WO2019244656A1

WO2019244656A1 - Storage device

Info

Publication number: WO2019244656A1
Application number: PCT/JP2019/022542
Authority: WO
Inventors: 周吾小川
Original assignee: 日本電気株式会社
Priority date: 2018-06-22
Filing date: 2019-06-06
Publication date: 2019-12-26
Also published as: JP7075067B2; JPWO2019244656A1

Abstract

According to the present invention, a controller provides access to a virtual storage medium comprising a plurality of storage medium areas. A data transmission means performs data transmission between a node to which the controller is connected and a memory of a host. When receiving an access request for the virtual storage medium via a network from a host, the controller divides, into partial areas, an area to be accessed on the virtual storage medium, and secures a buffer for each partial area. The controller controls, for each partial area, data transmission between an area of the storage medium which corresponds to the partial area, and a buffer corresponding to the partial area, and data transmission by the data transmission means between the buffer corresponding to the partial area and an area corresponding to the partial area in the memory of the host.

Description

Storage device

The present invention relates to a storage device, a data transfer method, and a recording medium.

For the purpose of sharing and expanding storage in a computer system, a virtual storage medium (hereinafter, referred to as a virtual storage medium) configured from an area of a storage medium connected to a storage device is transferred to a host computer (hereinafter, referred to as a virtual storage medium) via a network. (Referred to as Patent Document 1).

When such a storage device receives a read request designating an access target area of a virtual storage medium from a host, it secures a buffer in the storage device, reads data from the access target area, and stores the data in the buffer. At this time, in practice, the storage device reads data from the storage medium area allocated to the access target area and stores the data in the buffer. When all the read data has been stored in the buffer, the storage device starts the operation of reading the read data from the buffer and transferring the read data to the host memory via the network.

Further, when the storage device receives a write request specifying the access target area of the virtual storage medium from the host, the storage device secures a buffer in the storage device and transfers the write data stored in the memory of the host to the buffer via the network. . Then, when all the write data has been stored in the buffer, the storage device reads the write data from the buffer and writes the write data to the access target area of the virtual storage medium. At this time, the storage device actually writes the write data to the area of the storage medium allocated to the access target area.

Re-spec WO2016 / 002325

However, in the above-described storage device, in the case of a read request, the buffer is transferred from the buffer to the host memory unless all read data in the storage medium area allocated to the access target area of the virtual storage medium is stored in the buffer. Since the data transfer cannot be started, the read performance from the host to the virtual storage medium is reduced.

Further, in the above-described storage apparatus, in the case of a write request, unless all write data in the memory of the host is stored in the buffer, the buffer is transferred to the storage medium area allocated to the access target area of the virtual storage medium. Since the above write data cannot be written, the write performance from the host to the virtual storage medium decreases.

An object of the present invention is to provide a storage device that solves the above-mentioned problem, that is, the problem that the access performance from the host to the virtual storage medium is reduced.

A storage device according to an embodiment of the present invention includes:
A controller that provides access to a virtual storage medium configured from a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network;
Data transfer means between the node connected to the controller and the memory of the host,
The controller divides an access target area on the virtual storage medium accessed by the access request into partial areas, reserves a buffer for each of the partial areas, and, for each of the partial areas, Data transfer between an area of a storage medium and the buffer, and data transfer between the buffer corresponding to the partial area and the area corresponding to the partial area in the memory of the host by the data transfer means. It is configured to control.

Further, a data transfer method according to another aspect of the present invention includes:
Between a node connected to a controller that provides access to a virtual storage medium composed of a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network, and the memory of the host; Data transfer method,
Dividing an access target area on the virtual storage medium accessed by the access request into partial areas,
Securing a buffer for each partial area,
For each of the partial areas, data transfer between the area of the storage medium corresponding to the partial area and the buffer corresponding to the partial area; and the buffer and the host corresponding to the partial area by the data transfer unit. And data transfer to and from an area corresponding to the partial area in the memory.

Further, a computer-readable recording medium according to another embodiment of the present invention,
Computer
A controller that provides access to a virtual storage medium configured from a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network;
Causing the controller to function as data transfer means between the connected node and the memory of the host,
The controller divides an access target area on the virtual storage medium accessed by the access request into partial areas, reserves a buffer for each of the partial areas, and, for each of the partial areas, Data transfer between an area of a storage medium and the buffer corresponding to the partial area; and a transfer of the buffer corresponding to the partial area by the data transfer unit and an area corresponding to the partial area in the memory of the host. Record a program that is configured to control data transfer between.

According to the present invention having the above-described configuration, it is possible to enhance the access performance of the host to the virtual storage medium.

FIG. 1 is a diagram illustrating an entire system configuration according to a first embodiment of the present invention. FIG. 3 is a diagram illustrating a configuration of a controller that executes a process in response to an access request from an application according to the first embodiment of the present invention. FIG. 3 is a diagram illustrating an example of a format of a mapping table for managing a correspondence relationship between an identifier and an address of a storage medium mapped to each area of a virtual storage medium according to the first embodiment of this invention. FIG. 5 is a diagram illustrating a procedure in which a transfer storage area calculation unit calculates a storage medium including a transfer storage area corresponding to an access request and an address on the storage medium in the first embodiment of the present invention. FIG. 4 is a diagram illustrating an example of a correspondence relationship between a transfer storage area and a start address of a transfer memory area calculated from an address representing an area to be accessed on a virtual storage medium and an access size according to the first embodiment of this invention. FIG. 5 is a diagram illustrating an example of a format of an access processing table for managing information on data transfer processing required for an access request from an application according to the first embodiment of this invention. 4 is a flowchart illustrating a procedure in which a controller processes an access request to a virtual storage medium from an application according to the first embodiment of this invention. 4 is a flowchart illustrating a procedure in which a controller processes an access request to a virtual storage medium from an application according to the first embodiment of this invention. 4 is a flowchart illustrating a procedure in which an access processing execution management unit included in a controller executes a data transfer process for an access request in the first embodiment of the present invention. 4 is a flowchart illustrating a procedure in which an access processing execution management unit included in a controller executes a data transfer process for an access request in the first embodiment of the present invention. FIG. 9 is a block diagram of a storage device according to a second embodiment of the present invention. FIG. 11 is a diagram illustrating an example of a hardware configuration that realizes a controller and a data transfer unit in a storage device according to a second embodiment of the present invention.

Next, embodiments of the present invention will be described in detail with reference to the drawings.

[First Embodiment]
The present embodiment eliminates a bottleneck that occurs when accessing a virtual storage medium that combines storage areas of a plurality of storage media via a network, and improves access performance. About. Hereinafter, the background art of the present embodiment, the problem to be solved by the present embodiment, the outline of the means for solving the problem, the configuration of the present embodiment, the operation of the present embodiment, and the effect of the present embodiment will be sequentially described. I do.

[Background Art of the Embodiment]
The access to the virtual storage medium is performed by, for example, iSCSI (Internet Small Computer Interface) or NVM Express over Fabrics (Non-Volatile Memory Express over Fabrics; registered trademark; hereinafter referred to as NVMe-F). These are protocols that extend the standard of a general storage medium such as SCSI (Small Computer Interface) and NVM Express (Non-Volatile Memory Express; registered trademark; hereinafter referred to as NVMe), and are virtual protocols. The same access means as the storage medium of each standard connected to the computer is provided for the storage medium.

For access to these virtual storage media, RDMA (Remote Direct Memory Access) is used as a means for reducing overhead in data transfer processing via a network. RDMA automatically reduces data transfer overhead by automatically transferring data to a specified memory address by a network card (RDMA Network Interface Card; hereinafter, referred to as RNIC) between computers connected via a network. In the above-mentioned iSCSI RDMA extension, iSER (iSCSI Extensions for RDMA) and NVMe-oF, overhead is reduced by using RDMA for data transfer between a computer that provides a virtual storage medium and a computer that accesses the virtual storage medium. To improve the access performance.

[Problems to be solved by this embodiment]
When using RDMA for data transfer on a virtual storage medium, it is necessary to complete storage of data to be transferred before starting RDMA so that the RNIC can automatically perform data transfer. Therefore, when RDMA is used in data transfer for read access of the virtual storage medium, the RDMA transfer is started after waiting for all data read from the storage medium to be stored in a buffer which is a memory area to be transferred.

On the other hand, the virtual storage medium can be mapped to a plurality of storage medium areas, for example, RAID (Redundant Arrays of Inexpensive Disks). When a host accesses a virtual storage medium mapped to a plurality of storage medium areas, a case occurs in which an access target area is mapped across a plurality of storage media. In this case, access is made to each storage medium that is the mapping destination. In other words, in the case of the read access using RDMA for the data transfer described above, it is necessary to start the RDMA transfer after waiting for the completion of storing the read result from all the storage media to be mapped into the buffer.

In other words, when the destination of the read access from the host to the virtual storage medium is mapped to a plurality of storage media, the read data of the accessed storage medium cannot be transferred first, and the read performance from the host to the virtual storage medium decreases. . In particular, for example, when using a storage medium whose access time greatly changes depending on the internal state, such as an SSD (Solid State Drive) using a NAND Flash as a storage element, as the number of storage media waiting for completion of the read increases, The waiting time until the start of the RDMA transfer increases.

Also, when RDMA is used for write access to a virtual storage medium, writing to the storage medium cannot be started until transfer of all write data to the buffer is completed, and write performance to the virtual storage medium is degraded. Since the data transfer by the RDMA is controlled by the RNIC, it is difficult to confirm the state of the write data during the transfer. That is, when the write destination area is mapped to a plurality of storage media, write access cannot be started first from a storage medium for which data transfer from the host to the buffer has been completed, which causes performance degradation.

Therefore, when the access destination for the virtual storage medium is mapped to a plurality of storage media, the degradation of the access performance due to the waiting process before and after the RDMA transfer is a problem to be solved in the present embodiment.

[Summary of means for solving the problem]
To avoid or alleviate the waiting process before and after the RDMA transfer, which causes a reduction in the access performance of the virtual storage medium, it is necessary to divide the read data and write data transfer processes. By dividing the data transfer process, in the case of read access, the transfer can be started first from the data that has been read from the storage medium. In the case of write access, writing to the storage medium can be started individually from data that has been transferred from the host. On the other hand, the division of the data transfer process leads to an increase in overhead in the access process of the virtual storage medium. Therefore, it is desirable to divide the data transfer process to the minimum number necessary to avoid the waiting process.

In the present embodiment, a means for reducing queuing processing before and after data transfer that occurs when an access destination of a virtual storage medium via a network is mapped to a plurality of storage media and improving access performance of the virtual storage medium is provided. provide. Therefore, in the processing of an access request to a virtual storage medium, access is performed for each storage medium corresponding to the access request destination and RDMA transfer of data related to the access is independently performed. By separating the processing related to the access request for each storage medium, processing related to each storage medium can be performed in parallel without waiting for processing related to other storage mediums when accessing the storage medium and completing data transfer by RDMA. By doing so, the completion time of the access processing is reduced.

First, the host that accesses the virtual storage medium provides the controller that operates on the target node, which is a different node from the host, that provides access to the virtual storage medium, with the type of access request (read or write), The access is requested by sending address information such as LBA (Logical Block Address) indicating the area to be accessed on the virtual storage medium.

When the controller receives the access request from the host, the controller specifies, as the transfer storage area, the identifier of the storage medium corresponding to each part of the area to be accessed and the address indicating the area in the storage medium. Further, the controller obtains an offset from the head of the transfer storage area of each storage medium, which represents a corresponding location in the access target area of the virtual storage medium.

{Circle around (1)} The controller receives, in advance or together with an access request, address information indicating a memory area on the host for storing read data or write data for an area to be accessed on the virtual storage medium. Then, the controller obtains an address in the memory address space of the host corresponding to each transfer storage area as the transfer memory area from the address information. The address of each transfer memory area is obtained as an address in the memory space represented by an offset from the head in the area to be accessed in the virtual storage medium.

(4) Further, the controller secures a transfer buffer, which is a buffer for relaying transfer data, for each set of the transfer storage area and the transfer memory area on the memory of the target node. That is, a transfer buffer is reserved for each storage medium that is the mapping destination of the access target area of the virtual storage medium. The controller stores read data and write data for the storage medium in the transfer buffer, and performs data transfer with the transfer storage area. Further, the controller instructs the RNIC to perform RDMA transfer of read data and write data between the transfer buffer and the transfer memory area of the host.

When the access request from the host is a read, the controller performs a read access to each storage medium corresponding to each part of the area to be accessed, and stores the data in the transfer storage area in the transfer buffer. Then, the controller instructs the RNIC to perform RDMA transfer of read data between the transfer buffer corresponding to the storage medium for which the read access has been completed and the transfer memory area of the host. At this time, the controller starts the RDMA transfer of the read data irrespective of whether the storage of the read data from the transfer storage in another transfer buffer is completed or not completed. When the RDMA transfer from the transfer buffer to the transfer memory area is completed for all the read data, the controller notifies the host of the completion of the read access request.

On the other hand, if the access request from the host is a write, the controller instructs the RNIC to perform RDMA transfer of write data between the transfer memory area of the host corresponding to each part of the area to be accessed and the corresponding transfer buffer. Then, the controller performs write access sequentially from the storage medium corresponding to the transfer buffer for which the RDMA transfer has been completed, and stores the write data in the transfer storage area. At this time, the controller starts write access to the storage medium regardless of whether the RDMA transfer of the write data in the other transfer buffer has been completed or not completed. When all the write data has been stored in the transfer storage area on the storage medium from the transfer buffer, the controller notifies the host of the completion of the write access request.

[Configuration of the present embodiment]
FIG. 1 is a diagram illustrating a configuration of a storage system according to the first embodiment of the present invention. Referring to FIG. 1, the computer system according to the present embodiment includes a host 1, a target node 2, and a network 3. The host 1 is also called a host computer. The target node 2 is also called a target computer or a storage device.

The host 1 has a host memory 10 and a host communication unit 11. In the host 1, an application 12 that issues an access request for the virtual storage medium 23 to the target node 2 operates.

The host memory 10 is a memory connected to the host 1. The host memory 10 is accessed by the application 12. The host memory 10 has an area for storing read data as a result of a read request from the application 12 to the virtual storage medium 23, and write data to be written by the write request. This area is defined as a transfer memory area. The host memory 10 is a storage device such as a DRAM used as a memory of the host 1. Alternatively, the host memory 10 may be, for example, a GPU, an FPGA, or another device that is connected to the host 1 directly or via a network and performs input / output via the memory address space of the host 1.

The host communication unit 11 transmits and receives data between the host 1 and the target node 2 via the network 3. The host communication unit 11 performs RDMA transfer between the host memory 10 and the target memory 20 connected to the target node 2.

The application 12 is software or hardware that operates on the host 1. The application 12 issues an access request for the virtual storage medium 23 to the target node 2.

The target node 2 has a target memory 20, a target communication unit 21, a controller 22, and a storage medium 24.

(4) The target memory 20 is a memory connected to the target node 2. The target memory 20 is used as a buffer for storing read data and write data for the virtual storage medium 23. This buffer is defined as a target buffer.

The target communication unit 21 transmits and receives data between the target node 2 and the host 1 via the network 3. The target communication unit 21 performs RDMA transfer between the target memory 20 and the host memory 10 connected to the host 1.

The controller 22 is software or hardware that operates on the target node 2. The controller 22 performs access processing of the virtual storage medium 23. The controller 22 receives an access request for the virtual storage medium 23 from the application 12. The controller 22 controls access to each area of the storage medium 24 corresponding to the access request destination and transfer of read data and write data between the host memory 10 and the target memory 20 in order to execute the access request. .

The virtual storage medium 23 is a virtual storage medium constructed from the area of the storage medium 24 by the controller 22. The virtual storage medium 23 is recognized and accessed from the application 12 of the host 1 as a storage medium via the network 3. An area on the virtual storage medium 23 accessed from the application 12 is defined as an access target area.

(4) In the storage medium 24, an area of the virtual storage medium 23 is mapped. The area of the storage medium 24 to which the access target area of the virtual storage medium 23 is mapped is defined as a transfer storage area. The storage medium 24 is a device such as a solid state drive (SSD) connected to the target node 2 via a bus or a network. The storage medium 24 is not limited to the configuration of a single device shown in FIG.

The network 3 connects the host 1 and the target node 2 to each other. Network 3 supports RDMA transfer between host 1 and target node 2. Further, the network 3 is not limited to a mode in which only the host 1 and the target node 2 are connected as shown in FIG. For example, a mode in which a node (computer) other than the host 1 and the target node 2 are connected to the network 3 may be employed. Alternatively, the host 1 and the target node 2 may be connected via a switch.

Next, in the computer system according to the present embodiment, the configuration in which the controller 22 receives and executes an access request to the virtual storage medium 23 from the application 12 running on the host 1 will be described with reference to FIGS. 2, 3, 4, 5, and 5. This will be described with reference to FIG.

Referring to FIG. 2, the controller 22 includes an access request receiving unit 221, a mapping table 222, a transfer storage area calculation unit 223, a transfer memory area calculation unit 224, an access processing table 225, and an access processing execution management unit 226. And a data transfer control unit 227. The target memory 20 includes a target buffer 200.

The access request receiving unit 221 receives an access request from the application 12 running on the host 1. The access request includes a type of access request indicating whether the request is read or write. The access request includes a logical address indicating the head of the access target area on the virtual storage medium 23. The access request includes a memory address indicating the head of a storage area (transfer memory area) for read data or write data on the host memory 10 corresponding to the access target area. However, the address information of the transfer memory area corresponding to the access target area does not need to be included in the access request, and may be received in advance before the application 12 makes the access request. The access request receiving unit 221 controls the transfer storage area calculation unit 223, the transfer memory area calculation unit 224, the access processing table 225, and the access processing execution management unit 226 to process the received access request.

The mapping table 222 includes an identifier of the storage medium 24 mapped to each area (hereinafter, referred to as a block) on the virtual storage medium 23, and an address (hereinafter, a physical address and a physical address) in the storage medium 24 specified by the identifier. ) And the logical addresses in the virtual storage medium 23 are stored and managed in advance.

FIG. 3 is a diagram illustrating information included in the mapping table 222. The mapping table 222 records, for each block on the virtual storage medium 23, the identifier of the storage medium 24 of the mapping destination, and the correspondence between the logical address of the block in the virtual storage medium 23 and the physical address of the mapping destination storage medium 24. The mapping entry 2221. A block on the virtual storage medium 23 managed by one mapping entry 2221 of the mapping table 222 is called a logical block, and a block of the storage medium 24 mapped to the logical block is called a physical block. Each mapping entry 2221 includes a logical address 2222, which is the head address of each logical block of the virtual storage medium 23, a storage medium identifier 2223 indicating the identifier of the storage medium 24 to which each logical block is mapped, and the storage medium identifier. It includes a physical address 2224 which is the head address of the physical block of the mapping destination in the storage medium 24 represented by 2223, and a mapping size 2225 which indicates the size of the logical block and the physical block to be mapped.

For example, the mapping entry 2221 in the first row of the mapping table 222 indicates that a logical block of 64 MiB size starting from the logical address 0x00400000 in the virtual storage medium 23 has a physical address 0x0800000 in the storage medium 24 having the storage medium identifier 1. It is mapped to a physical block having a size of 64 MiB. The mapping entry 2221 in the second row of the mapping table 222 indicates that a logical block of 128 MiB size starting from the logical address 0x00800000 in the virtual storage medium 23 has a physical address 0xa00000000 in the storage medium 24 having the storage medium identifier 1 as the head. This indicates that the address is mapped to a physical block having a size of 128 MiB as an address.

As illustrated in the mapping entry 2221 on the first line of the mapping table 222 in FIG. 3, one logical block in the virtual storage medium 23 is mapped to a storage area of at most one storage medium 24. As illustrated in the mapping entries 2221 on the first and second lines of the mapping table 222 in FIG. 3, two logical blocks in the virtual storage medium 23 whose logical addresses are adjacent to each other are mapped to different storage media. ing. In this manner, by mapping each logical block in the virtual storage medium 23 to a storage area of at most one storage medium 24 and mapping two adjacent logical blocks to storage mediums different from each other, the data transfer processing can be performed. The division can be the minimum quantity required to avoid the queuing process.

The format of the mapping table 222 is not limited to the format shown in FIG. The mapping table 222 may have any format as long as the identifier of the storage medium 24 corresponding to each logical block of the virtual storage medium 23 and the correspondence between the logical address and the physical address can be determined. For example, when each mapping entry 2221 corresponds to each logical block obtained by dividing the area of the virtual storage medium 23 from the beginning by a certain size, the logical address 2222 and the mapping size 2225 for each logical block are unnecessary. It is also possible to use RAID for mapping between the virtual storage medium 23 and the storage medium 24. In this case, for example, in the case of RAID 0, the number of storage media 24 constituting the stripe and the mapping corresponding to the virtual storage medium 23 based on the size of the chunk, which is an area mapped continuously to each storage medium 24 The destination can be calculated. At this time, the information of each entry of the mapping table 222 excluding the RAID configuration information is unnecessary.

Based on the mapping table 222, the transfer storage area calculation unit 223 calculates an identifier of a storage medium 24 having a physical block allocated to a logical block including the logical address from a certain logical address of the access target area of the virtual storage medium 23. And a physical address corresponding to the logical address, and a size mapped from the physical address to a continuous area of the storage medium 24 are determined. Further, the transfer storage area calculation unit 223 calculates an offset from the head indicating the position of the area corresponding to the transfer storage area in the area of the access request destination.

FIG. 4 is an explanatory diagram of the operation of the transfer storage area calculation unit 223. Referring to FIG. 4, one logical block 41L of the virtual storage medium 23 is allocated to one physical block 41P of the storage medium 24. The leading logical address of the logical block 41L is Lb, the leading physical address of the physical block 41P is Pb, and the identifier of the storage medium 24 is x. Upon receiving the logical address La on the logical block 41L from the access request receiving unit 221, the transfer storage area calculating unit 223 refers to the mapping entry 2221 corresponding to the logical block 41L in the mapping table 42, and calculates the logical address La. The identifier x of the storage medium 24 assigned to the included logical block is determined. Further, the transfer storage area calculation unit 223 obtains the offset (La−Lb) of the logical address La by subtracting the leading logical address Lb of the logical block 41L from the logical address La. Then, the transfer storage area calculation unit 223 calculates the physical address Pa corresponding to the logical address La by adding the offset to the head physical address Pb of the physical block 41P. That is, Pa = Pb + (La−Lb). For example, if the logical address La is the start address of the access target area specified by the access request of the application 12, the physical address Pa indicates the start of the transfer storage area 43, and the offset (La-Lb) indicates the transfer storage area 43. Indicates the offset of Further, assuming that the mapping size of the logical block 41L and the physical block 41P is MS, the transfer storage area calculation unit 223 determines the size mapped from the physical address Pa to the continuous area of the storage medium 24 with the identifier x by MS- (La- Lb).

When the access target area of the virtual storage medium 23 is mapped to the transfer storage areas of the plurality of storage media 24, the transfer memory area calculation unit 224 determines the address of the transfer memory area on the host memory 10 corresponding to each transfer storage area. calculate.

FIG. 5 is an explanatory diagram of the operation of the transfer memory area calculation unit 224. The access target area 51 on the virtual storage medium 23 and the transfer storage area 52a on the

storage media

24a, 24b, and 24c mapped to the access target area 51 are shown. , 52b, and 52c, and a transfer memory area 53 on the host memory 10 corresponding to the access target area 51. In the example of FIG. 5, the area of the size Sa from the top of the access target area 51 is mapped to the storage medium 24a. The area of the size Sb following the access target area 51 is mapped on the storage medium 24b. The remaining area of the access target area 51 having the size Sc is mapped to the storage medium 24c. In the present embodiment, when an access request to the access target area 51 is made from the application 12, the access target area 51 is divided into three areas having the sizes of Sa, Sb, and Sc, and data transfer is performed in individual division units. Do. Therefore, the transfer memory area 53 is used by being divided into three areas having the sizes of Sa, Sb, and Sc. The transfer memory area calculation unit 224 calculates the head address of each divided area formed when the transfer memory area 53 is divided as described above.

The transfer memory area calculation unit 224 calculates the logical address of the divided area on the access target area 51 of the virtual storage medium 23 as Lx, sets the offset from the top of the access target area 51 to the logical address Lx as OF, Assuming that the address is M, the address Mx of the transfer memory area 53 corresponding to the logical address Lx is calculated as Mx = M + OF.

The access processing table 225 manages information on data transfer processing required for processing each access request from the application 12 to the virtual storage medium 23. The access processing table 225 is updated from the access request receiving unit 221 and is referred to by the access processing execution management unit 226 and the data transfer control unit 227. FIG. 6 is a diagram illustrating information included in the access processing table 225. The access processing table 225 stores an access request entry 2251 corresponding to each access request from the application 12 on a one-to-one basis. Each access request entry 2251 includes an access type 2252 and a transfer pair list. The transfer pair list includes one or more transfer pairs 2253 representing a pair of a transfer memory area and a transfer storage area related to the access request. The access request entry 2251 includes a transfer pair 2253 for each of a plurality of sets of a transfer memory area and a transfer storage area as a result of dividing the access request from the application 12 into a plurality. Further, the transfer pair 2253 includes a memory address 2254 indicating the head of the transfer memory area on the host memory 10, a storage identifier 2255 indicating the storage medium 24 including the transfer storage area, a storage medium address 2256 indicating the head of the transfer storage area, and It is composed of a transfer size 2257 representing the size of transfer data.

The format of the access processing table 225 is not limited to the format shown in FIG. For example, the access type 2252 may be set as a flag related to write access, and an access in which the flag is set may be written and an access in which the flag is not set may be read. Further, the transfer pair 2253 may include a flag indicating that the data transfer has been completed.

The access process execution management unit 226 refers to the access process table 225 to manage the execution status of data transfer required for processing an access request from the application 12 to the virtual storage medium 23. The access processing execution management unit 226 selects the unprocessed transfer pair 2253 from the access processing table 225 and secures the target buffer 200 on the target memory 20. Then, the data transfer control unit 227 starts transfer processing of read data or write data between the transfer memory area and the transfer storage area. When the data transfer process for the transfer pair 2253 ends, the access process execution management unit 226 releases the secured target buffer 200.

The data transfer control unit 227 transfers data between the transfer memory area in the host memory 10 and the secured target buffer 200 for the transfer pair 2253 specified by the access processing execution management unit 226, and stores the data in the target buffer 200. Data transfer to and from the transfer storage area on the medium 24 is performed.

If the access type 2252 for the transfer pair 2253 is read, the data transfer control unit 227 first performs read access to the storage medium 24 represented by the storage identifier 2255, and reads data from the transfer storage area to the target buffer 200. To transfer. When the data transfer to the target buffer 200 is completed, the data transfer control unit 227 next instructs the target communication unit 21 to transfer the read data stored in the target buffer 200 to the transfer memory area of the host memory 10. I do.

On the other hand, when the access type 2252 for the transfer pair 2253 is write, the data transfer control unit 227 first instructs the target communication unit 21 to transfer the write data stored in the transfer memory area of the host memory 10 to the target buffer. Transfer to 200. Then, when the data transfer to the target buffer 200 is completed, the data transfer control unit 227 performs write access to the storage medium 24 represented by the storage identifier 2255, and writes the write data from the target buffer 200 to the transfer storage area. Forward.

(4) The target buffer 200 is secured on the target memory 20 for each transfer pair 2253. The target buffer 200 temporarily holds read data or write data transferred between the transfer memory area and the transfer storage area.

[Operation of this embodiment]
Next, an operation of processing an access request from the application 12 to the virtual storage medium 23 in the embodiment of the present invention will be described with reference to FIGS. 1, 2, 3, 6, 7A, and 7B.

The application 12 transmits an access request to the virtual storage medium 23 of the target node 2 to the host communication unit 11. This access request includes the following parameters.
(A) Type of access request (b) Logical address indicating the access target area on the virtual storage medium 23 (c) Size of the access target area (d) In the memory space of the host 1 corresponding to the access target area Address pointing to area

There are two types of access requests: read and write. The logical address indicating the access target area is the first logical address of the access target area. That is, the access target area has continuous logical addresses, and the leading logical address is specified by the access request. The address corresponding to the access target area and indicating the area in the memory space of the host 1 is the head address of the memory area of the host memory 10 that stores the read data as a result of the read request or the write data written in the write request. .

The host communication unit 11 transmits the access request received from the application 12 to the target communication unit 21 of the target node 2 via the network 3. The target communication unit 21 transmits the received access request to the controller 22.

The access request receiving unit 221 of the controller 22 receives an access request for the virtual storage medium 23 from the application 12 (Step S101).

Next, the access request receiving unit 221 creates an access request entry 2251 corresponding to the access request received in step S101 on the access processing table 225, and stores the access type included in the received access request in the column of the access type 2252. Is registered (step S102). At this time, the access request receiving unit 221 leaves the transfer pair list column of the created access request entry 2251 blank.

Next, the access request receiving unit 221 initializes an offset representing a point of interest in the transfer storage area and the transfer memory area to 0 (step S103). Then, the access request receiving unit 221 pays attention to the initialized offset.

Next, the access request receiving unit 221 adds the value of the focused offset to the first logical address of the access target area in the virtual storage medium 23 included in the access request, and thereby focuses on the access target area on the virtual storage medium 23. Calculate the logical address of the point. (Step S104)

Next, the access request receiving unit 221 notifies the transfer storage area calculation unit 223 of the logical address of the point of interest on the virtual storage medium 23 obtained in step S104. The transfer storage area calculation unit 223 refers to the mapping table 222, and the identifier of the storage medium 24 including the transfer storage area corresponding to the notified logical address of the point of interest on the virtual storage medium 23, and the transfer storage area The physical address of the point of interest in is obtained (step S105).

Next, the access request receiving unit 221 notifies the transfer memory area calculating unit 224 of the start address of the memory area of the host memory 10 and the focused offset. The transfer memory area calculation unit 224 adds the focused offset to the notified start address of the memory area of the host memory 10, and the transfer memory of the host memory 10 corresponding to the logical address of the focused point on the virtual storage medium 23. The address of the point of interest in the area is obtained (step S106).

Next, the transfer storage area calculation unit 223 refers to the mapping table 222 and calculates the size mapped to the continuous area of the storage medium 24 from the notified logical address of the point of interest on the virtual storage medium 23 ( Step S107). Specifically, first, the transfer storage area calculation unit 223 determines the mapping entry 2221 corresponding to the logical block of the virtual storage medium 23 including the logical address of the point of interest. Next, the transfer storage area calculation unit 223 calculates a difference between the logical address of the noted point and the determined logical address 2222 of the corresponding mapping entry 2221. Then, the transfer storage area calculation unit 223 calculates the size mapped to the continuous area by subtracting the calculated difference from the mapping size 2225 of the corresponding mapping entry 2221.

Next, the access request receiving unit 221 performs the following processing (step S108). First, the access request receiving unit 221 calculates, as the untransferred area size, a value obtained by subtracting the focused offset from the size of the access target area specified by the access request. Next, the access request receiving unit 221 compares the size of the untransferred area with the size mapped in the continuous area of the storage medium obtained in step S107. Next, the access request receiving unit 221 determines the minimum value of the two sizes as the size of the transfer data.

Next, the access request receiving unit 221 obtains the identifier of the storage medium 24 obtained in step S105 and the physical address of the point of interest in the transfer storage area, the address of the point of interest in the transfer memory area obtained in step S106, and the address obtained in step S108. The combination of the sizes of the transfer data is registered in the column of the transfer pair 2253 of the access request entry 2251 created in the access processing table 225 in step S102 (step S109). That is, the identifier of the storage medium 24, the physical address of the point of interest in the transfer storage area, the address of the point of interest in the transfer memory area, and the size of the transfer data are respectively the storage identifier 2255, the storage medium address 2256, and the memory address 2254 in the transfer pair 2253. , Transfer size 2257.

Next, the access request receiving unit 221 adds the size of the transfer data obtained in step S108 to the focused offset (step S110). Then, the access request receiving unit 221 shifts the focus to the offset after the addition.

Next, the access request receiving unit 221 compares the value of the focused offset with the size of the access target area specified by the access request. When the value of the focused offset is smaller than the size of the access target area (Yes in Step S111), the access request receiving unit 221 executes the processing from Step S104 again.

On the other hand, when the value of the focused offset is equal to or larger than the size of the area to be accessed (No in step S111), the access request receiving unit 221 instructs the access process execution management unit 226 to execute the access process table to execute the access process 225 access request entry 2251 (step S112).

The access processing execution management unit 226 refers to the access processing table 225 and executes processing for the access request in accordance with the access request entry 2251 specified in step S112 (step S113).

When the access request receiving unit 221 completes the process for the access request according to the access request entry 2251 specified by the access process execution management unit 226, the access request receiving unit 221 deletes the specified access request entry 2251 registered in the access processing table 225. Further, the access request receiving unit 221 notifies the application 12 running on the host 1 that has issued the access request related to the access request entry 2251 of the completion of the access request, and ends the operation (step S114).

動作 By performing the processing from step S101 to S114, the operation of processing the access request from the application 12 to the virtual storage medium 23 in the embodiment of the present invention is completed. The processing order from steps S101 to S114 does not necessarily need to be the order shown in FIGS. 7A and 7B. For example, the processes of steps S104 to S105, step S106, and step S107 can be performed at the same time, or can be performed in a different order. In addition, the address indicating the area in the memory space of the host 1 received from the application 12 in step S101 may be received before the access request.

Next, referring to FIG. 1, FIG. 2, FIG. 3, FIG. 6, FIG. 8A, and FIG. 8B, in the embodiment of the present invention, an access request to the virtual storage medium 23 from the application 12 shown in FIG. The operation of executing the processing for the access request by the access processing execution management unit 226 in step S113 in the operation of processing the step S113 will be described.

The access process execution management unit 226 selects one arbitrary transfer pair 2253 with reference to the access request entry 2251 corresponding to the access request to be processed in the access process table 225 (step S201). Then, the access processing execution management unit 226 pays attention to the selected transfer pair 2253.

Next, the access process execution management unit 226 secures a target buffer 200 having a size equal to or larger than the transfer size 2257 of the focused transfer pair 2253 on the target memory 20 for the focused transfer pair 2253 (step S202). .

Next, the access process execution management unit 226 instructs the data transfer control unit 227 to execute the data transfer process of the focused transfer pair 2253 (step S203).

(4) When the access type 2252 of the access request to be processed is read (Yes in step S204), the data transfer control unit 227 performs the following process (step S205). First, the data transfer control unit 227 refers to the access processing table 225 to obtain the storage identifier 2255, the storage medium address 2256, and the transfer size 2257 of the focused transfer pair 2253. Next, the data transfer control unit 227 requests a read from the storage medium 24 indicated by the storage identifier 2255 of the transfer pair of interest 2253. In this read request, the storage medium address 2256 and the transfer size 2257 of the focused transfer pair 2253 are specified to specify the transfer storage area to be read-accessed, and in step S202, the storage destination of the read data is specified. The head address of the secured target buffer 200 is specified.

The storage medium 24 instructed for read access in step S205 reads data from the transfer storage area having the specified storage medium address 2256 as the head and the specified transfer size 2257, and reads the specified target buffer 200 , And notifies completion to the data transfer control unit 227 (step S206).

After confirming the completion of storing the read data in the target buffer 200, the data transfer control unit 227 instructs the target communication unit 21 to transfer data from the target buffer 200 to the transfer memory area of the host memory 10 (Step S207). . In the transfer instruction, the data transfer control unit 227 specifies the start address of the target buffer 200 and the transfer size 2257 of the transfer pair of interest 2253 as the transfer source to the target communication unit 21. In the transfer instruction, the data transfer control unit 227 specifies the memory address 2254 of the focused transfer pair 2253 as the transfer destination to the target communication unit 21.

The target communication unit 21 performs the following data transfer according to the transfer instruction (step S208). First, the target communication unit 21 reads the read data stored in the target buffer 200 specified as the transfer source, transfers the read data to the host communication unit 11 connected via the network 3, and transfers the read data to the host specified as the transfer destination. Instruct storage in the area specified by the memory address 2254 of the memory 10. The host communication unit 11 receives the instruction from the target communication unit 21 and stores the read data in the transfer memory area of the host memory 10 specified by the memory address 2254. When the storage of the read data in the transfer memory area by the host communication unit 11 is completed, the target communication unit 21 notifies the data transfer control unit 227 of the completion of the data transfer. Then, the process proceeds to step S213.

On the other hand, when the access type 2252 of the access request to be processed is write (No determination in step S204), the data transfer control unit 227 performs the following processing (step S209). First, the data transfer control unit 227 acquires the memory address 2254 and the transfer size 2257 of the transfer pair of interest 2253 with reference to the access processing table 225. Next, the data transfer control unit 227 instructs the target communication unit 21 to transfer data from the host memory 10 to the target buffer 200. In the transfer instruction, the data transfer control unit 227 specifies the memory address 2254 of the transfer pair of interest 2253 and the transfer size 2257 of the transfer pair of interest 2253 as the transfer source to the target communication unit 21. In the transfer instruction, the data transfer control unit 227 specifies the start address of the target buffer 200 as the transfer destination to the target communication unit 21.

(4) The target communication unit 21 performs the following transfer according to the transfer instruction (step S210). First, the target communication unit 21 reads the write data from the transfer memory area specified by the memory address 2254 and the transfer size 2257 of the host memory 10 through the host communication unit 11 and stores the write data in the target buffer 200 via the network 3. Next, the target communication unit 21 notifies the data transfer control unit 227 of the completion of the storage of the write data in the target buffer 200.

After confirming the completion of the storage of the write data in the target buffer 200, the data transfer control unit 227 issues a write request from the target buffer 200 to the storage medium 24 to the storage medium 24 indicated by the storage identifier 2255 of the transfer pair of interest 2253. Perform (step S211). In this write request, the data transfer control unit 227 specifies the head address of the target buffer 200 and the transfer size 2257 of the transfer pair of interest 2253 as the write source for the storage medium 24. In the write request, the data transfer control unit 227 specifies the storage medium address 2256 of the focused transfer pair 2253 as the write destination.

The storage medium 24 requested to write in step S211 performs the specified write access, and stores the write data of the specified target buffer 200 in the transfer storage area starting from the specified storage medium address 2256. Then, the completion is notified to the data transfer control unit 227 (step S212). Thereafter, the process proceeds to step S213.

In step S213, the data transfer control unit 227 notifies the access process execution management unit 226 that the transfer process for the focused transfer pair 2253 has been completed.

(4) The access process execution management unit 226 releases the target buffer 200 reserved for the attention transfer pair 2253 whose transfer process has been completed (step S214). Further, the access processing execution management unit 226 deletes the information on the focused transfer pair 2253 for which the execution of the transfer processing has been completed from the access request entry 2251 corresponding to the access request to be processed in the access processing table 225 (step S215). .

Next, the access process execution management unit 226 checks the number of transfer pairs 2253 included in the access request entry 2251 corresponding to the access request to be processed (step S216). If the number of transfer pairs 2253 confirmed in step S216 is one or more (Yes in step S217), the access processing execution management unit 226 executes the processing from step S201 again. If the number of transfer pairs 2253 confirmed in step S216 is 0 (No in step S217), the access process execution management unit 226 notifies the access request reception unit 221 of completion of the access request. Then, the operation ends (step S218).

By performing the processing from step S201 to S218 described above, the operation of executing the processing for the access request by the access processing execution management unit 226 in step S113 in the embodiment of the present invention is completed. The processing order from steps S201 to S218 does not necessarily need to be the order shown in FIGS. 8A and 8B. For example, the processes of step S214 and step S215 can be performed at the same time, or can be performed in a different order. Further, the processing from step S201 to step S215 may be executed in parallel for a plurality of transfer pairs 2253.

[Effects of the present embodiment]
By applying the present embodiment, when a host accesses a virtual storage medium in which storage areas of a plurality of storage media are combined, when an access destination is mapped to a plurality of storage media, occurrence of a queuing process before and after RDMA transfer occurs And improving the access performance. Further, data transfer between the area of the storage medium and the buffer and data transfer between the buffer and the memory of the host can be parallelized.

[Second embodiment]
FIG. 9 is a block diagram of the storage device 6 according to the second embodiment of the present invention. The storage device 6 includes a controller 61, a memory 62, a virtual storage medium 63, a plurality of storage media 64, and a data transfer unit 65. The storage device 6 is communicably connected to a host computer 8 via a network 7.

The virtual storage medium 63 is a virtual storage medium composed of a plurality of areas of the storage medium 64. The data transfer unit 65 is configured to transfer data between the storage device 6 and the host memory 81 of the host computer 8 via the network 7. The memory 62 is used as a buffer 621 for temporarily storing data read from the virtual storage medium 63 or data to be written to the virtual storage medium 63.

(4) The controller 61 provides access to the virtual storage medium 63. The controller 61 receives an access request to the virtual storage medium 63 from the host computer 8 via the network 7. The controller 61 controls the virtual storage medium 63, the storage medium 64, the memory 62, and the data transfer unit 65 to perform processing according to the received access request. For example, the controller 61 divides the access target area on the virtual storage medium 63 accessed by the access request into one or more partial areas. Further, the controller 61 secures a buffer 621 for each partial area in the memory 62. Further, the controller 61 controls, for each partial area of the virtual storage medium 63, data transfer between the area of the storage medium 64 corresponding to the partial area and the buffer 621 corresponding to the partial area. Further, the controller 61 controls the data transfer between the buffer 621 corresponding to the partial area and the area corresponding to the partial area in the host memory 81 by the data transfer unit 65 for each partial area of the virtual storage medium 63.

The controller 61 and the data transfer unit 65 include, for example, an information processing device 930 having an arithmetic processing unit 910 such as one or more microprocessors, a storage unit 920 such as a memory and a hard disk, and a program 940, as shown in FIG. It can be realized by. The program 940 is read into a memory from an external computer-readable recording medium when the information processing device 930 is started up, and controls the operation of the arithmetic processing unit 910. And a data transfer unit 65.

The storage device 6 configured as described above operates as follows.

When the controller 61 of the storage device 6 receives an access request for the virtual storage medium 63 from the host computer 8 via the network 7, the controller 61 divides the access target area on the virtual storage medium 63 into partial areas. Preferably, the controller 61 divides the access target area on the virtual storage medium 63 into a partial area composed of at most one area of the storage medium 64. Preferably, the controller 61 divides the access target area on the virtual storage medium 63 into partial areas so that the number of partial areas is minimized.

Next, the controller 61 secures a buffer 621 for each partial area in the memory 62. Next, for each partial area, the controller 61 performs data transfer between the area of the storage medium 64 corresponding to the partial area and the buffer 621 corresponding to the partial area. Further, the controller 61 transfers data between the buffer 621 corresponding to the partial area and the area corresponding to the partial area in the host memory 81 by the data transfer unit 65.

For example, when the type of the access request is a read request, the controller 61 performs, for each partial area, data transfer from the area of the storage medium 64 corresponding to the partial area to the buffer 621 corresponding to the partial area. Is completed, the data transfer from the buffer 621 to the area corresponding to the partial area in the host memory 81 is started. Further, for example, when the type of the access request is a write request, the controller 61 performs data transfer for each partial area from the area corresponding to the partial area in the host memory 81 to the buffer 621 corresponding to the partial area. After the data transfer is completed, the data transfer from the buffer 621 to the area of the storage medium 64 corresponding to the partial area is started.

As described above, according to the storage device 6 according to the present embodiment, the access performance from the host computer 8 to the virtual storage medium 63 is improved. The reason is that the storage device 6 divides the access target area on the virtual storage medium 63 accessed by the access request into partial areas, secures a buffer 621 for each partial area, and To transfer data between the area of the storage medium 64 and the buffer 621 corresponding to the partial area, and between the buffer 621 corresponding to the partial area and the area corresponding to the partial area in the host memory 81. In the case of a read request, data transfer from the partial area where the read data is stored in the buffer 621 to the host memory can be started, and in the case of a write request, writing from the partial area where the write data is stored in the buffer 621 to the recording medium This is because it is possible to start.

Although the present invention has been described with reference to the above embodiments, the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

The present invention enjoys the benefit of the priority claim based on the patent application of Japanese Patent Application No. 2018-119240 filed in Japan on June 22, 2018, and is described in the patent application. The contents are all included in this specification.

According to the present invention, a virtual storage medium configured from a storage medium area connected to a computer can be used for a service or the like provided to another computer via a network.

Some or all of the above embodiments may be described as in the following supplementary notes, but are not limited thereto.
[Appendix 1]
A controller that provides access to a virtual storage medium configured from a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network;
Data transfer means between the node connected to the controller and the memory of the host,
The controller divides an access target area on the virtual storage medium accessed by the access request into partial areas, reserves a buffer for each of the partial areas, and, for each of the partial areas, Data transfer between an area of a storage medium and the buffer corresponding to the partial area; and a transfer of the buffer corresponding to the partial area by the data transfer unit and an area corresponding to the partial area in the memory of the host. A storage device configured to control data transfer between the storage device.
[Appendix 2]
The storage device according to claim 1, wherein the controller divides the access target area into the partial areas each including at most one area of the storage medium in the division.
[Appendix 3]
3. The storage apparatus according to claim 2, wherein the controller divides the partial area so as to minimize the number of the partial areas.
[Appendix 4]
The controller is
From the host, the type of the access request, first address information indicating the access target area, and second address information indicating a transfer memory area on the memory for transmitting and receiving data to and from the access target area 3. The storage device according to claim 3, further comprising an access request receiving unit that acquires the request.
[Appendix 5]
The controller is
The access target area is divided into the partial areas each including at most one area of the storage medium, the size of each of the partial areas, the identifier of the storage medium corresponding to each of the partial areas, and the storage area. 5. The storage apparatus according to claim 4, further comprising a transfer storage area calculation unit that obtains third address information indicating a transfer storage area that is a medium area.
[Appendix 6]
The controller is
6. The transfer memory area according to claim 5, further comprising: a transfer memory area calculating unit that divides the transfer memory area into partial transfer memory areas corresponding to the respective partial areas and obtains fourth address information indicating the respective partial transfer memory areas. Storage device.
[Appendix 7]
The controller is
7. The storage device according to claim 6, further comprising an access processing execution management unit that secures the buffer in the node for each of the partial areas.
[Appendix 8]
The controller is
8. The storage apparatus according to claim 7, further comprising a data transfer control unit that controls data transfer between the transfer storage area corresponding to the partial area and the buffer corresponding to the partial area.
[Appendix 9]
9. The storage device according to claim 8, wherein the data transfer control means controls data transfer by the data transfer means between the buffer corresponding to the partial area and the partial transfer memory area corresponding to the partial area.
[Appendix 10]
The controller is
The storage device according to claim 9, further comprising a mapping table that stores information that associates the area of the virtual storage medium with the area of the plurality of storage media.
[Supplementary Note 11]
The controller is
For each of the partial areas, the fourth address information indicating the partial transfer memory area, the identifier of the storage medium, the third address information indicating the transfer storage area, and the size corresponding to the partial area 11. The storage device according to appendix 10, further comprising an access processing table for storing information of the transfer pair including the information.
[Supplementary Note 12]
The controller, when the type of the access request is a read request, for each of the partial areas, performs data transfer from the area of the storage medium corresponding to the partial area to the buffer corresponding to the partial area, After the data transfer from the area of the storage medium corresponding to the partial area to the buffer corresponding to the partial area is completed, from the buffer corresponding to the partial area to the area corresponding to the partial area in the memory of the host 12. The storage device according to any one of supplementary notes 1 to 11, configured to start the data transfer of (1).
[Appendix 13]
When the type of the access request is a write request, the controller performs, for each of the partial areas, data transfer from the area corresponding to the partial area in the memory of the host to the buffer corresponding to the partial area. After completion of the data transfer from the area corresponding to the partial area in the memory of the host to the buffer corresponding to the partial area, the data of the storage medium corresponding to the partial area is read from the buffer corresponding to the partial area. 13. The storage device according to any one of Supplementary Notes 1 to 12, configured to start data transfer to the area.
[Appendix 14]
The storage device according to any one of supplementary notes 1 to 13, wherein the data transfer means is configured to perform RDMA (Remote Direct Memory Access).
[Appendix 15]
Between a node connected to a controller that provides access to a virtual storage medium composed of a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network, and the memory of the host; Data transfer method,
Dividing an access target area on the virtual storage medium accessed by the access request into partial areas,
Securing a buffer for each partial area,
For each of the partial areas, data transfer between the area of the storage medium corresponding to the partial area and the buffer corresponding to the partial area, and the buffer corresponding to the partial area and the buffer in the memory of the host. A data transfer method for transferring data to and from an area corresponding to a partial area.
[Appendix 16]
Computer
A controller that provides access to a virtual storage medium configured from a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network;
Causing the controller to function as data transfer means between the connected node and the memory of the host,
The controller divides an access target area on the virtual storage medium accessed by the access request into partial areas, reserves a buffer for each of the partial areas, and, for each of the partial areas, Data transfer between an area of a storage medium and the buffer corresponding to the partial area; and a transfer of the buffer corresponding to the partial area by the data transfer unit and an area corresponding to the partial area in the memory of the host. A computer-readable recording medium recording a program configured to control data transfer between the computer and the computer.

1 Host 10 Host Memory 11 Host Communication Unit 12 Application 2 Target Node 20 Target Memory 200 Target Buffer 21 Target Communication Unit 22 Controller 221 Access Request Receiving Unit 222 Mapping Table 2221 Mapping Entry 2222 Logical Address 2223 Storage Medium Identifier 2224 Physical Address 2225 Mapping Size 223 Transfer storage area calculation section 224 Transfer memory area calculation section 225 Access processing table 2251 Access request entry 2252 Access type 2253 Transfer pair 2254 Memory address 2255 Storage identifier 2256 Storage medium address 2257 Transfer size 226 Access processing execution management section 227 Data transfer control section 23 virtual storage medium 24 storage medium 3 network 4 1 storage area 42 mapping table 421 logical address 422 storage medium identifier 423 physical address 43 transfer storage area 51 access target area 52 transfer storage area 53 transfer memory area 6 storage device 61 controller 62 memory 621 buffer 63 virtual storage medium 64 storage medium 7 network 8 host computer 81 host memory 910 arithmetic processing unit 920 storage unit 930 information processing device 940 program

Claims

A controller that provides access to a virtual storage medium configured from a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network;
Data transfer means between the node connected to the controller and the memory of the host,
The controller divides an access target area on the virtual storage medium accessed by the access request into partial areas, reserves a buffer for each of the partial areas, and, for each of the partial areas, Data transfer between an area of a storage medium and the buffer corresponding to the partial area; and a transfer of the buffer corresponding to the partial area by the data transfer unit and an area corresponding to the partial area in the memory of the host. A storage device configured to control data transfer between the storage device.
2. The storage device according to claim 1, wherein the controller divides the access target area into the partial areas each including at most one area of the storage medium.
3. The storage device according to claim 2, wherein the controller performs the division such that the number of the partial areas is minimized in the division.
The controller is
From the host, the type of the access request, first address information indicating the access target area, and second address information indicating a transfer memory area on the memory for transmitting and receiving data to and from the access target area 4. The storage device according to claim 3, further comprising an access request receiving unit that acquires the request.
The controller is
The access target area is divided into the partial areas each including at most one area of the storage medium, the size of each of the partial areas, the identifier of the storage medium corresponding to each of the partial areas, and the storage area. 5. The storage apparatus according to claim 4, further comprising a transfer storage area calculating unit that obtains third address information indicating a transfer storage area that is a medium area.
The controller is
6. The transfer memory area calculating unit according to claim 5, further comprising: a transfer memory area calculating unit that divides the transfer memory area into partial transfer memory areas corresponding to the respective partial areas, and obtains fourth address information indicating the respective partial transfer memory areas. Storage device.
The controller is
7. The storage apparatus according to claim 6, further comprising an access processing execution management unit that secures the buffer in the node for each of the partial areas.
The controller is
The storage apparatus according to claim 7, further comprising a data transfer control unit that controls data transfer between the transfer storage area corresponding to the partial area and the buffer corresponding to the partial area.
9. The storage device according to claim 8, wherein said data transfer control means controls data transfer by said data transfer means between said buffer corresponding to said partial area and said partial transfer memory area corresponding to said partial area.
The controller is
10. The storage device according to claim 9, further comprising a mapping table that stores information that associates an area of the virtual storage medium with an area of the plurality of storage media.
The controller is
For each of the partial areas, the fourth address information indicating the partial transfer memory area, the identifier of the storage medium, the third address information indicating the transfer storage area, and the size corresponding to the partial area 11. The storage device according to claim 10, further comprising an access processing table for storing information on a transfer pair including the information.
The controller, when the type of the access request is a read request, for each of the partial areas, performs data transfer from the area of the storage medium corresponding to the partial area to the buffer corresponding to the partial area, After the data transfer from the area of the storage medium corresponding to the partial area to the buffer corresponding to the partial area is completed, from the buffer corresponding to the partial area to the area corresponding to the partial area in the memory of the host The storage device according to any one of claims 1 to 11, wherein the storage device is configured to start the data transfer.
When the type of the access request is a write request, the controller performs, for each of the partial areas, data transfer from the area corresponding to the partial area in the memory of the host to the buffer corresponding to the partial area. After completion of the data transfer from the area corresponding to the partial area in the memory of the host to the buffer corresponding to the partial area, the data of the storage medium corresponding to the partial area is read from the buffer corresponding to the partial area. 13. The storage device according to claim 1, wherein the storage device is configured to start data transfer to an area.
14. The storage device according to claim 1, wherein the data transfer unit is configured to perform RDMA (Remote Direct Memory Access).
Between a node connected to a controller that provides access to a virtual storage medium composed of a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network, and the memory of the host; Data transfer method,
Dividing an access target area on the virtual storage medium accessed by the access request into partial areas,
Securing a buffer for each partial area,
For each of the partial areas, data transfer between the area of the storage medium corresponding to the partial area and the buffer corresponding to the partial area, and the buffer corresponding to the partial area and the buffer in the memory of the host. A data transfer method for transferring data to and from an area corresponding to a partial area.
Computer
A controller that provides access to a virtual storage medium configured from a plurality of storage medium areas and receives an access request to the virtual storage medium from a host via a network;
Causing the controller to function as data transfer means between the connected node and the memory of the host,
The controller divides an access target area on the virtual storage medium accessed by the access request into partial areas, reserves a buffer for each of the partial areas, and, for each of the partial areas, Data transfer between an area of a storage medium and the buffer corresponding to the partial area; and a transfer of the buffer corresponding to the partial area by the data transfer unit and an area corresponding to the partial area in the memory of the host. A computer-readable recording medium recording a program configured to control data transfer between the computer and the computer.