JP6905611B2 - Storage system and its control method - Google Patents

Description

The present invention relates to a storage system and a control method thereof, and is suitable for application to, for example, an information processing system including a plurality of storage nodes on which one or a plurality of SDSs (Software Defined Storage) are implemented. In the following, SDS refers to a storage device constructed by mounting software having a storage function on a general-purpose server device.

Conventionally, in an information processing system, a technique for accommodating capacity among a plurality of storage subsystems has been proposed.

For example, a first storage system that includes a first controller that receives I / O commands from the host and provides the host with a first volume, and a second volume that receives I / O commands from the host and provides the host with a second volume. A second storage system including a second controller is provided, the first storage area of the first volume is allocated from the first pool by the first controller, and the second of the first volumes. Storage area is allocated from the second pool by the first controller, the third storage area of the second volume is allocated from the first pool by the second controller, and the fourth of the second volumes. There is a technique for allocating the storage area of the above from the second pool by the second controller. For example, there is a technique described in Patent Document 1 below.

Japanese Unexamined Patent Publication No. 2012-043407

Considering the improvement of availability and reliability of the distributed storage system as a whole, it is desirable to make the controller redundant. It is also desirable from the viewpoint of availability and reliability that the data written from the host to the storage system is also redundantly stored in the system.

However, when making the controller and data redundant, it is necessary to carefully consider where to place the redundant data from the viewpoint of response performance and data protection of the distributed storage system consisting of multiple nodes as a whole. There is.

The present invention has been made in consideration of the above points, and an object of the present invention is to propose a storage system capable of protecting data while preventing deterioration of the response performance of the entire system and a control method thereof.

In order to solve such a problem, in the present invention, in a storage system having a plurality of storage nodes, each storage node manages the capacity provided by each storage device of the plurality of storage nodes in the cluster. The storage system includes a unit and a storage control unit that receives an I / O request, generates a command corresponding to the received I / O request, and transmits the command to the capacity control unit. The storage system is a different storage node. A plurality of the storage control units arranged in the storage control unit form a storage control unit group, and the first storage control unit of the storage control unit group is set to the first state of processing the I / O request. At the same time, a plurality of data stored in the storage device are stored in different storage nodes by setting the second storage control unit of the storage control unit group to a second state in which the I / O request is not processed. Data is made redundant by combining the above data, and the capacity control unit is arranged with the storage node in which the storage control unit in the first state is arranged and the storage control unit in the second state. When allocating the physical storage area of a plurality of storage nodes including at least the storage node to the storage control unit group and allocating the physical storage area to the storage control unit group, the storage control unit in the first state is used. The physical storage area of the storage node in which the storage node and the storage control unit in the second state are arranged is preferentially allocated to the storage control unit group.

Further, in the present invention, in the control method of the storage system having a plurality of storage nodes, each of the storage nodes has a capacity control unit that manages the capacity provided by each storage device of the plurality of storage nodes in the cluster. , A storage control unit that receives an I / O request, generates a command corresponding to the received I / O request, and transmits the command to the capacity control unit. In the storage system, the storage nodes are different from each other. The plurality of arranged storage control units form a storage control unit group, and the first storage control unit of the storage control unit group is set to the first state of processing the I / O request. , The second storage control unit of the storage control unit group is set to a second state in which the I / O request is not processed, and the data stored in the storage device is stored in a plurality of different storage nodes. The data is made redundant by combining data, and the capacity control unit is such that the storage node in which the storage control unit in the first state is arranged and the storage control unit in the second state are arranged. The first step of allocating the physical storage area of a plurality of storage nodes including at least a storage node to the storage control unit group and allocating the physical storage area to the storage control unit group, and the capacity control unit of the I In response to the command transmitted from the storage control unit set to the first state based on the / O request, data is stored in each physical storage area assigned to the storage control unit group to which the storage control unit belongs. Each includes a second step of writing or reading data from the physical storage area of one of the physical storage areas, and in the first step, the storage control unit in the first state. The physical storage area of the storage node in which the data is arranged and the storage control unit in the second state is arranged is preferentially allocated to the storage control unit group.

According to the storage system and the capacity allocation method of the present invention, at least two physical storage areas are allocated to the storage control unit group, and the data is protected because the data is redundantly stored in these physical storage areas. NS. Further, for the storage control unit group, the physical storage area of the storage node in which the storage control unit set in the first state is arranged and the storage in which the storage control unit set in the second state is arranged. Since the physical storage areas of the nodes are preferentially allocated, data can be quickly read / written to these physical storage areas.

According to the present invention, it is possible to realize a storage system capable of protecting data while preventing deterioration of the response performance of the entire system and a control method thereof. Issues, configurations and effects other than those described above will be clarified by the following description of the embodiments for carrying out the invention.

It is a block diagram which shows the whole structure of the information processing system by 1st Embodiment. It is a block diagram which shows the hardware configuration of a storage node. It is a block diagram which shows the logical structure of a storage node. It is a block diagram which provides the explanation of the flow of write processing in this information processing system. It is a block diagram which shows the detailed structure of the capacity control part. It is a block diagram which provides the explanation of the memory composition of a storage node. It is a figure which shows the structure of the storage control part pair management table. It is a figure which shows the structure of the physical chunk management table. It is a figure which shows the structure of the logical chunk management table. It is a figure which shows the structure of the free physical chunk number management table. It is a flowchart which shows the processing procedure of a write process. It is a flowchart which shows the processing procedure of a read process. It is a flowchart which shows the processing procedure of capacity allocation processing. It is a flowchart which shows the processing procedure of the physical chunk selection process. It is a flowchart which shows the process procedure of the failover process. It is a flowchart which shows the processing procedure of the re-redundancy processing. It is a block diagram which shows the whole structure of the information processing system by 2nd Embodiment. It is a block diagram which shows the configuration example of the storage control part pair considering a fault set. It is a block diagram which provides the explanation of the correspondence between the logical chunk and the physical chunk in the 2nd Embodiment. It is a figure which shows the structure of a node management table. It is a flowchart which shows the processing procedure of the physical chunk selection process by 2nd Embodiment. It is a block diagram which shows the whole structure of the information processing system by 3rd Embodiment. It is a block diagram which provides the explanation of the hierarchical control function in the information processing system of 3rd Embodiment. It is a figure which shows the structure of the physical chunk management table by 3rd Embodiment. It is a figure which shows the structure of the logical chunk management table by 3rd Embodiment. It is a figure which shows the structure of the free physical chunk number management table by 3rd Embodiment. It is a block diagram which shows the detailed structure of the capacity control part by 3rd Embodiment. It is a flowchart which shows the processing procedure of capacity allocation processing by 3rd Embodiment. It is a flowchart which shows the processing procedure of the physical chunk selection process by 3rd Embodiment. It is a block diagram which shows the whole structure of the information processing system by another embodiment. It is a flowchart which shows the processing procedure of capacity allocation processing by another Embodiment. It is a flowchart which shows the processing procedure of the physical chunk selection process by another embodiment. It is a flowchart which shows the processing procedure of capacity allocation processing by another Embodiment. It is a flowchart which shows the processing procedure of capacity allocation processing by another Embodiment. It is a flowchart which shows the processing procedure of the physical chunk selection process by another embodiment. It is a figure which shows the structure of a virtual volume management table.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. The following description and drawings are examples for explaining the present invention, and are appropriately omitted or simplified for the purpose of clarifying the description. Also, not all combinations of features described in the embodiments are essential to the means of solving the invention. The present invention is not limited to embodiments, and any application that fits the ideas of the present invention is included in the technical scope of the present invention. Those skilled in the art can make various additions and changes within the scope of the present invention. The present invention can also be implemented in various other forms. Unless otherwise specified, each component may be plural or singular.

In the following explanation, various information may be described by expressions such as "table", "table", "list", and "queue", but various information may be expressed by data structures other than these. good. The "XX table", "XX list", etc. may be referred to as "XX information" to indicate that they do not depend on the data structure. When explaining the contents of each information, expressions such as "identification information", "identifier", "name", "ID", and "number" are used, but these can be replaced with each other.

Further, in the following description, the reference code or the common number in the reference code is used when the same type of element is not distinguished, and the reference code of the element is used when the same type of element is described separately. Alternatively, the ID assigned to the element may be used instead of the reference code.

Further, in the following description, a process performed by executing a program may be described, but the program is executed by at least one or more processors (for example, a CPU) to appropriately store a predetermined process. Since it is performed while using resources (for example, memory) and / or interface device (for example, communication port), the main body of processing may be a processor. Similarly, the subject of processing for executing a program may be a controller having a processor, a device, a system, a computer, a node, a storage system, a storage device, a server, a management computer, a client, or a host. The subject of processing performed by executing a program (for example, a processor) may include a hardware circuit that performs part or all of the processing. For example, the subject of processing performed by executing a program may include a hardware circuit that executes encryption and decryption, or compression and decompression. The processor operates as a functional unit that realizes a predetermined function by operating according to a program. A device and system including a processor is a device and system including these functional parts.

The program may be installed from the program source into a device such as a calculator. The program source may be, for example, a program distribution server or a storage medium readable by a computer. When the program source is a program distribution server, the program distribution server includes a processor (for example, a CPU) and a storage resource, and the storage resource may further store the distribution program and the program to be distributed. Then, when the processor of the program distribution server executes the distribution program, the processor of the program distribution server may distribute the program to be distributed to other computers. Further, in the following description, two or more programs may be realized as one program, or one program may be realized as two or more programs.

(1) First Embodiment (1-1) Configuration of Information Processing System According to the Present Embodiment FIG. 1 is a diagram showing a configuration of an information processing system 1 according to the present embodiment. The information processing system 1 includes a plurality of host devices 3 and a plurality of host devices 3 connected to each other via a network 2 composed of, for example, Fiber Channel, Ethernet (registered trademark), or LAN (Local Area Network). The storage node 4 and the management node 5 of the above are provided.

The host device 3 makes a read request or a write request to the storage node 4 in response to a user operation or a request from an implemented application program (hereinafter, these are collectively referred to as an I / O (Input / Output) request as appropriate). ) Is a general-purpose computer device that transmits. The host device 3 may be a virtual computer device such as a virtual machine.

The storage node 4 is a physical server device that provides a storage area for reading and writing data to the host device 3, and as shown in FIG. 2, CPUs (Central Processing) connected to each other via an internal network 10 It is configured to include a Unit) 11, a memory 12, a plurality of storage devices 13, and a communication device 14. Each storage node 4 includes one or more CPU 11, a memory 12, a storage device 13, and a communication device 14.

The CPU 11 is a processor that controls the operation of the entire storage node 4. The memory 12 is composed of a volatile semiconductor memory such as a SRAM (Static RAM (Random Access Memory)) or a DRAM (Dynamic RAM), and is used for temporarily holding various programs and necessary data. When at least one or more CPUs 11 execute the program stored in the memory 12, various processes as a whole of the storage node 4 as described later are executed.

The storage device 13 is a large-capacity non-volatile one or a plurality of types such as an SSD (Solid State Drive), a SAS (Serial Attached SCSI (Small Computer System Interface)) hard disk drive, or a SATA (Serial ATA (Advanced Technology Attachment)) hard disk drive. It is composed of a sex storage device and provides a physical storage area for reading / writing data in response to an I / O request from the host device 3 (FIG. 1).

The communication device 14 is an interface for the storage node 4 to communicate with the host device 3 and other storage nodes 4 or the management node 5 via the network 2 (FIG. 1). For example, a NIC (Network Interface Card) or It consists of FC (Fibre Channel) cards and the like. The communication device 14 controls the protocol at the time of communication with the host device 3, another storage node 4, or the management node 5.

The management node 5 is a computer device used by the system administrator to manage the entire information processing system 1. The management node 5 manages a plurality of storage nodes 4 as a group called a cluster 6. Although FIG. 1 shows an example in which only one cluster 6 is provided, a plurality of clusters 6 may be provided in the information processing system 1. Cluster 6 may be referred to as a distributed storage system.

FIG. 3 shows the logical configuration of the storage node 4 according to the present embodiment. As shown in FIG. 3, each storage node 4 includes a front-end driver 20, a back-end driver 21, one or more storage control units 22, and a capacity control unit 23.

The front-end driver 20 controls the communication device 14 (FIG. 2), and provides the storage control unit 22 with an abstract interface during communication with the host device 3 and other storage nodes 4 or management nodes 5 (the CPU 11 (FIG. 2). It is software having a function provided in FIG. 2). Further, the back-end driver 21 is software having a function of controlling each storage device 13 (FIG. 2) in the own storage node 4 and providing the CPU 11 with an abstract interface at the time of communication with these storage devices 13.

The storage control unit 22 is software that functions as a controller for SDS (Software Defined Storage). The storage control unit 22 may be called a storage control software or a storage control program. The storage control unit 22 receives an I / O request from the host device 3 and issues an I / O command in response to the I / O request to the capacity control unit 23.

In the case of the present embodiment, each storage control unit 22 mounted on the storage node 4 is managed as a pair forming a redundant configuration together with another storage control unit 22 arranged on another storage node 4. In the following, this pair will be referred to as a storage control unit pair 25.

Note that FIG. 3 shows a case where one storage control unit pair 25 is configured by the two storage control units 22, and also in the following, the storage control unit pair 25 is configured by the two storage control units 22. However, one redundant configuration may be configured by three or more storage control units 22.

In the storage control unit pair 25, one storage control unit 22 is set to a state in which it can receive an I / O request from the host device 3 (a state of the active system, hereinafter referred to as an active mode). The other storage control unit 22 is set to a state in which it does not accept a read request or a write request from the host device 3 (a state of a standby system, hereinafter referred to as a passive mode).

Then, in the storage control unit pair 25, a failure has occurred in the storage control unit 22 (hereinafter, this is referred to as an active storage control unit 22) set in the active mode and the storage node 4 in which the active storage control unit 22 is arranged. In some cases, the state of the storage control unit 22 (hereinafter, this is referred to as a passive storage control unit 22) that has been set to the passive mode is switched to the active mode. As a result, when the active storage control unit 22 cannot operate, the passive storage control unit 22 can take over the I / O processing executed by the active storage control unit 22.

The capacity control unit 23 allocates a physical storage area provided by the storage device 13 in the own storage node 4 or another storage node 4 to each storage control unit pair 25, and is given by the storage control unit 22. The software has a function of reading / writing the specified data to the corresponding storage device 13 according to the above-mentioned I / O command. The capacity control unit 23 may be called a capacity control software or a capacity control program.

In this case, when the capacity control unit 23 allocates the physical storage area provided by the storage device 13 in the other storage node 4 to the storage control unit pair 25, the capacity control unit 23 is mounted on the other storage node 4. I / O given from the active storage control unit 22 of the storage control unit pair 25 by exchanging data with the capacity control unit 23 via the network 2 in cooperation with the capacity control unit 23. Read / write the data to the storage area according to the command.

In the information processing system 1 having the above configuration, as shown in FIG. 4, the capacity control unit 23 has a physical storage area provided by each storage device 13 in each storage node 4 having a predetermined size. It is managed by dividing it into physical storage areas (hereinafter referred to as physical chunks) PCs.

Further, the capacity control unit 23 associates a dedicated pool PL with each storage control unit pair 25 (FIG. 3), and associates these pool PLs with a logical storage area having the same size as the physical chunk PC (hereinafter, this). Is assigned as appropriate to the logical chunk LC, and one or more physical chunk PCs are associated with the logical chunk LC.

Further, one or a plurality of virtual logical volumes (hereinafter, referred to as virtual volumes) VVOLs are defined on the pool PL of each storage control unit pair 25, and these virtual volume VVOLs are provided to the host device 3. ..

When the host device 3 writes data to the virtual volume VVOL, the host device 3 has an identifier (LUN: Logical Unit Number) of the virtual volume (hereinafter, referred to as a write target virtual volume) VVOL to which the data is written, and its logical unit number. A write request specifying a write destination area (hereinafter, this is referred to as a write destination area) WAR of the data in the write target virtual volume VVOL is transmitted to any storage node 4 in the corresponding cluster 6.

The front-end driver 20 of the storage node 4 that has received this write request is the active of the storage control unit pair 25 (FIG. 3) associated with the write target virtual volume VVOL specified in the received write request via the pool PL. The write request to the front-end driver 20 of each storage node 4 in which the storage control unit 22 (FIG. 3) or the passive storage control unit 22 is arranged, and the write target transmitted from the host device 3 together with the write request. Data (hereinafter referred to as write data) is transferred.

Further, the front-end driver 20 of the storage node 4 that has received the write request and the write data associates the write request and the write data with the write target virtual volume VVOL specified in the write request via the pool PL. It is delivered to the storage control unit 22 of the storage control unit pair 25.

Then, the active storage control unit 22 of the storage control units 22 to which the write request and the write data are delivered associates the write destination area WAR in the write target virtual volume VVOL with the write target virtual volume VVOL. A storage area (hereinafter, referred to as a logical area) is allocated from the logical chunk LC constituting the pool PL as needed.

Further, the active storage control unit 22 sets the address of the write destination area WAR in the write target virtual volume VVOL specified in the write request, the chunk number of the logical chunk LC in which the logical area is assigned to the write destination area WAR, and the chunk number of the logical chunk LC. An I / O command converted to the offset position of the logical area is generated, and the generated I / O command is transmitted to the capacity control unit 23 in the own storage node 4 together with the write data.

Then, when the capacity control unit 23 receives the I / O command and the write data, the capacity control unit 23 receives the I / O command in each storage device 13 that provides each physical chunk PC associated with the logical chunk LC specified by the I / O command. Data is stored in the storage area at the offset position.

In this way, in the information processing system 1, the data from the host device 3 is redundantly stored in a plurality of physical chunk PCs associated with the corresponding logical chunk LCs. Therefore, the number of physical chunk PCs assigned to the logical chunk LC is determined by the setting contents of the redundancy method in the information processing system 1.

For example, in the case of setting to duplicate and store data, two physical chunk PCs are associated with one logical chunk LC, and in the case of setting to multiplex and store data more than triplet, or Erasure. -When a setting such as Coding is made to create and store redundant data from data, three or more required number of physical chunk PCs are associated with one logical chunk LC.

When a plurality of physical chunk PCs are associated with one logical chunk LC and data is multiplexed and stored in the plurality of physical chunk PCs, one physical chunk PC from the plurality of physical chunk PCs is the "master". Is set, and all the remaining physical chunk PCs are set to "mirror". Then, as described later, the data read from the physical chunk PC is performed from the physical chunk PC set as the "master". Further, in the case of EC (Erasure Coding), a plurality of physical chunk PCs are associated with one logical chunk LC, and master data and redundant data are stored in these plurality of physical chunk PCs in a predetermined pattern.

On the other hand, when the host device 3 reads data from the virtual volume VVOL, it stores the LUN of the virtual volume (hereinafter, this is referred to as a read target virtual volume) VVOL and the read destination of the data in the read target virtual volume VVOL. A read request for which an area (hereinafter, this is referred to as a read destination area) is specified is transmitted to any storage node 4 in the cluster 6 including the read target virtual volume VVOL.

The front-end driver 20 of the storage node 4 that has received this read request is the active storage control unit 22 of the storage control unit pair 25 that is associated with the read target virtual volume VVOL specified in the received read request via the pool PL. Alternatively, the read request is transferred to each storage node 4 in which the passive storage control unit 22 is arranged.

Further, the front-end driver 20 of the storage node 4 that has received the read request associates the read request with the read target virtual volume VVOL specified in the read request via the pool PL, and the storage control unit pair 25. It is handed over to the storage control unit 22 of.

Thus, the active storage control unit 22 of the storage control units 22 to which the read request is delivered assigns the address of the read destination area in the read target virtual volume VVOL to the logical chunk in which the logical area is assigned to the read destination area. The I / O command converted into the chunk number of the LC and the offset position of the logical area is generated, and the generated I / O command is transmitted to the capacity control unit 23 in the own storage node 4.

When the capacity control unit 23 receives this I / O command, the capacity control unit 23 is in the physical chunk PC set as the “master” among the physical chunk PCs associated with the logical chunk LC specified by the I / O command. Data is read from the storage area at the offset position specified by the I / O command, and the read data is transferred as read data to the active storage control unit 22 of the source of the I / O command. Thus, the read data is then transferred by the active storage control unit 22 to the host device 3 that is the source of the read request via the network 2.

(1-2) Allocation of physical chunks to logical chunks By the way, as described above, a plurality of physical chunk PCs are associated with one logical chunk LC, and data is stored in each of these physical chunk PCs to make data redundant. When adopting the redundancy method, it is desirable to select a plurality of physical chunk PCs associated with one logical chunk LC from the physical chunk PCs provided by different storage nodes 4 from the viewpoint of data protection. This is because, for example, when a plurality of physical chunk PCs in the same storage node 4 are associated with one logical chunk LC, data lost occurs when the storage node 4 cannot read data due to a failure or the like. This is because it will occur.

Therefore, in the present information system 1, when the capacity control unit 23 assigns the logical chunk LC to the storage control unit pair 25 and associates a plurality of physical chunk PCs with the logical chunk LC, the plurality of physical chunk PCs are different from each other. It is decided to select from the physical chunk PCs provided by the plurality of storage nodes 4.

On the other hand, when the physical chunk PC associated with the logical chunk LC is selected from the physical chunk PCs in the storage node 4 different from the storage node 4 in which the active storage control unit 22 is arranged, the active storage control unit is selected. When the capacity control unit 23 (capacity control unit 23 in the same storage node 4 as the active storage control unit 22) that has received the I / O command from 22 reads / writes data to the physical chunk PC, the physical chunk Communication with the storage node 4 that provides the PC is required, and there is a problem that the response performance of the entire system deteriorates accordingly. Therefore, when associating a plurality of physical chunk PCs with the logical chunk LC, one of the physical chunk PCs is provided by the storage device 13 in the storage node 4 in which the active storage control unit 22 is arranged. It is desirable to select from these from the viewpoint of the response performance of the entire system.

Further, considering that the passive storage control unit 22 is switched to the active mode when a failure occurs in the storage node 4 in which the active storage control unit 22 is arranged in the storage control unit pair 25, for the same reason as described above. As for one of the physical chunk PCs associated with the logical chunk LC, it is better to select from the physical chunk PCs provided by the storage device 13 in the storage node 4 in which the passive storage control unit 22 is arranged. It is also desirable from the viewpoint of performance.

Therefore, in the information processing system 1, when the capacity control unit 23 assigns the logical chunk LC to the storage control unit pair 25 and associates a plurality of physical chunk PCs with the logical chunk LC, the active storage control of the storage control unit pair 25 is performed. The physical chunk PC provided by the storage device 13 in the storage node 4 in which the unit 22 is arranged, and the physical provided by the storage device 13 in the storage node 4 in which the passive storage control unit 22 of the storage control unit pair 25 is arranged. The capacity control unit 23 is equipped with a capacity priority allocation function that preferentially associates a chunk PC with the logical chunk LC.

However, for the logical chunk LC in the pool PL assigned to one storage control unit pair 25, the storage node in which the active storage control unit 22 and the passive storage control unit 22 constituting the storage control unit pair 25 are arranged. When the physical chunk PCs are associated with each other indefinitely from 4, the storage node 4 has the active storage control unit 22 and the passive storage control unit 22 arranged in the storage node 4 with respect to the logical chunk LC of the other storage control unit pair 25. There is a problem that the physical chunk PC cannot be associated with the storage device 13 of the above.

Therefore, such a capacity priority allocation function includes a storage node 4 in which the active storage control unit 22 of the storage control unit pair 25 is arranged with respect to the storage control unit pair 25, and a passive storage control unit of the storage control unit pair 25. It also includes a function of suppressing the capacity of the physical chunk PC allocated from the storage node 4 in which the 22 is arranged.

As a means for realizing such a capacity priority allocation function, the capacity control unit 23 includes a capacity allocation processing unit 30, a physical chunk selection processing unit 31, a failover processing unit 32, and a re-redundancy processing, as shown in FIG. The unit 33 is provided. Further, as shown in FIG. 6, in the memory 12 of each storage node 4, in addition to the above-mentioned front-end driver 20, back-end driver 21, one or more storage control units 22 and capacity control unit 23, a storage control unit The pair management table 34, the physical chunk management table 35, the logical chunk management table 36, the free physical chunk number management table 37, and the virtual volume management table 70 are stored.

The capacity allocation processing unit 30 is a program having a function of associating a physical chunk PC with a logical chunk LC assigned to the storage control unit pair 25. Further, the physical chunk selection processing unit 31 is, for example, a program that is called in the process of associating the physical chunk PC with the logical chunk LC by the capacity allocation processing unit 30 and has a function of selecting the physical chunk PC associated with the logical chunk LC.

Further, the failover processing unit 32 monitors each storage node 4 in which the active storage control unit 22 is arranged for each storage control unit pair 25, and when a failure occurs in the storage node 4, the storage control unit pair 25 is monitored. This is a program having a function of passing the I / O processing executed by the active storage control unit 22 of the storage control unit 22 to the passive storage control unit 22 of the storage control unit pair 25.

In practice, when the failover processing unit 32 detects a failure of the storage node 4 in which the active storage control unit 22 of the storage control unit pair 25 is arranged, the failover processing unit 32 activates the state of the passive storage control unit 22 of the storage control unit pair 25. By switching to the mode, the I / O processing executed by the active storage control unit 22 of the storage control unit pair 25 is taken over by the passive storage control unit 22 of the storage control unit pair 25.

Further, the re-redundation processing unit 33 monitors each storage node 4 that provides each physical chunk PC associated with the logical chunk LC assigned to the storage control unit pair 25 for each storage control unit pair 25. , A program having a function of associating a physical chunk PC provided by another storage node 4 with a logical chunk LC in place of the physical chunk PC provided by the storage node 4 when a failure occurs in any of the storage nodes 4. Is. By the processing of the re-redundancy processing unit 33, the storage destination of the data written in the logical chunk LC is re-redundant.

On the other hand, the storage control unit pair management table 34 is a table used by the capacity control unit 23 to manage the configuration of each storage control unit pair 25, and as shown in FIG. 7, the storage control unit pair number column 34A, It is configured to include an active side placement destination node number field 34B, a passive side placement destination node number field 34C, and a LUN field 34D. In the storage control unit pair management table 34, one row corresponds to one storage control unit pair 25.

Then, in the storage control unit pair number column 34A, all the numbers (pair numbers) unique to the storage control unit pair 25 assigned to each storage control unit pair 25 defined in the corresponding cluster 6 are stored. ..

Further, in the active side placement destination node number column 34B, a number (node number) unique to the storage node 4 assigned to the storage node 4 in which the active storage control unit 22 of the corresponding storage control unit pair 25 is placed is stored. Then, the node number of the storage node 4 in which the passive storage control unit 22 of the storage control unit pair 25 is arranged is stored in the passive side placement destination node number column 34C.

Further, in the LUN column 34D, the LUN which is the identification information for identifying the virtual volume provided to the host 3 is stored, and the storage control unit pair number and the active side allocation destination node number for managing the virtual volume identified by the LUN are stored. , It is managed in association with the passive side placement destination node number.

When the front-end driver 20 of each storage node 4 receives an I / O request (read request or write request) from the host 3, the front-end driver 20 acquires the LUN included in the I / O request and displays the storage control unit pair management table 34. The storage control unit pair number, the active side placement destination node number, and the passive side placement destination node number associated with the LUN are specified. As a result, the front-end driver 20 of each storage node 4 can specify the storage control unit pair that manages the virtual volume that is the I / O request destination and the placement destination node of the storage control unit pair.

Therefore, in the case of the example of FIG. 7, the storage control unit pair 25 to which the pair number "1" is assigned is passive because the active storage control unit 22 is arranged in the storage node 4 to which the node number "1" is assigned. It is shown that the storage control unit 22 is arranged in the storage node 4 to which the node number "2" is assigned.

The physical chunk management table 35 is a table used by the capacity control unit 23 to manage the physical chunk PCs defined in the cluster 6, and as shown in FIG. 8, the physical chunk number column 35A and the affiliation node number column It is configured to include 35B, a drive number field 35C, and an in-drive offset field 35D. In the physical chunk management table 35, one row corresponds to one physical chunk PC.

Then, in the physical chunk number column 35A, all the numbers (physical chunk numbers) unique to the physical chunk PC assigned to each physical chunk PC in the cluster 6 are stored, and the belonging node number column 35B corresponds to the corresponding. The node number of the storage node 4 that provides the physical chunk PC is stored.

Further, in the drive number column 35C, a number (drive number) unique to the storage device 13 assigned to the storage device 13 that provides the physical chunk PC in the storage node 4 that provides the corresponding physical chunk PC is stored. NS. Further, the in-drive offset column 35D stores the offset position of the physical chunk PC in the storage area provided by the storage device 13 from the beginning of the storage area.

Therefore, in the case of the example of FIG. 8, the physical chunk PC having the physical chunk number “0” is “0” from the beginning of the storage device 13 having the drive number “0” mounted on the storage node 4 having the storage number “0”. It is shown that the storage area has a predetermined size starting from a position offset by "0x00000".

The logical chunk management table 36 is a table used by the capacity control unit 23 to manage the logical chunk LC defined in the cluster 6, and as shown in FIG. 9, the logical chunk number column 36A and the allocation destination storage control. It is configured to include a unit pair number column 36B, a master physical chunk number column 36C, and a mirror physical chunk number column 36D. In the logical chunk management table 36, one row corresponds to one logical chunk LC.

A number (logical chunk number) unique to each logical chunk LC assigned to each logical chunk LC in the cluster 6 is stored in the logical chunk number column 36A, and the allocation destination storage control unit pair number column 36B stores the unique number (logical chunk number). , The pair number of the storage control unit pair 25 to which the corresponding logical chunk LC is assigned is stored.

Further, in the master physical chunk number column 36C, the physical chunk number of the physical chunk PC set as the "master" among the plurality of physical chunk PCs associated with the corresponding logical chunk LC is stored, and the mirror physical chunk number is stored. Among the plurality of physical chunk PCs, the physical chunk number of the physical chunk PC set in the "mirror" is stored in the column 36D.

Therefore, in the case of the example of FIG. 9, the logical chunk LC having the logical chunk number “0” is assigned to the storage control unit pair 25 having the pair number “0”, and the logical chunk LC is assigned to the “master”. It is shown that the physical chunk PC having the chunk number set to "0" and the physical chunk PC having the chunk number set to "mirror" set to "4" are associated with each other.

FIG. 36 shows the configuration of the virtual volume management table 70. The virtual volume management table 70 is a table used for managing the correspondence between the area of each virtual volume and the logical chunks associated with the area, and as shown in FIG. 36, the LUN column 70A and the VVOL address column. It is configured to include 70B, a logical chunk number field 70C, and a logical chunk address field 70D.

The LUN column 70A and the VVOL address column 70B store the LUN and the address (VVOL address) of the virtual volume, which are identification information for identifying the virtual volume provided to the host 3, respectively, and are stored in the logical chunk number column 70C and the logical chunk. In the address field 70D, the logical chunk number of the logical chunk assigned to the area of the virtual volume identified by the LUN and the VVOL address and the address (logical chunk address) of the logical chunk are stored, respectively. The information in the LUN column 70A, the VVOL address column 70B, the logical chunk number column 70C, and the logical chunk address column 70D is managed in association with each virtual volume area.

In the VVOL address field 70B, for example, a range of addresses indicating the area of the virtual volume may be stored, or the start address of the area of the virtual volume may be stored. Similarly, the logical chunk address field 70D may store, for example, a range of addresses indicating the area of the logical chunk, or may store the start address of the area of the logical chunk.

When the storage control unit 22 of each storage node 4 receives the I / O request (read request or write request), the storage control unit 22 acquires the LUN and VVOL address included in the I / O request, and uses the virtual volume management table 70. The logical chunk number and the logical chunk address associated with the LUN and VVOL address are specified. As a result, the storage control unit 22 of each storage node 4 can specify the area of the logical chunk allocated to the area of the virtual volume that is the I / O request destination.

Regarding the virtual volume management table 70, for example, the virtual volume management table 70 for all virtual volumes provided by the cluster (distributed storage system) 6 may be shared by each node, and for example, for each storage control unit pair 25, the virtual volume management table 70 may be shared. The virtual volume management table 70 related to the virtual volume provided by the storage control unit pair 25 may be managed.

The free physical chunk number management table 37 capacity-controls the total number of unused physical chunk PCs (hereinafter, referred to as free physical chunk PCs) that are not yet associated with any logical chunk LC in each storage node 4. It is a table used for management by the unit 23, and is configured to include a node number column 37A and a free physical chunk number column 37B as shown in FIG. In the free physical chunk number management table 37, one row corresponds to one storage node 4.

The node number column 37A stores the node numbers of all the storage nodes 4 existing in the cluster 6, and the free physical chunk number column 37B stores the total number of empty physical chunks in the corresponding storage node 4. Will be done.

Therefore, in the case of the example of FIG. 10, the storage node 4 having the node number “0” has no free physical chunk PC (the number of free physical chunks is “0”), and the storage node 4 having the node number “1” has no free physical chunk PC. It is shown that there are "10" free physical chunk PCs.

(1-3) Various processes executed in the storage node (1-3-1) Write processing In FIG. 11, the active storage control unit 22 of the storage control unit pair 25 is associated with the storage control unit pair 25. The flow of write processing executed when a write request for writing a virtual volume VVOL is received is shown.

When the active storage control unit 22 receives such a write request, the write process shown in FIG. 11 is started. First, the active storage control unit 22 that has received the write request has a capacity virtualization function, a local copy function, and a remote. Execute necessary processing related to necessary functions such as copy function (S1).

After that, the active storage control unit 22 sets the address of the write destination area specified in the write request in the write target virtual volume (write target virtual volume) VVOL with the chunk number of the corresponding logical chunk LC and the corresponding logical chunk LC. Generates an I / O command converted to the offset position of the corresponding logical area in the logical chunk LC (S2), and transmits the generated I / O command to the capacity control unit 23 in the own storage node 4 (S3). .. In the following, it is assumed that the logical area has already been allocated from the logical chunk LC to the write destination area of the write target virtual volume VVOL specified in the write request.

The capacity control unit 23 that has received this I / O command is associated with all the logical chunk LCs that allocate the logical area to the write destination area in the write target virtual volume VVOL specified in the write request from the host device 3. The physical chunk PC of is selected as the write destination physical chunk PC (S4).

Subsequently, the capacity control unit 23 is in a state where any of the selected physical chunk PCs to be written is blocked due to a failure of the storage device 13 or the like (hereinafter, this state is referred to as “blocking”), or data. Is being copied (hereinafter referred to as "rebuilding"), and data copying to the storage area in the storage device 13 corresponding to the read destination area specified in the write request has not been completed yet. It is determined whether or not it is in any of the above states (S5).

Then, when the capacity control unit 23 obtains a negative result in this determination, the capacity control unit 23 proceeds to step S7. On the other hand, when the capacity control unit 23 obtains an affirmative result in the determination in step S5, it is a "blocked" write-destination physical chunk PC or a "rebuilding" storage area corresponding to the write-destination area. The physical chunk PC whose data copy to is not yet completed is excluded from the write destination physical chunk PC (S6), and then the process proceeds to step S7.

Further, when the capacity control unit 23 proceeds to step S7, the capacity control unit 23 refers to the physical chunk management table 35 (FIG. 8), and for all the write destination physical chunk PCs, the storage device 13 that provides the write destination physical chunk PCs, respectively. The drive number and the offset position of the write destination physical chunk PC in the storage device 13 are acquired (S7).

Then, the capacity control unit 23 stores the write data received by the active storage control unit 22 at that time together with the write request in the corresponding storage areas in all the write destination physical chunk PCs based on the acquired information. (S8). This completes this write process.

(1-3-2) Read processing On the other hand, in FIG. 12, the active storage control unit 22 of the storage control unit pair 25 makes a read request for reading the virtual volume VVOL associated with the storage control unit pair 25. The flow of processing executed when it is received is shown.

When the active storage control unit 22 receives such a read request, the read process shown in FIG. 12 is started. First, the active storage control unit 22 that has received the read request starts the capacity virtualization function, the local copy function, and the remote. Execute necessary processing related to necessary functions such as copy function (S10).

After that, the active storage control unit 22 sets the address of the read destination area specified in the read request in the read target virtual volume (read target virtual volume) VVOL with the chunk number of the corresponding logical chunk LC and the corresponding logical chunk LC. Generates an I / O command converted to the offset position of the corresponding logical area in the logical chunk LC (S11), and transmits the generated I / O command to the capacity control unit 23 in the own storage node 4 (S12). ..

Upon receiving this I / O command, the capacity control unit 23 allocates a logical area to the read destination area specified in the read request in the read target virtual volume VVOL specified in the read request from the host device 3, and the logical chunk LC. Of all the physical chunk PCs associated with, the physical chunk PC set as the “master” is selected as the read destination physical chunk PC (S13).

Subsequently, the capacity control unit 23 determines whether or not the selected read destination physical chunk PC is in either the “blocked” state or the “rebuilding” state (S14). Then, when the capacity control unit 23 obtains an affirmative result in this determination, the physical chunk PC associated with the logical chunk LC in which the logical area is allocated to the read destination area of the read target virtual volume VVOL specified in the read request. Among them, the physical chunk PC for which the determination in step S14 has not been performed is selected as the new read destination physical chunk PC (S15). Further, the capacitance control unit 23 returns to step S14, and then repeats the loop of step S14-step S15-step S14 until a negative result is obtained in step S14.

Then, when the capacity control unit 23 eventually obtains a negative result in step S14 by detecting a physical chunk PC that matches the conditions, the drive number of the storage device 13 that provides the read destination physical chunk PC and the storage device 13 are used. The offset position of the lead destination physical chunk PC is acquired from the physical chunk management table 35 (FIG. 8) (S16).

Further, the capacity control unit 23 reads the data specified in the read request from the host device 3 from the storage device 13 (S17) based on the acquired information, and reads the read read data into the I / O command. After returning to the active storage control unit 22 of the transmission source of (S18), this read process is terminated. The active storage control unit 22 that has received the read data then transfers the read data to the host device 3 that is the source of the read request.

When the data to be data is distributed and stored in a plurality of physical chunk PCs together with redundant data by EC (Erasure Coding), the capacity control unit 23 performs all of these plurality in step S13 described above. Select the physical chunk PC as the lead destination physical chunk PC.

Further, in step S14, the capacity control unit 23 determines whether or not at least one of the selected read-destination physical chunk PCs is in either the “blocked” or “rebuilding” state. Then, when the capacity control unit 23 obtains a negative result in this determination, the capacity control unit 23 executes the steps S16 and S17 for each read destination physical chunk PC in the same manner as described above to obtain data from the read destination physical chunk PCs, respectively. read out. Further, the capacity control unit 23 generates the original data based on these read data, and in step S18, the generated data is used as read data in the active storage control unit 22 of the transmission source of the I / O command. Reply, and then end this read process.

On the other hand, when the capacity control unit 23 obtains an affirmative result in step S14, the capacity control unit 23 performs steps S16 and S17 for the remaining read destination physical chunk PCs that are not “blocked” or “rebuilding” in the same manner as described above. Data is read from each of these read destination physical chunk PCs by executing the above. Further, the capacity control unit 23 restores the data stored in the read destination physical chunk PC "in blockage" or "rebuilding" based on these read data, and uses the restored data to obtain the original data. Is generated, and in step S18, the generated data is returned as read data to the active storage control unit 22 of the transmission source of the I / O command, and then the read process is terminated.

(1-3-3) Capacity allocation process On the other hand, FIG. 13 shows the initial or additional storage capacity (1-3-3) from the active storage control unit 22 in the same storage node 4 to the storage control unit pair 25 to which the active storage control unit 22 belongs. The processing procedure of the capacity allocation processing executed by the capacity allocation processing unit 30 (FIG. 5) of the capacity control unit 23 requested to allocate the physical chunk PC) is shown.

When such a request is given, the capacity allocation processing unit 30 starts the capacity allocation process shown in FIG. 13, and first, referring to the storage control unit pair management table 34 (FIG. 7), the storage to which the storage capacity should be allocated is stored. All the node numbers of the storage nodes 4 in which the storage control units 22 constituting the control unit pair 25 (hereinafter, referred to as the target storage control unit pair 25) are arranged are acquired (S20).

Subsequently, the capacity allocation processing unit 30 sets the storage node 4 in which the active storage control unit 22 in the target storage control unit pair 25 is arranged as the priority node (S21). Further, the capacity allocation processing unit 30 sets the lower limit of the total capacity of the free physical chunks in the storage node 4 in which the active storage control unit 22 in the target storage control unit pair 25 is arranged as the active free capacity threshold value (S22). .. It should be noted that the value of the free space threshold value for active use may be set as a constant in the program in advance even if it is set by a system administrator or the like when the information processing system 1 is installed. May be good.

Next, the capacity allocation processing unit 30 calls the physical chunk selection processing unit 31 (FIG. 5) and requests the selection of the physical chunk PC to be associated with the logical chunk LC to be assigned to the target storage control unit pair 25 (S23). Thus, the physical chunk selection processing unit 31 that has received this request assigns the physical chunk PCs to be assigned to the logical chunk LC of the target storage control unit pair 25 from the free physical chunk PCs in the cluster 6 to the priority node (here, in this case). The free physical chunk PC in the storage node 4) in which the active storage control unit 22 is arranged is preferentially selected, and the chunk number of the selected free physical chunk PC is notified to the capacity allocation processing unit 30.

Then, when the chunk number of the free physical chunk PC is notified from the physical chunk selection processing unit 31, the capacity allocation processing unit 30 secures the physical chunk PC of the chunk number as the “master” physical chunk PC (S24). ..

Next, the capacity allocation processing unit 30 sets the storage node 4 in which the passive storage control unit 22 of the target storage control unit pair 25 is arranged as the priority node (S25). Further, the capacity allocation processing unit 30 sets the lower limit of the total capacity of the free physical chunk PCs in the storage node 4 in which the passive storage control unit 22 in the target storage control unit pair 25 is arranged as the passive free capacity threshold value (S26). ). It should be noted that the value of the free space threshold value for passive use may be set as a constant in the program in advance even if it is set by a system administrator or the like when the information processing system 1 is installed. May be good.

Further, the capacity allocation processing unit 30 sets the storage node 4 that provides the physical chunk PC set as the “master” secured in step S24 as the exclusion node (S27).

Subsequently, the capacity allocation processing unit 30 calls the physical chunk selection processing unit 31 (FIG. 5) and requests the selection of the physical chunk PC to be associated with the logical chunk LC to be assigned to the target storage control unit pair 25 (S28). Thus, the physical chunk selection processing unit 31 that has received this request assigns the physical chunk PCs to be assigned to the logical chunk LC of the target storage control unit pair 25 from the free physical chunk PCs in the cluster 6 to the priority node (here, in this case). The passive storage control unit 22 is preferentially selected from the free physical chunk PCs in the storage node 4) in which the passive storage control unit 22 is arranged, and the chunk number of the selected free physical chunk PC is notified to the capacity allocation processing unit 30.

Then, when the chunk number of the free physical chunk PC is notified from the physical chunk selection processing unit 31, the capacity allocation processing unit 30 secures the physical chunk PC of the chunk number as the physical chunk PC of the “mirror” (S29). ..

Subsequently, the capacity allocation processing unit 30 creates a new logical chunk LC, and associates the created logical chunk LC with the physical chunk PC secured in step S24 and the physical chunk PC secured in step S29 ( S30). Further, the capacity allocation processing unit 30 allocates the logical chunk LC created in step S30 to the pool PL of the target storage control unit pair 25 (S31).

Then, the capacity allocation processing unit 30 ends the additional capacity allocation processing after this. The capacity allocation processing unit 30 of each storage node 4 may set the passive free capacity threshold value and the active free capacity threshold value to the same value or may be set to different values. Further, each of the passive free space threshold value and the active free space threshold value may be a value common to the storage nodes 4 or may be a different value for each storage node 4.

The capacity allocation processing unit 30 of the storage node 4 may set the passive free capacity threshold value to a value higher than the active free capacity threshold value. In this case, when the free space of the storage node 4 is lower than the free space threshold for passive but exceeds the free space threshold for active, a new mirror physical chunk is used for the data of the passive storage control unit 22 from the storage area of the storage node. Is not assigned, but a new master physical chunk can be assigned for the data of the active storage control unit 22. As a result, in the storage node 4, the master physical chunk accessed from the active storage control unit 22 can be preferentially assigned to the mirror physical chunk.

(1-3-4) Physical chunk selection process In FIG. 14, a physical chunk PC corresponding to a logical chunk to be assigned from the capacity allocation processing unit 30 to the target storage control unit pair 25 is selected from the capacity allocation processing unit 30 in steps S23 and S28 of the capacity allocation process described above. The processing content of the physical chunk selection processing executed by the physical chunk selection processing unit 31 requested to do so is shown.

When the request is given by the capacity allocation processing unit 30, the physical chunk selection processing unit 31 starts the physical chunk selection processing shown in FIG. 14, and first, the priority node notified from the capacity allocation processing unit 30 together with the request. Get the node number of (S40).

Further, when the exclusion node is set, the physical chunk selection processing unit 31 also acquires the node number of the exclusion node notified from the capacity allocation processing unit 30 together with the request (S41).

Further, the physical chunk selection processing unit 31 notifies the active free capacity threshold value (in the case of the request in step S23) or the passive free space threshold value (in the case of the request in step S28) notified from the capacity allocation processing unit 30 together with the request. (S42).

Subsequently, in the physical chunk selection processing unit 31, the priority node is not an exclusion node, and the total capacity of the priority node is the active free capacity threshold (in the case of the request in step S23) or the passive free capacity threshold (step S28). (In the case of the request) It is determined whether or not there are more free physical chunk PCs (S43). This determination is made with reference to the free physical chunk number management table 37 (FIG. 10).

Then, when the physical chunk selection processing unit 31 obtains an affirmative result in this determination, it refers to the physical chunk management table 35 (FIG. 8) and selects one free physical chunk PC from the free physical chunk PCs in the priority node. After selecting (S44) and notifying the capacity allocation processing unit 30 of the chunk number of the selected free physical chunk PC (S47), the physical chunk selection process ends.

On the other hand, when the physical chunk selection processing unit 31 obtains a negative result in the determination in step S43, the physical chunk selection processing unit 31 selects one storage node 4 from the storage nodes 4 other than the priority node and the exclusion node in the cluster 6 ( S45). As a method of selecting the storage node 4 at this time, for example, a method of selecting the storage node 4 having the largest number of free physical chunk PCs can be applied by referring to the free physical chunk number management table 37.

Subsequently, the physical chunk selection processing unit 31 selects one free physical chunk PC from the free physical chunk PCs in the storage node 4 selected in step S45 (S46). Further, the physical chunk selection processing unit 31 notifies the capacity allocation processing unit 30 of the chunk number of the selected free physical chunk PC (S47), and then ends the physical chunk selection processing.

(1-3-5) Failover Processing On the other hand, FIG. 15 shows a processing procedure of failover processing that is periodically executed by the failover processing unit 32 (FIG. 5) of the capacity control unit 23.

When the failover processing unit 32 starts this failover processing, it first determines whether or not the processing after step S51 has been executed for all the storage control unit pairs 25 in the cluster 6 (S50).

Then, when the failover processing unit 32 obtains a negative result in this determination, it selects one unselected storage control unit pair 25 from all the storage control unit pairs 25 in the cluster 6 in step S51 (S51) and selects it. It is determined whether or not a failure has occurred in the storage node 4 in which the active storage control unit 22 of the storage control unit pair 25 (hereinafter, this is referred to as the first selection storage control unit pair 25) is arranged (S52). ).

When the failover processing unit 32 obtains a negative result in this determination, the failover processing unit 32 returns to step S50, and after that, the first selected storage control unit pair 25 is sequentially switched to another unprocessed storage control unit pair 25, and after step S50. Repeat the process of.

On the other hand, when the failover processing unit 32 obtains an affirmative result in the determination in step S52, the failover processing unit 32 switches the state of the passive storage control unit 22 of the first selected storage control unit pair 25 to the active mode, and the storage control unit pair. The state of the active storage control unit 22 of 25 is switched to the passive mode (S53).

Subsequently, the failover processing unit 32 refers to the logical chunk management table 36 (FIG. 9) and determines whether or not there is a logical chunk LC assigned to the first selected storage control unit pair 25 (S54). ).

Then, when the failover processing unit 32 obtains a negative result in this determination, it returns to step S50, and after that, the first selected storage control unit pair 25 is sequentially switched to the other unprocessed storage control unit pair 25, and after step S50. Repeat the process of.

On the other hand, when the failover processing unit 32 obtains an affirmative result in the determination in step S54, the failover processing unit 32 “masters” the physical chunk PC associated with the logical chunk LC assigned to the first selected storage control unit pair 25. And "Mirror" settings are switched (S55).

Specifically, the failover processing unit 32 is stored in the active placement destination node number column 34B (FIG. 7) of the row corresponding to the first selected storage control unit pair 25 in the storage control unit pair management table 34 (FIG. 7). The storage control unit pair management table 34 is updated so that the node number stored in the node number and the node number stored in the passive placement destination node number field 34C (FIG. 7) of the row are exchanged (S55).

Next, the failover processing unit 32 returns to step S50, and then repeats the processing after step S50 while sequentially switching the first selected storage control unit pair 25 to another unprocessed storage control unit pair 25.

Then, when the failover processing unit 32 obtains an affirmative result in step S50 by completing the processing after step S52 for all the storage control unit pairs 25 defined in the cluster 6, the failover processing ends.

(1-3-6) Re-redundancy processing On the other hand, FIG. 16 shows a processing procedure of the re-redundancy processing that is periodically executed by the re-redundancy processing unit 33 (FIG. 5) of the capacity control unit 23.

When the re-redundancy processing unit 33 starts this re-redundancy processing, it first determines whether or not the processing after step S61 has been executed for all the storage control unit pairs 25 in the cluster 6 (S60).

Then, when the re-redundation processing unit 33 obtains a negative result in this determination, it selects one storage control unit pair 25 that has not yet been selected in step S61 from all the storage control unit pairs 25 in the cluster 6 (S61). ), Provide any physical chunk PC associated with any logical chunk LC assigned to the selected storage control unit pair (hereinafter, this is referred to as a second selected storage control unit pair) 25. It is determined whether or not the storage device 13 or the storage node 4 on which the storage device 13 is mounted has a failure (S62).

When the re-redundancy processing unit 33 obtains a negative result in this determination, the process returns to step S60, and then, while sequentially switching the second selected storage control unit pair 25 to another unprocessed storage control unit pair 25, the step The processing after S60 is repeated.

On the other hand, when the re-redundancy processing unit 33 obtains an affirmative result in the determination in step S62, the physical chunk PC provided by the storage node 4 in which the failure has occurred (hereinafter, this is referred to as a failure physical chunk) PC. It is determined whether or not the PC is a physical chunk PC set in the "master" (S63).

Then, when the re-redundancy processing unit 33 obtains an affirmative result in this determination, the re-redundancy processing unit 33 sets the storage node 4 in which the active storage control unit 22 of the second selected storage control unit pair 25 is arranged as the priority node (S64). Further, when the re-redundancy processing unit 33 obtains a negative result in the determination in step S63, the re-redundancy processing unit 33 sets the storage node 4 in which the passive storage control unit 22 of the second selection storage control unit pair 25 is arranged as the priority node (S65). ).

Subsequently, the re-redundancy processing unit 33 is a storage node 4 that provides any physical chunk PC associated with any logical chunk LC assigned to the second selection storage control unit pair 25. Then, the storage node 4 in which the failure has not occurred is set as the exclusion node (S66).

Next, the re-redundancy processing unit 33 calls the physical chunk selection processing unit 31 (FIG. 5) to execute the physical chunk selection process described above for FIG. 14, thereby replacing the fault physical chunk PC detected in step S62. The physical chunk PC is selected (S67).

Then, the re-redundancy processing unit 33 subsequently detects the physical chunk (hereinafter, this is referred to as a selected physical chunk) PC selected by the physical chunk selection processing unit 31 in step S67 as a fault physical chunk in step S62. It is associated with the corresponding logical chunk (hereinafter, this is referred to as a re-redundancy target logical chunk) LC assigned to the second selected storage control unit pair 25 instead of the PC (S68).

Specifically, the re-redundation processing unit 33 has the master physical chunk number column 36C of the row corresponding to the re-redundant target logical chunk PC in the logical chunk management table 36 (FIG. 9) (the fault physical chunk PC is the “master”). Case) or the chunk number of the fault physical chunk PC stored in the mirror physical chunk number field 36D (when the fault physical chunk PC is "mirror") is rewritten to the chunk number of the selected physical chunk PC.

Further, in step S68, the re-redundancy processing unit 33 has a free physical chunk number column 37B (FIG. 10) in the row corresponding to the storage node 4 that provides the selected physical chunk PC in the free physical chunk number management table 37 (FIG. 10). The numerical value stored in 10) is updated to a value reduced by "1".

Further, when the fault physical chunk PC is a physical chunk PC set as the “master”, the re-redundation processing unit 33 corresponds to the re-redundation target logical chunk LC together with the fault physical chunk PC in this step S68. Among the attached physical chunk PCs set in the "mirror", the physical chunk PC provided by the storage node 4 in which the active storage control unit 22 is arranged is switched to the "master". Specifically, the re-redundancy processing unit 33 is a chunk stored in the master physical chunk number column 36C (FIG. 9) of the row corresponding to the re-redundancy target logical chunk LC in the logical chunk management table 36 (FIG. 9). The number is replaced with the chunk number stored in the mirror physical chunk number column 36D (FIG. 9) of that line.

After that, the re-redundancy processing unit 33 sets the state of the selected physical chunk PC described above to “rebuilding” (S69). Further, the re-redundancy processing unit 33 executes a rebuild process of restoring the data stored in the failed physical chunk PC to the selected physical chunk PC (S70).

Specifically, when the data stored in the fault physical chunk PC is mirrored to another physical chunk PC, the re-redundancy processing unit 33 transfers the data stored in the other physical chunk PC. Make a full copy to the selected physical chunk PC. If the data stored in the failed physical chunk PC is part of the Erasure-Coding data, the data is restored using other data, and the restored data is stored in the selected physical chunk PC. do.

Then, when the rebuilding process is completed, the re-redundancy processing unit 33 returns to step S60, and after that, the second selected storage control unit pair 25 is sequentially switched to another storage control unit pair 25 that has not been processed, and steps S60 and subsequent steps are performed. Repeat the process of.

Then, when the re-redundancy processing unit 33 eventually obtains an affirmative result in step S60 by completing the processing after step S61 for all the storage control unit pairs 25 defined in the cluster 6, this re-redundancy processing is performed. To finish.

(1-4) Effect of the present embodiment In the information processing system 1 of the present embodiment configured as described above, at least two physical chunk PCs are assigned to the storage control unit pair 25, and these physics are used. The data is protected because the data is duplicated and stored in the chunk PC.

Further, in the information processing system 1, as these two physical chunk PCs, the physical chunk PC provided by the storage device 13 in the storage node 4 in which the active storage control unit 22 constituting the storage control unit pair 25 is arranged, and the physical chunk PC. A physical chunk PC provided by the storage device 13 in the storage node 4 in which the passive storage control unit 22 constituting the storage control unit pair 25 is arranged is selected.

Therefore, the active storage control unit 22 constituting the storage control unit pair 25 and the passive storage control unit 22 switched to the active mode can quickly access the corresponding physical chunk PC among these physical chunk PCs. It is possible to quickly read / write data to the physical chunk PC.

Therefore, according to the present information processing system 1, data protection can be performed while preventing deterioration of the response performance of the entire system.

(2) Second Embodiment As in the first embodiment, a logical chunk LC is assigned to the storage control unit pair 25, and data is made redundant by associating a plurality of physical chunk PCs with the logical chunk LC. When selecting a physical chunk PC to be associated with a logical chunk LC, it is desirable to consider the fault set as well. Here, the "fault set" means a group of storage nodes 4 in which a failure may occur due to a single power failure.

For example, in order to prevent data loss or the like from occurring due to a failure of one storage node 4, a plurality of physical chunk PCs associated with one logical chunk LC are provided from among the physical chunk PCs provided by different storage nodes 4. Even if selected, if these storage nodes 4 are getting power from the same power supply, read / write data to all physical chunk PCs associated with the logical chunk LC when the power supply goes down. Cannot be done.

Therefore, in the present embodiment, when associating a plurality of physical chunk PCs with the logical chunk LC, the physical chunk PCs associated with the logical chunk LC are selected in consideration of the fault set. Specifically, as the physical chunk PCs associated with one logical chunk LC, the physical chunk PCs provided by the storage nodes 4 belonging to different fault sets are selected.

Further, even if the physical chunk PCs associated with one logical chunk LC are selected from the physical chunk PCs provided by the storage nodes 4 belonging to different fault sets in this way, the storage control to which the logical chunk LC is assigned is selected. When the storage nodes 4 in which the storage nodes 4 constituting the unit pair 25 are arranged belong to the same fault set, an I / O request from the host device 3 when a power failure occurs in the fault set. Can no longer be supported.

Therefore, in the present embodiment, the storage control unit pair 25 is configured by the two storage control units 22 arranged in the storage nodes 4 belonging to different fault sets.

FIG. 17, which is shown by assigning the same reference numerals to the portions corresponding to those in FIG. 1, shows a schematic configuration example of the information processing system 40 according to the present embodiment in which the above-mentioned “fault set” is conscious. Here, the two storage nodes 41 "storage node 1" and "storage node 2" belong to the fault set 42 "fault set 1", and the storage nodes 41 "storage node 4" and "storage node 4" are "fault set 1". It is assumed that the storage node 41 "storage node (2n-1)" and "storage node 2n" belongs to the fault set 42 "fault set 2" and belongs to the fault set 42 "fault set n".

Since the hardware configuration of each storage node 41 is the same as that of the storage node 4 of the first embodiment, the description thereof will be omitted here.

FIG. 18 is shown by assigning the same reference numerals to the portions corresponding to FIG. 3, and shows a configuration example of each storage control unit pair 25 defined in the information processing system 40 of the present embodiment. As shown in FIG. 18, in the case of the present embodiment, each storage control unit pair 25 is composed of two storage control units 22 arranged in storage nodes 41 belonging to different fault sets 42, respectively.

For example, in the case of the example of FIG. 18, the storage control unit pair 25 called “storage control unit pair 1 (SCP1)” is arranged in the storage node 41 called “storage node 1” belonging to the fault set called “fault set 1”. It is composed of a storage control unit 22 called "storage control unit 1" and a storage control unit 22 called "storage control unit 2" arranged in a storage node 41 called "storage node 3" belonging to a fault set called "fault set 2". Has been done.

Further, the storage control unit pair 25 called "storage control unit pair 2 (SCP2)" is a storage called "storage control unit 3" arranged in the storage node 41 called "storage node 3" belonging to the fault set called "fault set 2". It is composed of a control unit 22 and a storage control unit 22 called “storage control unit 4” arranged in a storage node 41 called “storage node 2” belonging to a fault set called “fault set 1”.

For example, the system administrator may perform the configuration setting of each storage control unit pair 25 for each storage node 4 via the management node 5 after grasping the fault set of each storage node 4. good. Further, one of the storage nodes 4 (for example, the storage node 4 selected as a representative in the cluster 6) becomes a storage node 41 belonging to a different fault set 42 with reference to the node management table 44 described later with reference to FIG. 20. The storage control unit pair 25 may be configured from the two storage control units 22 arranged respectively.

FIG. 19 shows an example in which two physical chunk PCs are associated with one logical chunk LC in the information processing system 40 of the present embodiment. As shown in FIG. 19, in the case of the present embodiment, a plurality of physical chunk PCs provided by the storage devices 13 in the storage nodes 41 belonging to different fault sets are associated with each logical chunk LC. Be done.

For example, in the case of the example of FIG. 19, for the logical chunk LC "A", the storage device 13 in the storage node 41 called "storage node 3" belonging to the fault set "fault set 1" provides "A". The physical chunk PC "A" provided by the storage device 13 in the storage node 41 "storage node 5" belonging to the fault set "fault set 3" is associated with each other.

For the logical chunk LC "B", the physical chunk PC "B" provided by the storage device 13 in the storage node 41 "storage node 1" belonging to the fault set "fault set 1" and "B" It is associated with a physical chunk PC called "B" provided by the storage device 13 in the storage node 41 called "storage node 3" belonging to the fault set called "fault set 2".

20 shows a front-end driver 20, a back-end driver 21, one or more storage control units 22, a storage control unit pair management table 34, and physical chunk management according to the first embodiment described above for FIGS. 6 to 10 and the like. In addition to the table 35, the logical chunk management table 36, and the free physical chunk number management table 37, the node management table 44 stored in the memory 12 (FIG. 2) of each storage node 41 of the present embodiment is shown.

The node management table 44 is a table used by the capacity control unit 43 (FIG. 18) of the present embodiment to manage the fault set to which each storage node 41 belongs, and as shown in FIG. 20, the node number column. It is configured to include 44A and a fault set number field 44B.

Then, in the node number column 44A, all the node numbers assigned to each storage node 4 existing in the cluster 6 are stored, and in the fault set number column 44B, the fault set to which the corresponding storage node 4 belongs is assigned. A unique number (fault set number) is stored in the fault set.

Therefore, in the case of the example of FIG. 20, for example, the storage node 41 to which the node number "1" is assigned and the storage node 41 to which the node number "3" is assigned are assigned the fault set number "1". The storage node 41, which belongs to the fault set and is given the node number "2", and the storage node 41, which is given the node number "4", are assigned to the fault set given the fault set number "2". It is shown to belong.

In FIG. 21, instead of the physical chunk selection processing unit 31 of the capacity control unit 23 of the first embodiment described above with respect to FIG. 14, the physical chunk selection processing unit of the capacity control unit 43 (FIG. 18) of the present embodiment is shown. A processing procedure of the physical chunk selection processing according to the present embodiment executed by 45 (FIG. 5) is shown. Since the processing contents of the capacity control unit 43 other than this are the same as those of the capacity control unit 23 of the first embodiment, the description here will be omitted.

When the physical chunk selection processing unit 45 of the capacity control unit 43 of the present embodiment is called in step S23 or step S28 of the capacity allocation processing described above with respect to FIG. 13, the physical chunk selection processing shown in FIG. 21 is started. , Steps S80 to S83 are processed in the same manner as steps S40 to S43 of FIG.

Then, when the physical chunk selection processing unit 45 obtains an affirmative result in the judgment of step S83, the physical chunk selection processing unit 45 processes steps S84 and S89, respectively, in the same manner as in steps S44 and S47 of the capacity allocation process described above for FIG. , Ends this physical chunk selection process.

On the other hand, when the physical chunk selection processing unit 45 obtains a negative result in the determination in step S83, the physical chunk selection processing unit 45 is any other storage node 41 belonging to the same fault set as the priority node, and the storage node 41 is A free physical chunk PC that is not an excluded node and whose total capacity in the storage node 41 is equal to or greater than the active free capacity threshold (in the case of the request in step S23) or the passive free capacity threshold (in the case of the request in step S28). It is determined whether or not a certain storage node 41 exists (S85). This determination is made with reference to the free physical chunk number management table 37 (FIG. 10).

Then, when the physical chunk selection processing unit 45 obtains an affirmative result in this determination, the physical chunk selection processing unit 45 selects one storage node 41 from the storage nodes 41 satisfying the condition of step S83 (S87). As a method of selecting the storage node 41 at this time, for example, referring to the free physical chunk number management table 37, the storage node 41 having the largest number of free physical chunk PCs among the storage nodes 41 satisfying the condition of step S85. The method of selecting is applicable. However, the storage node 41 may be selected by another method.

Subsequently, the physical chunk selection processing unit 45 selects one free physical chunk PC from the free physical chunk PCs in the selected storage node 41 (S88). Further, the physical chunk selection processing unit 45 notifies the capacity allocation processing unit 30 (FIG. 5) of the chunk number of the physical chunk PC selected in step S88 (S89), and then ends the physical chunk selection processing.

On the other hand, when the physical chunk selection processing unit 45 obtains a negative result in the determination in step S85, the physical chunk selection processing unit 45 refers to the free physical chunk number management table 37, and 41 in any storage node belonging to a fault set different from the priority node. Therefore, the storage node 41 is not an exclusion node, and the total capacity in the storage node 41 is the active free capacity threshold (in the case of the request in step S23) or the passive free capacity threshold (in the case of the request in step S28). Case) Select one storage node 41 having the above free physical chunk PCs (S86).

Further, the physical chunk selection processing unit 45 selects one free physical chunk from the free physical chunk PCs in the storage node 41 selected in step S86 (S88). Further, the physical chunk selection processing unit 45 notifies the capacity allocation processing unit 30 of the chunk number of the physical chunk PC selected in step S88 (S89), and then ends the physical chunk selection processing.

As described above, in the information processing system 40 of the present embodiment, the physical chunk PC associated with the logical chunk LC is selected in consideration of the fault set in addition to the configuration of the first embodiment. It is possible to reliably avoid the occurrence of a situation in which data cannot be read / written to all the physical chunk PCs associated with the storage control unit pair 25 due to a single power failure.

Therefore, according to the present embodiment, in addition to the effect obtained by the first embodiment, it is possible to obtain the effect that an information processing system having higher availability and reliability can be constructed.

(3) Third Embodiment FIG. 22 showing the corresponding portion with FIG. 1 with the same reference numerals shows the overall configuration of the information processing system 50 according to the third embodiment. The information processing system 50 is different from the information processing system 1 of the first embodiment in that a hierarchical control function and a capacity allocation function corresponding to the hierarchy control function are implemented in each storage node 51, respectively. Since the other functions of the information processing stem 50 of the present embodiment are substantially the same as those of the information processing system 1 of the first embodiment, the description thereof will be omitted here.

First, the hierarchical control function will be described. The layer control function groups the storage area provided by the storage device 13 into a plurality of storage layers (tiers) according to the response speed of the storage device 13, and the response speed is higher for more frequently accessed data. It is a function to store in the storage area of the storage hierarchy.

Therefore, in the case of the present embodiment, each storage node 51 is equipped with a plurality of types of storage devices 13 having different response speeds, and the physical chunk PCs provided by the same type of storage devices 13 have storage areas of the same storage hierarchy. It is managed as.

For example, when each storage node is equipped with three types of storage devices 13 of SSD, SAS hard disk device, and SATA hard disk device, the physical chunk PC provided by the SSD having the fastest response speed is the storage area of the first storage layer. The physical chunk PC provided by the SATA hard disk device, which has the next fastest response speed, is the storage area of the second storage layer, and the physical chunk PC provided by the SATA hard disk device, which has the slowest response speed, is the storage area of the third storage layer. Be managed.

The access frequency of each data stored in the storage area of each storage layer is managed, the data having the highest access frequency is moved to the storage area of the first storage layer, and the data having the next highest access frequency. Is moved to the storage area of the second storage layer, and the process of moving the data having the lowest access frequency to the storage area of the third storage layer is periodically executed.

According to such a hierarchical control function, while maintaining the response performance to the frequently accessed data, the infrequently accessed data can be stored and held by the low-cost storage device 13, whereby the system as a whole can be stored and held. There is an advantage that the cost can be kept low.

In the information processing system 50 of the present embodiment, as shown in FIG. 23, different storage layers are used for the pool PLs assigned to the storage control unit pairs 25 in order to support such a layer control function. A plurality of logical chunk LCs associated with the physical chunk PCs of the above are assigned. In the following, the description will be made assuming that each storage control unit pair 25 is associated with logical chunk LCs for three storage layers (first to third storage layers), respectively, but the logical chunks associated with the storage control unit pair 25 will be described. The number of storage layers of LC may be other than 3.

Then, when the active storage control unit 22 of each storage control unit pair 25 is given a write request to set the virtual volume VVOL associated with the storage control unit pair 25 to which it belongs as the write target virtual volume VVOL, the write request is given. Initially, the logical area of the logical chunk LC to which the physical chunk PC of the first storage layer having the highest response performance is associated is allocated to the write destination area in the write target virtual volume VVOL specified in.

Further, the active storage control unit 22 subsequently monitors the access frequency of each data written to the write target virtual volume VVOL, and according to the access frequency to these data, the most frequently accessed data is the virtual volume VVOL. The logical area of the logical chunk LC to which the data corresponds to the written storage area in is switched to the logical area in the logical chunk LC belonging to the highest storage hierarchy as needed. Further, the active storage control unit 22 moves the data to the corresponding storage area in the physical chunk PC associated with the logical chunk LC after switching accordingly.

Further, for the data having the next highest access frequency, the active storage control unit 22 assigns the logical area of the logical chunk LC corresponding to the storage area to which the data is written in the virtual volume VVOL to the logical chunk LC belonging to the next highest storage hierarchy. The data is switched to the logical area, and the data is moved to the corresponding storage area in the physical chunk PC associated with the logical chunk LC after the switching.

Further, the active storage control unit 22 sets the logical area of the logical chunk LC corresponding to the storage area in which the data is written in the virtual volume VVOL to the logical area of the logical chunk LC belonging to the lowest storage layer for the data having the lowest access frequency. The data is moved to the corresponding storage area of the physical chunk PC associated with the logical chunk LC after switching to.

As a means for realizing the capacity allocation function of the present embodiment, the memory 12 (FIG. 2) of each storage node 51 of the present embodiment is replaced with the physical chunk management table 35 described above with respect to FIG. The physical chunk management table 52 shown in FIG. 24 is stored, and the logical chunk management table 53 shown in FIG. 25 is stored in place of the logical chunk management table 36 described above in FIG. 9, and the free physics described above in FIG. 10 is further stored. Instead of the chunk number management table 37, the free physical chunk number management table 54 shown in FIG. 26 is stored.

The physical chunk management table 52 of the present embodiment is the same as the physical chunk number column 35A, the affiliation node number column 35B, the drive number column 35C, and the in-drive offset column 35D of the physical chunk management table 35 described above with respect to FIG. In addition to the physical chunk number column 52A, the affiliation node number column 52B, the drive number column 52D, and the in-drive offset column 52E in which the information of the above is stored, a media type column 52C is provided. The media type column 52C stores the media type (SSD, SAS, SATA, etc.) of the storage device 13 that provides the corresponding physical chunk PC.

Therefore, in the case of the example of FIG. 24, the physical chunk PC having the physical chunk numbers "0" to "2", "4", or "5" is the physical chunk provided by the storage device 13 having the media type "SSD". , The physical chunk PC having the physical chunk number "3" is the physical chunk provided by the storage device 13 having the media type "SAS", and the physical chunk PC having the physical chunk numbers "6" and "7" has the media type. It is shown to be a physical chunk provided by the storage device 13 of "SATA".

Further, in the logical chunk management table 53 of the present embodiment, the logical chunk number column 36A of the logical chunk management table 36 described above with respect to FIG. 9, the allocation destination storage control unit pair number column 36B, the master physical chunk number column 36C, and the mirror physical In addition to the logical chunk number column 53A in which the same information as the chunk number column 36D is stored, the allocation destination storage control unit pair number column 53B, the master physical chunk number column 53D, and the mirror physical chunk number column 53E, the media type column 53C Is provided. The media type column 53C stores the media type of the storage device 13 that provides the physical chunk PC associated with the corresponding logical chunk LC.

Therefore, in the case of the example of FIG. 25, the logical chunk LC having the logical chunk number "0" or "1" is the logic associated with the physical chunk PC provided by the storage device 13 having the media type "SSD". It is shown that the logical chunk LC which is a chunk and has the logical chunk number “2” is the logical chunk associated with the physical chunk PC provided by the storage device 13 having the media type “SAS”.

Further, in the free physical chunk number management table 54 of the present embodiment, the same information as the node number column 37A and the free physical chunk number column 37B of the free physical chunk number management table 37 described above with respect to FIG. 10 is stored. In addition to the number column 54A and the free physical chunk number column 54B, the free physical chunk number columns 54C, 54D, and 54E corresponding to each storage hierarchy are provided. Then, in these free physical chunk number columns 54C to 54E, the number of free physical chunk PCs among the physical chunk PCs provided by the media type storage device 13 constituting the corresponding storage hierarchy is stored.

Therefore, in the case of the example of FIG. 26, for example, in the storage node 51 having the node number "1", the space provided by the storage device 13 of the media type "SSD" constituting the first storage hierarchy at present is provided. There are "5" physical chunk PCs, and there are "2" free physical chunk PCs provided by the media type storage device 13 called "SAS" that constitutes the second storage hierarchy, and the third storage hierarchy is used. It is shown that there are "3" free physical chunk PCs provided by the constituent "SATA" media type storage device 13. As described above, it is assumed that which media type storage device 13 belongs to which storage layer is predetermined.

FIG. 27, which is shown by assigning the same reference numerals to the portions corresponding to those in FIG. 5, shows the configuration of the capacitance control unit 55 of the present embodiment. The capacity control unit 55 of the present embodiment is the same as the capacity control unit 23 (FIG. 5) of the first embodiment, except that the processing contents of the capacity allocation processing unit 56 and the physical chunk selection processing unit 57 are different. It is configured in.

FIG. 28 shows the capacity control of the present embodiment in which the active storage control unit 22 in the same storage node 51 requests the allocation of the first or additional storage capacity to the storage control unit pair 25 to which the active storage control unit 22 belongs. The processing procedure of the capacity allocation processing executed by the capacity allocation processing unit 56 (FIG. 27) of unit 55 is shown. In the case of the present embodiment, the active storage control unit 22 requests the capacity control unit 55 to allocate the storage capacity by designating the storage hierarchy.

When such a request is given, the capacity allocation processing unit 56 starts the capacity allocation process shown in FIG. 28, first confirms the designated storage hierarchy (S90), and then steps S91 to S93. No. 13 is processed in the same manner as in steps S20 to S22 of the capacity allocation process of the first embodiment described above.

Subsequently, the capacity allocation processing unit 56 calls the physical chunk selection processing unit 57 (FIG. 27) and selects a physical chunk PC to be associated with the logical chunk LC of the storage hierarchy to be assigned to the target storage control unit pair 25 confirmed in step S90. (S94). Thus, the physical chunk selection processing unit 57 that has received this request assigns the physical chunk PCs to be assigned to the logical chunk LC of the designated storage hierarchy of the target storage control unit pair 25 from the free physical chunk PCs in the cluster 6. Priority is selected from the free physical chunk PCs in the priority node (here, the storage node 51 in which the active storage control unit 22 is arranged), and the chunk number of the selected free physical chunk PC is notified to the capacity allocation processing unit 56. ..

Next, the capacity allocation processing unit 56 processes steps S95 to S98 in the same manner as steps S24 to S27 of the capacity allocation processing of the first embodiment described above with respect to FIG. 13. After that, the capacity allocation processing unit 56 calls the physical chunk selection processing unit 57 (FIG. 27), and the physical chunk PC to be associated with the logical chunk LC of the storage hierarchy to be assigned to the target storage control unit pair 25 confirmed in step S90. Request selection (S99). Thus, the physical chunk selection processing unit 57 that has received this request assigns the physical chunk PCs to be assigned to the logical chunk LC of the designated storage hierarchy of the target storage control unit pair 25 from the free physical chunk PCs in the cluster 6. Priority is selected from the free physical chunk PCs in the priority node (here, the storage node 51 in which the passive storage control unit 22 is arranged), and the chunk number of the selected free physical chunk PC is notified to the capacity allocation processing unit 56. ..

Subsequently, the capacity allocation processing unit 56 processes steps S100 to S102 in the same manner as steps S29 to S31 of the capacity allocation process described above for FIG. 13, and then ends the capacity allocation process.

On the other hand, FIG. 29 selects a physical chunk PC associated with the logical chunk LC to be assigned from the capacity allocation processing unit 56 to the target storage control unit pair 25 in step S94 or step S99 of the capacity allocation processing of the present embodiment described above with respect to FIG. The processing content of the physical chunk selection processing executed by the physical chunk selection processing unit 57 requested to do so is shown.

When the request is given by the capacity allocation processing unit 56, the physical chunk selection processing unit 57 starts the physical chunk selection processing shown in FIG. 29. First, steps S110 to S112 are described in FIG. The same process as in steps S40 to S42 of the physical chunk selection process of the embodiment is performed.

Subsequently, in the physical chunk selection processing unit 57, the priority node is not an exclusion node, and the total capacity of the storage device of the specified hierarchy in the priority node is the active free space threshold (in the case of the request in step S93), or , It is determined whether or not there is a free physical chunk PC equal to or higher than the passive free space threshold value (in the case of the request in step S99) (S113). This determination is made with reference to the free physical chunk number management table 54 (FIG. 26).

When the physical chunk selection processing unit 57 obtains an affirmative result in this determination, the physical chunk selection processing unit 57 refers to the physical chunk management table 52 (FIG. 24), and refers to the free physical provided by the storage device 13 of the designated storage hierarchy in the priority node. After selecting one free physical chunk PC from the chunk PCs (S114) and notifying the capacity allocation processing unit 56 of the chunk number of the selected free physical chunk PC (S117), the physical chunk selection process ends.

On the other hand, when the physical chunk selection processing unit 57 obtains a negative result in the determination in step S113, the physical chunk selection processing unit 57 selects one storage node 51 from the storage nodes 51 other than the priority node and the exclusion node in the cluster 6 ( S115). As a method of selecting the storage node 51 at this time, for example, the number of free physical chunk PCs among the physical chunk PCs provided by the storage device 13 of the designated storage hierarchy is referred to by referring to the free physical chunk number management table 37. The method of selecting the storage node 51 having the largest number can be applied. However, the storage node 51 may be selected by another method.

Subsequently, the physical chunk selection processing unit 57 selects one free physical chunk PC from the free physical chunk PCs provided by the storage device 13 of the designated storage hierarchy in the storage node 51 selected in step S115 ( S116). Further, the physical chunk selection processing unit 57 notifies the capacity allocation processing unit 56 of the chunk number of the selected free physical chunk PC (S117), and then ends the physical chunk selection processing.

As described above, according to the present embodiment, the same effect as that of the first embodiment can be obtained for the information processing system 50 equipped with the hierarchical control function.

(4) Other Embodiments In the first to third embodiments described above, as shown in FIGS. 1, 17, and 22, the storage device 13 that provides the host device with a physical storage area is provided. Although the case where the storage nodes 4, 41 and 51 are mounted is described, the present invention is not limited to this, and for example, as shown in FIG. 30 in which the corresponding portions corresponding to those in FIG. 1 are designated by the same reference numerals. The storage device 13 may not be mounted on each storage node 60, and an external storage device 61 on which the storage device 13 related to the storage node 60 is mounted may be connected.

In this case, the hardware configuration of each storage node 60 is such that the storage device 13 is removed from FIG. 2, and the logical configuration of each storage node 60 is the same as the logical configuration of the storage node 4 configured with respect to FIG. Just do it. In this case, the control contents of the front-end driver 20, the back-end driver 21, the storage control unit 22, and the capacity control unit 23 are the same as those in the first embodiment.

Therefore, in the case of this example, each storage control unit pair 25 is mounted on the external storage device 61 connected to the storage node 60 in which the active storage control unit 22 constituting the storage control unit pair 25 is arranged. The storage device 13 provided by the physical chunk PC provided by the storage device 13 and the external storage device 61 connected to the storage node 60 in which the passive storage control unit 22 constituting the storage control unit pair 25 is arranged is provided. A logical chunk LC associated with the physical chunk PC to be used is assigned.

Further, in the above-described first to third embodiments, the above-mentioned physical chunk selection processing steps S45 and S46 for FIG. 14, the above-mentioned physical chunk selection processing steps S87 and S88 for FIG. 21, and FIG. 29. In steps S115 and S116 of the physical chunk selection process described above, the physical chunk PCs associated with the logical chunk LC assigned to the storage control unit pair 25 from among the free physical chunk PCs in the storage node 4 having the largest number of free physical chunk PCs. However, the present invention is not limited to this, and priority is given to physical chunk PCs in the vicinity of the active storage control unit 22 and the passive storage control unit 22 that constitute the storage control unit pair 25. May be associated with the logical chunk LC assigned to the storage control unit pair 25.

The “physical chunk PC near the location of the storage control unit 22” refers to a physical chunk PC having few network devices such as switches that the storage control unit 22 passes through when accessing the physical chunk PC. Therefore, the nearest physical chunk PC to which the storage control unit 22 is arranged is the physical chunk PC in the storage node 4 in which the storage control unit 22 is arranged.

In this way, the physical chunk PCs in the vicinity of the active storage control unit 22 and the passive storage control unit 22 that constitute the storage control unit pair 25 are associated with the logical chunk LC that is preferentially assigned to the storage control unit pair 25. However, the active storage control unit 22 and the passive storage control unit 22 can quickly access the physical chunk PC, and can quickly read / write data to the physical chunk PC.

Further, in the above-described first embodiment, as described above with respect to FIGS. 13 and 14, the storage node 4 in which the active storage control unit 22 of the storage control unit pair 25 is arranged and the storage control unit pair 25. By setting the storage node 4 in which the passive storage control unit 22 is arranged as the priority node (see step S21 and step S25 in FIG. 13), the physical chunk selection processing unit 31 can be used as a free physical chunk PC of these storage nodes 4. Has been described in the case where is preferentially selected as the physical chunk PC associated with the logical chunk LC to be assigned to the storage control unit pair 25, but the present invention is not limited to this, and the physical chunk selection processing unit 31 is the cluster 6 The free physical chunk PCs associated with such a logical chunk LC may be selected based only on the number of free physical chunk PCs of each storage node 4 in the storage node 4.

Specifically, as shown in FIG. 31, in which the corresponding portions corresponding to those in FIG. 13 are designated by the same reference numerals, the capacity allocation processing unit 30 omits steps S21 and S25 in FIG. The active storage control unit 22 of the 25 and the storage node 4 in which the passive storage control unit 22 is arranged are not set as priority nodes. Further, as shown in FIG. 32 in which the corresponding portions corresponding to those in FIG. 14 are designated by the same reference numerals, the physical chunk selection processing unit 31 determines whether or not the priority node is set after step S42 (S120), and negates. If a result is obtained, the process proceeds to step S43, and if a positive result is obtained, the process proceeds to step S45.

However, it is usually better to select a free physical chunk PC in the storage node 4 in which the active storage control unit 22 of the storage control unit pair 25 is arranged as the physical chunk PC associated with the logical chunk LC assigned to the storage control unit pair 25. Since the response performance of the system as a whole at the time is good, step S21 is executed and processing is omitted for step S25 of FIG. 13 as shown in FIG. 33 in which the corresponding portions corresponding to those of FIG. 13 are designated by the same reference numerals. The capacity allocation processing unit 30 may be configured in the above. Even in this case, the physical chunk selection processing unit 31 may be constructed so as to execute the processing of the flowchart of FIG. 32.

Further, in the third embodiment described above, as described above with respect to FIGS. 28 and 29, the capacity allocation processing unit 56 (FIG. 27) of the capacity control unit 55 (FIG. 27) controls storage for all layers. The storage node 4 in which the active storage control unit 22 of the unit pair 25 is arranged and the storage node 4 in which the passive storage control unit 22 of the storage control unit pair 25 is arranged are set as priority nodes (steps S21 and step of FIG. 13). By (see S25), the physical chunk selection processing unit 57 (FIG. 27) preferentially allocates the free physical chunk PCs of these storage nodes 4 to the storage control unit pair 25 as physical chunk PCs associated with the logical chunk LC. The case where the selection is made has been described, but the present invention is not limited to this, and only when the designated storage hierarchy is the highest storage hierarchy, the physical chunk selection processing unit 57 causes each storage node in the cluster 6 to be selected. The free physical chunk PCs associated with such a logical chunk LC may be selected based only on the number of 4 free physical chunk PCs.

In this case, as shown in FIG. 34 in which the corresponding portions corresponding to those in FIG. 28 are designated by the same reference numerals, the storage layer confirmed in step S90 after the processing in step S91 is executed is the highest storage layer in the capacity allocation processing unit 56. Whether or not it is determined (S130), and if a negative result is obtained, the storage node 4 in which the active storage control unit 22 in the target storage control unit pair 25 is arranged is not set as the priority node, and the step S93 is performed. move on. Further, the capacity allocation processing unit 56 determines whether or not the storage layer confirmed in step S90 after executing the process in step S95 is the highest storage layer (S131), and if a negative result is obtained, the target storage. The process proceeds to step S97 without setting the storage node 4 in which the active storage control unit 22 in the control unit pair 25 is arranged as the priority node.

Further, in this case, as shown in FIG. 35 in which the corresponding portions corresponding to those in FIG. 29 are designated by the same reference numerals, the physical chunk selection processing unit 31 determines whether or not the priority node is set after step S112 ( S132) If a negative result is obtained, the process proceeds to step S115, and if a positive result is obtained, the process proceeds to step S114.

By doing so, the storage area can be allocated from the own node for the highest storage layer that requires response performance, and the storage area can be evenly allocated from each storage node 4 for the other storage layers. can.

Further, the hypervisor may be operated on the server, one or a plurality of virtual machines may be operated on the hypervisor, and various programs shown in FIG. 3 may be operated on the virtual computer. That is, various programs (control software 20, redundancy unit 22, cluster control unit 23) may operate on the hardware of the physical computer or may operate on the virtual computer. Similarly, the compute node 2 may be an application program (host program) running on a virtual computer or a physical host computer (host computer). When the information processing system 1 has a plurality of servers, some of the servers may be located at different sites. Further, a part or all of the server of the information processing system 1 may be on the cloud and the service may be provided to the user via the network.

A configuration (hyper-converged infrastructure) in which a virtual machine on which various programs (control software 20, redundancy unit 22, cluster control unit 23) operates and a virtual computer on which a host program operates are on the same server (node). However, the configuration may be on different servers connected via a network.

The present invention can be applied to an information processing system including a plurality of storage nodes on which one or more SDSs are mounted.

1,40,50 ... Information processing system, 3 ... Host device, 4,41,51,60 ... Storage node, 6 ... Cluster, 11 ... CPU, 13 ... Storage device, 22 ... Storage control Units, 23, 43, 55 ... Capacity control unit, 25 ... Storage control unit pair, 30, 56 ... Capacity allocation processing unit, 31, 45, 57 ... Physical chunk selection processing unit, 32 ... Failover processing unit , 33 ... Redundancy processing unit, 34 ... Storage control unit Pair management table, 35, 52 ... Physical chunk management table, 36, 53 ... Logical chunk management table, 37, 54 ... Free physical chunk number management Table, 42 ... fault set, 44 ... node management table, 61 ... external storage device, LC ... logical chunk, LP ... pool, PC ... physical chunk, VVOL ... virtual volume.

Claims (13)

In a storage system with multiple storage nodes
Each of the storage nodes
A capacity control unit that manages the capacity provided by each storage device of the plurality of storage nodes in the cluster, and
A storage control unit that accepts an I / O request, generates a command corresponding to the received I / O request, and transmits the command to the capacity control unit.
With
The storage system
A plurality of the storage control units arranged in the different storage nodes form a storage control unit group.
The first storage control unit of the storage control unit group is set to the first state for processing the I / O request, and the second storage control unit of the storage control unit group is set to the I / O. Set to a second state that does not process the request,
The data stored in the storage device is made redundant by combining a plurality of data stored in different storage nodes.
The capacity control unit
The physical storage area of a plurality of storage nodes including at least the storage node in which the storage control unit in the first state is arranged and the storage node in which the storage control unit in the second state is arranged is stored. When allocating to the control unit group and allocating the physical storage area to the storage control unit group, the storage node in which the storage control unit in the first state is arranged and the storage control unit in the second state A storage system characterized in that the physical storage area of the storage node in which is arranged is preferentially allocated to the storage control unit group. It is possible to operate a plurality of the storage control unit groups on one storage node.
The storage system according to claim 1, wherein a physical storage area is allocated to each storage control unit group in the storage node. A claim characterized in that, when a failure occurs in the storage node in which the storage control unit in the first state is arranged, the storage control unit in the second state is changed to the first state. Item 1. The storage system according to item 1. Prior to the occurrence of the failure, the master data is stored in the physical storage area of the storage node including the storage node in which the storage control unit in the first state is arranged, and the storage control in the second state is performed. Mirror data is stored in the physical storage area of the storage node including the storage node in which the unit is arranged.
The storage system according to claim 3, wherein the mirror data is switched to master data after the failure occurs. The physical storage area of the storage node in which the storage control unit in the first state is arranged has priority over the physical storage area of the storage node in which the storage control unit in the second state is arranged. The storage system according to claim 1, wherein the storage system is assigned to. Data redundancy is performed by storing the same data in a plurality of the storage nodes, and the data includes master data and mirror data that the storage control unit in the first state writes.
The storage node in which the storage control unit in the first state is arranged stores master data to be read by the storage control unit in the first state, and the storage control unit in the second state is arranged. The storage system according to claim 2, wherein the storage control unit in the first state stores mirror data that is not read in the storage node. The capacity control unit controls the physical storage area of the storage node in which the storage control unit in the first state is arranged and the storage node in which the storage control unit in the second state is arranged. Assign preferentially to department groups,
The free capacity of the physical storage area of the storage node in which the storage control unit in the first state is arranged and the storage node in which the storage control unit in the second state is arranged is smaller than a predetermined threshold value. The storage system according to claim 1, further comprising allocating a physical storage area of the storage node other than the storage node in which the storage control unit in the first or second state is arranged. The threshold value of the storage node in which the storage control unit in the first state is arranged is smaller than the threshold value of the storage node in which the storage control unit in the second state is arranged. The storage system according to claim 7. The capacity control unit
When allocating the physical storage area of the other storage node to the storage control unit group, the other storage node is selected so as not to be stored in the same storage node as other data of the data redundant data set. The storage system according to claim 1, wherein the storage system is assigned. The storage system according to claim 9, wherein the data redundancy is performed by storing the same data in a plurality of storage nodes. The first aspect of claim 1 is that data redundancy is performed by generating a redundant code from the data and storing the redundant code in the storage node different from the storage node in which the data is stored. Described storage system. A plurality of storage control unit groups are provided, and the storage control unit groups are provided.
The storage system according to claim 1, wherein a plurality of the storage control units constituting different storage control unit groups are arranged on the same storage node. In the control method of a storage system having multiple storage nodes,
Each of the storage nodes
A capacity control unit that manages the capacity provided by each storage device of the plurality of storage nodes in the cluster, and
A storage control unit that accepts an I / O request, generates a command corresponding to the received I / O request, and transmits the command to the capacity control unit.
Have,
In the storage system
A plurality of the storage control units arranged in the different storage nodes form a storage control unit group.
The first storage control unit of the storage control unit group is set to the first state for processing the I / O request, and the second storage control unit of the storage control unit group is set to the I / O. Set to a second state that does not process the request,
The data stored in the storage device is made redundant by combining a plurality of data stored in different storage nodes.
The capacity control unit includes at least the storage node in which the storage control unit in the first state is arranged and the storage node in which the storage control unit in the second state is arranged. When allocating the physical storage area to the storage control unit group and allocating the physical storage area to the storage control unit group, the storage node in which the storage control unit in the first state is arranged and the second A first step of preferentially allocating the physical storage area of the storage node in which the storage control unit in the state is arranged to the storage control unit group, and
The capacity control unit is assigned to the storage control unit group to which the storage control unit belongs in response to the command transmitted from the storage control unit set to the first state based on the I / O request. Each of the physical storage areas is provided with a second step of writing data to each of the physical storage areas or reading data from the physical storage area of one of the physical storage areas.
In the first step,
The physical storage area of the storage node in which the storage control unit in the first state is arranged and the storage node in which the storage control unit in the second state is arranged is given priority to the storage control unit group. A method of controlling a storage system, characterized in that it is assigned.

Images