 102. The access execution unit 101 reads the write data from the storage medium using the procedure in steps S11 to S14 in FIG. 8 and stores it in the work memory 102 (step S55).

The buffer data movement instruction unit 106 refers to the buffer area management table 110, and acquires the virtual medium identifier 1102 and the virtual medium address 1103 representing the storage location of the write data for the write data stored in the work memory 102 in step S55. . Then, the buffer data movement instruction unit 106 sets the storage destination address of the write data on the work memory 102 in the storage execution management table 107, the storage data pointer 1075, and the write data size in the issue size 1074. Further, the buffer data movement instruction unit 106 sets the storage medium and address information of the write data in the previously acquired virtual medium identifier 1102 and virtual medium address 1103. Further, the buffer data movement instruction unit 106 sets the storage destination change prohibition flag 1079 corresponding to the write request in the storage execution management table 107 to “1” (step S56).

The buffer data movement instruction unit 106 instructs the access execution unit 101 to execute a write request related to the write data stored in the work memory 102 registered in the storage execution management table 107 in step S56 (step S57).

When the access execution unit 101 instructed in step S57 completes execution of the write request, the buffer data movement instruction unit 106 refers to the virtual medium identifier 1102 and the virtual medium address 1103 from the entry in the buffer area management table 110 selected in step S54. Then, for the entries corresponding to the virtual medium identifier 1102 and the virtual medium address 1103 in the mapping management table 109, the setting values of the virtual buffer identifier 1093 and the virtual buffer address 1094 are changed to a value (for example, 0xFFFFFFFF) indicating that the entry is invalid. Update (step S58).

The buffer data movement instructing unit 106 updates the entry in the buffer area management table 110 selected in step S54 to a state in which valid write data is not stored, that is, a state in which data can be newly stored. Further, the buffer data movement instruction unit 106 uses the value of the free capacity 1104 in the buffer area management table 110 corresponding to the virtual buffer 22 in which the free capacity is insufficient as the capacity of the write data that has been moved to the virtual medium 21. After adding the minutes, the process is re-executed from step S51 (step S59).

When the processing from the above steps S51 to S59 is performed and the free space of the storage area of each storage medium constituting the virtual buffer 22 is insufficient, the write data is moved from the virtual buffer 22 to the virtual medium 21 to free up the free space. The securing operation is completed.

The access execution unit 101 instructed to execute the write request in step S57 stores a write request for storing the data stored in the work memory 102 in the virtual medium 21 in accordance with the processing from step S202 shown in FIG. Execute processing to be issued to the medium.

Further, when the buffer data movement instruction unit 106 re-executes the process from step S51 after the process of step S53, for example, the process for determining and securing the lack of free space at an appropriate frequency by pausing the process for an arbitrary time, for example. May be performed.

As described above, in the system according to the first embodiment, a part of the storage area of a storage medium (for example, SSD) is used as a buffer. At this time, when a response delay occurs in a storage medium having a storage area constituting the buffer, it is necessary to avoid that the data storage destination for the buffer becomes a storage medium in the response delay state. Therefore, in addition to the randomness of the write request that affects the performance and life of the storage medium, the response delay related to the storage medium including the original storage destination storage area of the write data and the storage medium having the storage area as the storage destination in the buffer Whether to store the write data in the buffer is determined in consideration of the occurrence of this.

Also, writing to the storage medium in the response delay state is avoided by selecting the storage area for the write data in the buffer from the storage areas included in the storage medium that is not in the response delay state. Further, since the storage medium constituting the buffer is included in the storage medium itself, it is necessary to avoid the influence when a response delay occurs in the storage medium that is executing the write. Therefore, when occurrence of a response delay is detected in the storage medium that is executing the write, the storage medium that is not in the response delay state is selected and executed for the data stored by the write that is being executed.

As described above, in the first embodiment, a part of the storage area included in each storage medium constituting the storage such as SSD is used as a shared buffer area for the storage medium. In order to realize the sharing of the buffer area, the host 1 (storage control device) has a buffer target determination function (buffer target determination unit 103), a buffer storage destination determination function (buffer storage destination determination unit 104), a response delay determination. Function (response delay determination unit 105), buffer data movement instruction function (buffer data movement instruction unit 106), storage execution management function (management of storage execution management table 107), storage medium state management function (storage medium state management table 108) Management), a mapping management function (management of the mapping management table 109), and a buffer area management function (management of the buffer area management table 110).

In order to realize a buffer with performance, capacity, and non-volatility according to the number of storage media, the buffer target determination function stores the data requested to be written by the host in a buffer secured on the storage medium. Make a decision. The buffer target determination function determines whether write data should be temporarily stored in the buffer for each write request to the storage medium constituting the storage apparatus. The buffer target determination function buffers the write data when the storage medium, which is the original storage destination for each write request, is in a response delay state, or when it is determined that the performance of the storage medium is reduced and the life is consumed. (Temporary buffer; virtual buffer). For example, when the storage medium is an SSD, the write determination of a write that leads to write degradation in the NAND flash in the SSD causes a write amplification. Judgment can be made according to the size.

The buffer storage location determination function determines the storage medium and address to be stored in the buffer. The storage destination determination function in the buffer determines a storage medium that is a storage destination and a storage destination address in the storage medium for the write data that is determined to be stored in the buffer in the buffer target determination function. The storage destination determination function in the buffer is the response delay state of each storage medium acquired from the storage medium state management function and the storage medium acquired from the buffer area management function for the storage medium and address that is the storage destination in the write data buffer. This is determined using the usage status of the buffer storage area.

The response delay determination function determines the state of response delay in each storage medium, and when a response delay occurs, determines that the storage destination is another storage medium for the write access being executed and makes a write request again. . The response delay determination function determines whether or not each storage medium is in a response delay state from the execution time of the write command for each storage medium managed by the storage execution management function. The response delay determination function notifies the storage medium state management function of the determination result. In addition, the response delay determination function, for a write access being executed in a response-delayed storage medium, regardless of whether or not the write access is storing data in the buffer, Is instructed to write again.

The buffer data movement instruction function secures the free space of the buffer by storing the write data in the buffer at the address of the original storage medium when the buffer capacity is insufficient. The buffer data movement instruction function reads the write data stored in the buffer and the address of the original storage medium and the address when the free space in which the write data can be stored in the buffer storage area of each storage medium decreases. Is written in the area pointed to, and free space is secured by making the data re-storable in the area where the write data is stored on the buffer.

The storage execution management function manages the write request being executed for each storage medium. The storage execution management function manages the write access start time, the write data storage destination address, and the data stored by the write access for the write access being executed for each storage medium. The storage execution management function also manages the storage medium and the address where the data stored by the write access is supposed to be stored if the write access to be managed stores data in the buffer. To do.

The storage medium state management function manages the response delay state of each storage medium determined by the response delay determination function. The storage medium state management function receives and manages response delay state information regarding each storage medium determined by the response delay determination function.

The mapping management function manages the storage medium and storage destination address information that are the original storage destinations for each data stored in the buffer. The mapping management function manages whether or not the write data corresponding to the address of each storage medium is stored with respect to the buffer configured by the storage area for the buffer secured on each storage medium. In addition, when the write data is stored in the buffer, the mapping management function stores the storage medium and address information on the buffer, and the storage medium that is the original storage destination for each stored write data. Manages storage address information.

The buffer area management function manages the usage status of the storage areas secured on each storage medium constituting the buffer. The buffer area management function has already stored buffered write data for each address in the storage area for the buffer storage area secured on each storage medium, or can write new write data. Manage whether there is.

The system according to the first embodiment can realize, for example, a buffer having performance, capacity, and non-volatility corresponding to the number of SSDs used in a storage that uses SSDs as storage media at low cost. Specifically, a part of the SSD storage area is secured as a buffer area, and by combining the buffer areas secured by a plurality of SSDs, the SSD is included in the storage apparatus (connected). Configure a shared buffer. Since the number of areas constituting the shared buffer is at most equal to the number of SSDs connected to the system, a buffer capacity proportional to the number of SSDs can be obtained.

Also, since a part of the SSD, which is a non-volatile storage medium, is used as a buffer, data loss due to a power failure can be avoided, and at the same time, no additional storage medium is required and the mounting cost can be reduced. Then, in order to prevent random writing with a small data size for each SSD, data is stored in a write-once format in a storage area (virtual buffer) on each SSD secured as a buffer area.

Since the random write with a small data size for the SSD is converted into a buffer area, that is, continuous writing to the SSD, the performance of the SSD can be improved and the life can be extended by reducing the random write as in the conventional buffer. That is, write data for which a random write request has been made to each storage medium is written continuously to a shared buffer in each storage medium (that is, for data addition rather than data update). By suppressing the fragmentation of effective write data, for example, the amount of effective write data movement associated with erasure in units of blocks, which is a cause of write amplification that occurs when SSD is used as a storage medium. Can be reduced, and the performance improvement and life extension of the SSD can be realized.

In addition, by performing striping on the buffer area for a plurality of SSDs, buffer performance proportional to the number of SSDs can be obtained. Further, by providing redundancy for the data stored in the buffer using RAID or the like between the buffer areas of each SSD, it is possible to avoid data loss due to a failure in the SSD and to protect the data stored in the buffer. .

As described above, in the storage system according to the first embodiment, a non-volatile, sufficient performance and capacity buffer for the SSD as a storage medium is provided at a lower cost than when the method according to the first embodiment is not applied. It is possible to reduce random data with a small data size, improve performance, and extend the service life. In addition, since the performance and capacity in proportion to the number of SSDs that are storage media for the buffer are realized, it is possible to prevent the buffer from becoming a bottleneck for write access to the SSD.

A part or all of the above embodiments can be described as follows, but is not limited to the following.
[Form 1]
This is the same as the storage system according to the first aspect described above.
[Form 2]
The access execution unit
The storage system according to mode 1, wherein when the determination unit determines to store write data in the buffer area based on the data size of the write data from the host, the write data from the host is added to the buffer area.
[Form 3]
The storage control device
When the determination unit determines to store the write data in the buffer area,
Of the storage media forming the buffer area, the determination as to which storage medium the first storage area defined as the storage destination of the write data is made for each storage medium in which the first storage area is defined. The storage system according to mode 1 or 2, further comprising a storage destination determination unit that performs based on the response status.
[Form 4]
The storage location determination unit
The storage medium forming the buffer area is determined based on the free capacity of each of the first storage areas, with respect to which storage medium the first storage area defined as the storage destination of the write data is set. The storage system of form 3.
[Form 5]
The storage control device
Performing threshold processing on the execution time of the write access performed for each of the storage media, determining whether or not the storage media is in a response delay state according to the result of the threshold processing;
When it is determined that the storage medium that is executing write access is in the response delay state, the write data corresponding to the write access that is being executed by the storage medium that is in the response delay state is sent to the storage destination determination unit. The storage system according to mode 3 or 4, further comprising a delay determination unit that instructs to store the data in the buffer area.
[Form 6]
The storage control device
It is determined whether or not the free space of each of the first storage areas is greater than or equal to a predetermined reference value, and the data stored in the first storage area that has been determined that the free capacity is less than the predetermined reference value The storage system according to any one of aspects 1 to 5, further comprising a data moving unit that moves the data to the second storage area of the storage medium designated as the write destination from the host.
[Form 7]
The storage according to any one of forms 1 to 6, wherein the storage control device manages the first storage area defined in each of the two or more storage media as a shared buffer for the two or more storage media. system.
[Form 8]
The storage control apparatus according to the second aspect described above.
[Form 9]
This is the same as the storage control method according to the third aspect described above.
[Mode 10]
It is as the program which concerns on the above-mentioned 4th viewpoint.
Forms 8 to 10 can be developed into forms 2 to 7, as with form 1.

In addition, each disclosure of the above cited patent documents, etc. shall be incorporated by reference into this document. Within the scope of the entire disclosure (including claims) of the present invention, the embodiments and examples can be changed and adjusted based on the basic technical concept. In addition, various combinations or selections of various disclosed elements (including each element in each claim, each element in each embodiment or example, each element in each drawing, etc.) within the scope of the entire disclosure of the present invention. Is possible. That is, the present invention of course includes various variations and modifications that could be made by those skilled in the art according to the entire disclosure including the claims and the technical idea. In particular, with respect to the numerical ranges described in this document, any numerical value or small range included in the range should be construed as being specifically described even if there is no specific description.

DESCRIPTION OF SYMBOLS 1 Host 2, 201 Storage apparatus 2A-2C Storage medium 3 Network 21 Virtual medium 22 Virtual buffer 101, 211 Access execution part 102 Work memory 103 Buffer object judgment part 104 In-buffer storage destination decision part 105 Response delay judgment part 106 Buffer data movement Instructing unit 107 Storage execution management table 108 Storage medium state management table 109 Mapping management table 110 Buffer area management table 202 Storage controller 212 Judgment unit 1071 Issue time 1072 Issue destination storage medium 1073 Issue destination address 1074 Issue size 1075 Stored data pointer 1076 Request Storage medium 1077 Request address 1078 Storage location change flag 1079 Storage location change prohibition flag 1081 Storage media identifier 1082 Response delay flag 1091 Virtual media information Child 1092 virtual media address 1093 virtual buffer identifier 1094 virtual buffer address 1101 virtual buffer address 1102 virtual media identifier 1103 virtual media address 1104 space

Claims (10)

1. A storage device having at least two or more storage media;
   The storage area of each of the two or more storage media is defined by being divided into a first storage area and a second storage area, and the first storage area is provided as a buffer area and the second storage area is provided to a host. A storage control device that manages the storage area
   Including
   The storage control device
   An access execution unit for controlling access to the first and second storage areas;
   A determination unit that determines whether to store the write data in the buffer area based on at least one of a data size of the write data from the host and a response status of the storage medium designated to write the write data from the host; ,
   A storage system.
2. The access execution unit
   The storage system according to claim 1, wherein when the determination unit determines to store write data in the buffer area based on a data size of write data from the host, the write data from the host is added to the buffer area. .
3. The storage control device
   When the determination unit determines to store the write data in the buffer area,
   Of the storage media forming the buffer area, the determination as to which storage medium the first storage area defined as the storage destination of the write data is made for each storage medium in which the first storage area is defined. The storage system according to claim 1, further comprising a storage destination determination unit that performs based on a response status.
4. The storage location determination unit
   The storage medium forming the buffer area is determined based on the free capacity of each of the first storage areas, with respect to which storage medium the first storage area defined as the storage destination of the write data is set. The storage system according to claim 3.
5. The storage control device
   Performing threshold processing on the execution time of the write access performed for each of the storage media, determining whether or not the storage media is in a response delay state according to the result of the threshold processing;
   When it is determined that the storage medium that is executing write access is in the response delay state, the write data corresponding to the write access that is being executed by the storage medium that is in the response delay state is sent to the storage destination determination unit. The storage system according to claim 3, further comprising a delay determination unit that instructs to store the data in the buffer area.
6. The storage control device
   It is determined whether or not the free space of each of the first storage areas is greater than or equal to a predetermined reference value, and the data stored in the first storage area that has been determined that the free capacity is less than the predetermined reference value The storage system according to any one of claims 1 to 5, further comprising a data moving unit that moves the data from the host to the second storage area of the storage medium designated as a writing destination.
7. 7. The storage control device according to claim 1, wherein the storage control device manages the first storage area defined in each of the two or more storage media as a shared buffer for the two or more storage media. 8. Storage system.
8. Controlling a storage device having at least two or more storage media;
   The storage area of each of the two or more storage media is defined by being divided into a first storage area and a second storage area, and the first storage area is provided as a buffer area and the second storage area is provided to a host. A storage control device that manages as a storage area,
   An access execution unit for controlling access to the first and second storage areas;
   A determination unit that determines whether to store the write data in the buffer area based on at least one of a data size of the write data from the host and a response status of the storage medium designated to write the write data from the host; ,
   A storage control device.
9. Including a storage device having at least two or more storage media;
   The storage areas of each of the two or more storage media are defined by being divided into a first storage area and a second storage area, the first storage area is a buffer area, and the second storage area is a host. In a storage system managed as a storage area to be provided,
   Controlling access to the first and second storage areas;
   Determining whether to store the write data in the buffer area based on at least one of the data size of the write data from the host and the response status of the storage medium designated to write the write data from the host;
   Including a storage control method.
10. Controlling a storage device having at least two or more storage media;
    The storage area of each of the two or more storage media is defined by being divided into a first storage area and a second storage area, and the first storage area is provided as a buffer area and the second storage area is provided to a host. A storage area management program that is executed by a computer installed in a storage control device,
    Processing for controlling access to the first and second storage areas;
    A process of determining whether to store write data in the buffer area based on at least one of a data size of write data from the host and a response status of a storage medium designated to write the write data from the host;
    A program that executes

Images