JP2006011932A - Storage apparatus and exclusive control method thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To exclusively use a logical volume by making an FC_SAN initiator and an iSCSI_SAN initiator coexist. <P>SOLUTION: Respective HBAs 1C of a host 1A is connected to CHAs 4A for FC of a storage apparatus via an FC_SAN. Respective HBAs 1D of a host 1B is connected to CHAs 4B for iSCSI via an iSCSI_SAN. A shared memory 5 is provided with a table 5A for managing LUs 9A and 9B, a table 5B for associating information for specifying an initiator capable of using the LUs with key information, and a table 5C for managing an iSCSI_NAME. In the case of ac FC initiator, the tables 5A and 5B are referred to, and in the case of an iSCSI initiator, the tables 5A to 5C are referred to. Different kinds of protocols can be supported by common table groups 5A to 5C. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a storage apparatus and a storage apparatus exclusive control method, and more particularly to a storage apparatus and a storage apparatus exclusive control method capable of accessing a logical volume using different types of protocols.

  The storage device is configured by arranging storage devices such as hard disk drives and semiconductor memory devices in an array. The storage device provides a logical storage area based on, for example, RAID (Redundant Array of Independent Inexpensive Disks) such as RAID1 and RAID5. This logical storage area is also called a logical volume (LU).

  The host computer accesses the logical volume via the communication port of the storage device, and reads / writes data (I / O). The host computer may reserve the logical volume and use the logical volume exclusively in order to maintain data consistency and the like. When the use of the logical volume is finished, the host computer releases the reserved state.

Here, LUN (Logical Unit Number) security is known as a technology that enables only a specific host computer to access a specific logical volume (Patent Documents 1 and 2). In this technology, a WWN (World Wide Name) set in each host computer (host bus adapter) and a LUN are managed in association with each other. When the host computer requests access, it checks whether or not the WWN and the desired LUN are associated with each other, and if they are associated with each other, the access is permitted.
JP 2001-265655 A Japanese Patent Application No. 10-333839

  By the way, as described in the above-mentioned document, a storage area network (SAN) using serial SCSI (Small Computer System Interface) called a fiber channel protocol is known as a storage apparatus. The SAN is a shared storage developed to reduce the dependency on the host computer and to share the storage by a plurality of host computers via a network. In a SAN, a large amount of data can be transferred in units of blocks via optical fibers or metal cables.

  On the other hand, in the field of general-purpose computer communication, a technique for transferring data via an IP (Internet Protocol) network has been developed. In such general-purpose data communication, for example, as is known as the Internet, data is transferred between a plurality of nodes using a TCP / IP (Transmission Control Protocol / Internet Protocol) packet.

  Data transfer can be performed using general-purpose Internet technology, but block-level transfer such as SAN cannot be performed. However, in recent years, iSCSI_SAN has been proposed that enables block-level transfer between remote nodes using Internet technology. In iSCSI (Internet SCSI), a SCSI command or data is wrapped in a TCP / IP packet, and the packet is transferred via an IP network. In contrast to this iSCSI, a conventional SAN is also called a Fiber Channel SAN (FC_SAN). The advent of iSCSI enables the direct connection of storage to IP networks and enables the effective use of existing IP network products such as routers and switches.

  Here, since FC_SAN and iSCSI_SAN have different protocols, there is a problem in, for example, how to specify a host computer (initiator) when performing exclusive control of a logical volume. This is because each communication protocol has a different host computer identification method.

  If the Fiber Channel protocol and iSCSI are simply used together, the host computer identification process for the Fiber Channel protocol and the host computer identification process for the iSCSI are constructed independently of each other. However, storage devices are increasing in capacity and performance year by year, and it is not uncommon to have thousands of logical volumes and connect to hundreds of host computers. Therefore, if an exclusive control system is constructed for each communication protocol, the required data size becomes enormous, and the memory resources of the storage apparatus are compressed.

  The present invention has been made in view of the above-described problems, and one object of the present invention is to provide a storage device and a storage device that can manage the correspondence between a host device and a logical volume in common with different protocols. It is to provide an exclusive control method. One object of the present invention is to provide a storage apparatus and a storage apparatus exclusive control method capable of realizing a logical volume exclusive control common to different protocols by reducing a necessary data size. Further objects of the present invention will become clear from the description of the embodiments described later.

  In order to solve the above problems, the storage apparatus of the present invention includes access management information for use in common with a plurality of different protocols. That is, the storage apparatus according to the present invention is based on the first higher-level interface control unit that controls data exchange with at least one communication port of the first higher-level device based on the first protocol, and on the second protocol. A second higher-level interface control unit that controls data exchange with at least one communication port of at least one second higher-level device; and a lower-level interface control unit that controls data transfer with at least one or more storage devices; A memory unit shared by the first and second upper interface control units and the lower interface control unit, at least one logical volume provided on the storage device, and communication ports of the first and second upper devices An access management information storage unit for storing access management information configured by associating a logical volume with a logical volume; Comprising by referring to the processes management information, first, an access control unit that controls whether to execute the access request to the logical volume from the second upper-level device.

  Here, the first protocol and the second protocol have a common command system and can be configured as mutually related protocols. As one example, for example, the first protocol is a fiber channel protocol and the second protocol is an iSCSI (Internet SCSI) protocol.

  The access management information is configured by associating each communication port of the first and second host devices that perform data transfer based on different first and second protocols with a logical volume. Then, the access control unit refers to access management information common to each protocol, and controls whether or not to permit an access request from a higher-level device. As a result, the required data size can be reduced and the access management structure can be simplified compared to the case where access management information is individually provided for each protocol.

  The access management information identifies first management information in which volume identification information for identifying a logical volume and first port identification information for identifying a communication port of the first higher-level device are associated, and a communication port of the second higher-level device. It can be configured by associating with the second management information storing the second port identification information.

  Here, for example, the first management information and the second management information are stored by storing the pseudo first port identification information having a data structure that is externally common to the first port identification information in the first management information. You may associate. For example, the pseudo first port identification information can be generated based on a rule different from the generation rule of the first port identification information so as to be treated as invalid first port identification information. If the first port identification information is WWN and the second port identification information is iSCSI_NAME, the WWN associated with the second management information is assumed to be a pseudo WWN in order to guide from the first management information to the second management information.

  The pseudo WWN is information that is common to the WWN only in the data structure (for example, data size) on the outer shape and is different in substance. For example, a significant bit is set in a total of 8 bytes of the WWN according to the regular rule, including the upper 4 bytes and the lower 4 bytes. On the other hand, in the pseudo WWN, for example, null data is set in the upper 4 bytes. In the lower 4 bytes of the pseudo WWN, for example, the registration position (entry position, entry number) of iSCSI_NAME in the second management information is set. The pseudo WWN is treated as invalid in terms of WWN interpretation. If the access control unit is configured to refer to the second management information in the case of an invalid WWN, the mechanism of the first management information related to the first protocol is used as it is, and the second protocol is added with a slight configuration. Access management can be performed.

  All or some of the functions, means, and steps constituting the storage apparatus and the exclusive control method for the storage apparatus according to the present invention can be configured by a computer program. This computer program can be distributed in a fixed manner on a storage medium such as a hard disk, a semiconductor memory, or an optical disk. Alternatively, the computer program can be distributed via a communication network such as the Internet.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the present embodiment, as will be described in detail later, a first higher-level interface control unit that controls data exchange with at least one communication port of the first higher-level device based on the first protocol, and a second Based on the protocol, a second higher-level interface control unit that controls data exchange with at least one communication port of the second higher-level device, and a lower-level interface that controls data transfer with at least one or more storage devices A storage apparatus including a control unit, a memory unit shared by the first and second upper interface control units and the lower interface control unit, and at least one logical volume provided on the storage device is used.

  In this storage device, a step of holding common access management information configured by associating each communication port of the first and second host devices with a logical volume, and referring to the held common access management information To determine whether or not the access request to the logical volume can be executed from the first and second host devices, and if execution of the access request is permitted, execute a process according to the content of the access request By executing the above, exclusive control of a logical volume that is common among different protocols is realized.

  FIG. 1 is an explanatory diagram showing the overall concept of the present embodiment. This storage system includes a plurality of host computers (hereinafter referred to as “hosts”) 1A and 1B, and a storage apparatus 3 used by each of the hosts 1A and 1B. The host 1A includes a plurality of communication ports 1C (host bus adapters), and is connected to the storage device 3 via the network 2A (FC_SAN). The host 1B includes a plurality of communication ports 1D (network cards), and is connected to the storage device 3 via the network 2B (iSCSI_SAN). Hereinafter, the communication port is abbreviated as “port”.

  The storage apparatus 3 includes a channel adapter (hereinafter referred to as “CHA”) 4A that controls data exchange with a plurality of ports 1C of the host 1A based on the fiber channel protocol, and a plurality of hosts 1B based on iSCSI. A CHA 4B that controls data exchange with the port 1D of the disk, a plurality of disk adapters (hereinafter referred to as “DKA”) 7 that respectively control data exchange with each disk drive 8, and each CHA 4A, 4B and each DKA 7 A shared memory 5 and a cache memory 6 to be shared, and a plurality of logical volumes (hereinafter referred to as “LU”) 9A and 9B set on a storage area of each disk drive 8 are provided. The CHA 4A includes a plurality of ports 4A1 corresponding to the respective ports 1C, and the CHA 4B includes a plurality of ports 4B1 corresponding to the respective ports 1D.

  Although not shown, the hosts 1A and 1B are connected so that bidirectional data communication is possible via a communication network such as a LAN (Local Area Network), a WAN (Wide Area Network), or the Internet. can do.

  In the shared memory 5, tables 5A, 5B, and 5C for managing access to the logical volumes 9A and 9B are stored in advance. The table 5A stores information for identifying the LUs 9A and 9B such as LUN and LU number, for example, and the table 5B includes ports 1C and 1D such as WWN, for example. Information for identifying each is stored.

  Here, since each port 1C communicates using FC_SAN, it can be uniquely identified by WWN. On the other hand, since each port 1D communicates using iSCSI_SAN, it cannot be specified using WWN.

  Therefore, in this embodiment, iSCSI_NAME is adopted. The iSCSI_NAME is information that can be uniquely identified all over the world, and is information that does not depend on the installation location of the iSCSI node and remains unchanged throughout its lifetime. Two types of iSCSI_NAME are defined. One of them is called IQN, which is an iSCSI Qualified Name using an IP domain name. The other is called EUI, which uses the IEEE EUI-64 Format using the 64-bit Extended Unique Identifier of the Institute of Electrical and Electronic Engineers (IEEE). In either case, the name is unique in the world and does not change throughout the life, and is written in text format.

  However, the data size of WWN is different from the data size of iSCSI_NAME. While WWN is 8 bytes, iSCSI_NAME has a maximum size of 223 bytes according to the standard. It is also conceivable that the data registration size of the table 5B is 223 bytes in accordance with iSCSI_NAME. However, in this case, the size of the table 5B increases and the useless area also increases. This is because in this storage system, FC_SAN and iSCSI_SAN are mixed and not all nodes (initiators) have iSCSI_NAME. In addition, the change in the structure of the table 5B causes a change in other units (tables and programs) corresponding to FC_SAN, which increases the cost.

  Therefore, in this embodiment, a new table 5C is introduced and the table 5C and the table 5B are associated with each other in order to use the existing framework constructed corresponding to FC_SAN without changing as much as possible.

  In other words, a pointer to the table 5C is included in the table 5B for managing the WWN, and the iSCSI_NAME is managed in the table 5C. The pointer to the table 5C registered in the table 5B is described in the form of WWN, but is not treated as a valid WWN because it does not comply with the WWN generation rule. Therefore, when the CHA 4A for FC_SAN refers to the table 5B, there is no possibility of erroneously interpreting the pseudo WWN as a pointer to the table 5C. On the other hand, when the CHA 4B for iSCSI_SAN finds an invalid pseudo WWN by referring to the table 5B, it is guided to this pseudo WWN and refers to the table 5C to detect the target iSCSI_NAME.

  In this way, the storage apparatus 3 can control access to the logical volumes 9A and 9B from the hosts 1A and 1B according to different protocols. Therefore, for example, the host 1A can exclusively use one logical volume 9A, and the host 1B can exclusively use the other logical volume 9B. However, the present invention is not limited to this, and as will be apparent from the embodiments described later, the hosts 1A and 1B can share one or more logical volumes.

  As described above, in this embodiment, an existing access management system constructed for FC_SAN can be used as it is, and can be extended to iSCSI access management with a relatively low cost and simple configuration. Hereinafter, details of the present embodiment will be further described.

  FIG. 2 is a block diagram showing an outline of the overall configuration of the storage system. As will be described later, this system includes a plurality of hosts 10A and 10B and at least one storage device 100.

  Each of the hosts 10A and 10B is realized as, for example, a server, a personal computer, a workstation, or the like. Each of the hosts 10A and 10B is connected to a plurality of client terminals located outside the figure via a communication network (not shown) and provides an information processing service to each client terminal.

  For example, the host 10A can be configured to include an application program (hereinafter referred to as “application”) group 11A, a path management program 12A, and a plurality of host bus adapters (hereinafter referred to as “HBA”) 13A. Similarly, the host 10B can be configured to include an application group 11B, a path management program 12B, and a plurality of HBAs 13B.

  The application groups 11A and 11B are information processing programs that use data stored in the storage device 3, such as various database software. The path management programs 12A and 12B, for example, control the amount of data flowing through the HBAs 13A and 13B (load balance function), and switch communication routes when a failure occurs in either of the HBAs 13A and 13B (failover function). . That is, the path management programs 12A and 12B control the communication path.

  The HBA 13A performs data transfer based on the fiber channel protocol, and the HBA 13B performs data transfer based on iSCSI. As will be described later, the HBA 13B wraps a SCSI command or data in a TCP / IP packet or extracts a SCSI command from the TCP / IP. Thus, since the processing load of the HBA 13B is larger than that of the HBA 13A, it is preferable to install a TCP / IP offload engine (TOE) that specializes in TCP / IP protocol processing. In the figure, one host 10A participating in FC_SAN and one host 10B participating in iSCSI_SAN are shown. However, the present invention is not limited to this, and a plurality of hosts 10A, 10B can be provided.

  The network CN2A that connects the host 10A and the storage apparatus 100 is configured as FC_SAN. The network CN2B that connects the host 10B and the storage apparatus 100 is configured as iSCSI_SAN. In iSCSI_SAN, block level data transfer is performed using an IP network.

  As will be described later, the storage apparatus 100 includes at least one or more Fiber Channel CHAs (hereinafter “CHF”) 110A, at least one iSCSI CHA (hereinafter “CHI”) 110B, The DKA 120 includes a disk drive 130, a cache memory 140, a shared memory 150, a switch unit 160, and a service processor (hereinafter “SVP”) 170.

  The CHF 110A controls data transfer with the host 10A and includes a plurality of ports 111A. The storage apparatus 100 can be provided with a plurality of CHFs 110A. Similarly, the CHI 110B controls data transfer with the host 10B and includes a plurality of ports 111B. The storage apparatus 100 can be provided with a plurality of CHIs 110B. Details of the CHFs 110A and CHI 110B will be described later, but they differ only in terms of protocol processing relating to the external networks CN2A and CN2B, and the processing in the storage apparatus 100 is common.

  Each DKA 120 controls data communication with each disk drive 130. Each DKA 120 and each disk drive 130 are connected via a communication network CN3 such as a SAN, for example, and perform block-unit data transfer according to a fiber channel protocol. The same applies to CHF 110A and CHI 110B, but each DKA 120 includes, for example, a printed circuit board on which a processor, a memory, and the like are mounted, and a control program (both not shown) stored in the memory. Predetermined functions are realized by the cooperative work of the software and the software.

  Each disk drive 130 can be realized as, for example, a hard disk drive (HDD), a semiconductor memory device, an optical disk device, a magneto-optical disk device, or the like. Disk drives with different storage methods may be mixed. Each disk drive 130 is a physical storage device. Although different depending on the RAID configuration, for example, a RAID group 131 is constructed by a set of four disk drives 130, and LUs 132A and 132B, which are virtual storage areas, can be set on the RAID group 131. .

  Note that it is not always necessary that all storage resources of the storage apparatus 100 exist in the storage apparatus 100. For example, the storage apparatus 100 can take in the external storage resource as if it were its own storage resource by mapping the LUN of the storage resource existing outside the storage apparatus 100 to its own LUN or LU. .

  For example, the cache memory 140 stores user data and the like. The cache memory 140 is composed of, for example, a volatile or non-volatile memory. Data written from the hosts 10A and 10B (write data) and data read from the hosts 10A and 10B (read data) are delivered via the cache memory 140.

  The shared memory (or control memory) 150 is composed of, for example, a volatile or nonvolatile memory. For example, control information, management information T, and the like are stored in the shared memory 150. A plurality of shared memories 150 and cache memories 140 can be provided. Further, the cache memory 140 and the shared memory 150 can be mixed and mounted on the same memory board. Alternatively, a part of the memory can be used as a cache area and the other part can be used as a control area.

  The switch unit 160 connects each CHF 110A, CHI 110B, each DKA 120, the cache memory 140, and the shared memory 150 to each other. The switch unit 160 can be configured as, for example, an ultrafast crossbar switch.

  The SVP 170 can be connected to each CHF 110A, CHI 110B, and each DKA 120 via an internal network CN4 such as a LAN, for example. However, the present invention is not limited to this, and the SVP 160 may be connected to only one or both of the CHF 110A and the CHI 110B, and information indicating the internal state may be collected via either the CHF 110A or the CHI 110B. The SVP 170 can be connected to a plurality of management terminals (all not shown) via a communication network such as a LAN, for example, and various types of information collected in the storage apparatus 100 can be used raw or statistically. Can be provided to an external management terminal.

  FIG. 3 is a block diagram showing a schematic configuration of the CHI 110B. The CHI 110B includes, for example, an iSCSI port control unit 111B (simply described as “port” in FIG. 2), a channel processor (hereinafter “CHP”) 112, a local memory 113, a microprogram adapter (hereinafter “ MPA ") 114 and a data transfer adapter (hereinafter" DTA ") 115. In the figure, only the configuration related to one iSCSI port is shown, but a plurality of ports can actually be provided.

  The iSCSI port control unit 111B is in charge of conversion processing between TCP / IP as an external protocol used in the external network CN2B and a Fiber Channel protocol as an internal protocol used for data transfer inside the storage apparatus 100. .

  The CHP 112 performs overall control of the CHI 110B, and processes SCSI commands such as a read command and a write command, for example. The SCSI command extracted from the TCP / IP packet by the iSCSI port control unit 111B is stored in the shared memory 150 from the CHP 112 via the MPA 114. The DKA 120 refers to the shared memory 150 as needed, and when it finds an unprocessed command, it processes the command. The processing result by the DKA 120 is reported from the shared memory 150 to the CHP 112 via the MPA 114.

  The write data extracted from the TCP / IP packet by the iSCSI port control unit 111B is stored in the cache memory 140 from the CHP 112 via the DTA 115. Read data read from the disk drive 130 by the DKA 120 is stored in the cache memory 140. This read data is read into the CHP 112 from the cache memory 140 via the DTA 115.

  The local memory 113 is composed of, for example, a nonvolatile memory. The local memory 113 stores management information and control information used by the iSCSI port control unit 111B or the CHP 112. For example, the local memory 113 can store an index table T1B, a KEY table T2B, an iSCSI_NAME table T3B, a resource table T4B, and the like. Here, the index table T1B is a copy of a part of an index table T1 described later. Similarly, the KEY table T2B is a copy of a necessary part of the KEY table T2, and the iSCSI_NAME table T3B is a copy of the iSCSI_NAME table T3.

  That is, of the tables T1 to T3 stored in the shared memory 150, only data that can be used by the CHI 110B is copied to the local memory 113. The copy data in the local memory 113 is updated at an appropriate timing. Note that the resource table T4B is a table for managing resources allocated to the HBA 13B.

  FIG. 4 is a block diagram showing a schematic configuration of the CHF 110A. As described with reference to FIG. 3, the description will focus on the configuration related to one port in the figure, but in practice, a plurality of ports can be provided.

The CHF 110A includes an FC port control unit 111A, a CHP 112, a local memory 113, an MPA 114, and a DTA 115. Comparing CHF 110A and CHI 110B, in CHF 110A, FC port control unit 111A is provided instead of iSCSI port control unit 111B. The FC port control unit 111A controls the fiber channel protocol. In the local memory 113 of the CHF 110A, an index table T1A, a KEY table T2A, a WWN table T3A, and a resource table T4A are stored in the same manner as that of the CHI 110B. Here, the tables T1A, T2A, and T4A are the same as the tables T1B, T2B, and T4B described above, and thus the description thereof is omitted.
The WWN table T3A is not a part of the iSCSI_NAME table T3, but is a table used by each CHF 110A to manage the WWN of each FC initiator. The WWN table T3A can be configured, for example, by associating WWN, S_ID, LUN, and the like. It can also be generated with other structures.

  Next, both Fiber Channel and iSCSI protocols will be briefly described. First, FIG. 5 is an explanatory diagram showing the layer configuration of the fiber channel. The fiber channel is composed of an FC-0 layer, an FC-1 layer, an FC-2 layer, and an FC-4 layer in order from the bottom. The FC-3 layer defines a common service, but has not been standardized yet.

  The FC-0 layer defines physical media such as optical fiber cables, metal cables, connectors, transmission speed, distance, signal processing, and the like.

  The FC-1 layer defines data encoding / decoding and the like as known as the 8B / 10B encoding system.

  The FC-2 layer defines frame assembly, flow control, data transmission procedure, and the like, which are the minimum transmission units. The flow control is defined by, for example, service classes classified into classes 1, 2, 3, 4, 5, 6, F and the like. Class 2, class 3, and class F are relatively frequently used. A few classes are briefly described.

  Class 1 is a service that can use the entire physical bandwidth between two ports, and has a maximum data transfer amount compared to other classes. In class 2, two ports perform data transmission without occupying a physical band. Another port can be connected in class 2 to any one of the two ports connected in class 2. In class 2, data reception confirmation and content confirmation (ACK) are performed in frame units.

  On the other hand, unlike class 2, class 3 confirms the reception of a frame but does not confirm the contents. In class 3, transmission error processing is left to the upper layer (FC-4). In principle, the communication quality of class 3 is lower than that of class 2. However, since the fiber channel employs the 8B / 10B encoding method in the FC-1 layer and keeps the bit error rate (BER) low, there is no practical problem. In class 3, data transmission at a higher speed than class 2 can be realized by omitting the confirmation by the ACK frame.

  Class F is used to transmit data for controlling the fabric topology. When fabric switches (FC switches) are connected to each other via an E port (Expansion Port), data transmission between the E ports conforms to class F.

  The FC-3 layer is a layer that defines fabric services such as name service and multicast.

  The FC-4 layer is a layer in charge of mapping to higher-level protocols such as FCP (Fibre Channel Protocol), IP, and FICON (Fibre Connect).

  As shown in FIG. 5, in the case of the fabric topology, a fabric switch is interposed between the host and the storage apparatus, and each N port (Node Port) is connected in the FC-0 layer. Since each FC-0 layer is independent, data transfer can be performed even if one is an optical fiber cable and the other is a metal cable.

  FIG. 6 is an explanatory diagram showing a data structure. One sequence 200 is composed of a plurality of frames 210 performing a series of work. Although not shown, one exchange is constituted by a group of sequences 200. For example, one read operation for reading a series of data from the storage apparatus 100 corresponds to one exchange.

  The frame 210 is a minimum transmission unit of the fiber channel. One frame 210 includes an SOF (Start of Frame) 211 indicating the start of a frame, a header 212, a data field 213, a CRC (Cyclic Redundancy Check) 214, and an EOF (End of Frame) 215 indicating the end of the frame. Composed. Here, the SOF is 4 bytes, the header is 24 bytes, the data field 213 has a variable length, but the maximum value is 2112 bytes, the CRC is 4 bytes, and the EOF is 4 bytes.

  The header 212 includes, for example, a destination address D_ID (Destination_ID) 212A, a source address S_ID (Source_ID) 212B, an F_CTL (Frame Control) 212C, a SEQ_ID (Sequence_ID) 212D, and a SEQ_CNT (Sequence_Count) 212E etc. are included. In addition, the exchange ID and the like are also included in the header 212, but are omitted.

  FIG. 7 is an explanatory diagram showing an iSCSI layer structure. The physical layer, data link layer, IP layer, TCP layer, iSCSI layer, SCSI layer, and SCSI application layer are stacked in order from the bottom.

  In the IP layer, data transfer specifying an IP address is performed, and in the TCP layer, data transfer specifying a TCP port number is performed. In the iSCSI layer, data transfer specifying iSCSI_NAME is performed, and in the SCSI layer, data transfer specifying LU number and LBA (Logical Block Address) is performed. The iSCSI layer is located between the SCSI layer and the TCP layer. The SCSI command and the SCSI response are stored in a capsule called iSCSI_PDU, and data is transferred through the TCP connection.

  FIG. 8 is an explanatory diagram showing a data structure used in iSCSI. The command output from the SCSI application layer of the host through the SCSI layer is a command frame 340 including a command and data (or only a command) as shown in FIG. The command frame 340 is a 6-byte command frame in which an operation code such as a write command or a read command is included in the first byte.

  When the command frame 340 reaches the iSCSI layer, the iSCSI layer converts the SCSI command frame into an iSCSI_PDU (Protocol Data Unit) 330. When this PDU 330 passes through the TCP layer, it is converted into a TCP packet 320. Further, this TCP packet 320 becomes an IP packet 310 by passing through the IP layer. Then, a MAC (Media Access Control) address is added to complete the variable length MAC frame 300. The data field can store 1500 bytes of data.

  The SCSI command frame 340 includes, for example, a basic header segment 341, an additional header segment 342, and a data segment 343. The basic header segment 341 can include, for example, an operation code 341A, a data segment length 341B, a LUN number 341C, an initiator task tag 341D, and an operation code specification field 341E.

  Under the protocol configuration shown in the figure, the physical layer and the data link layer, the IP layers, the TCP layers, the iSCSI layers, the SCSI layers, and the SCSI application layers sequentially session with each other to output the I The storage apparatus 100 can process the / O request.

  For example, in a session between the physical layer and the data link layer, an ARP (Address Resolution Protocol) request and an ARP response to the ARP are performed to acquire each other's MAC address. In a session between IP layers, a ping request and a ping response to the ping request are performed to confirm whether or not the other party (IP address) exists.

  In a session between TCP layers, three packets for synchronizing sequence numbers are exchanged. In a session between iSCSI layers, an iSCSI connection is established by a login phase in which a login request and a login response are exchanged (for example, the IP address, the iSCSI_NAME of the login request source host, and the TCP port number are used). . In a session between SCSI layers, for example, a SCSI command such as a read command or a write command is transmitted from the host to the storage apparatus 100. In a session between SCSI application layers, write data is transmitted from the host to the storage apparatus 100, or read data is transmitted from the storage apparatus 100 to the host.

  Next, the structure of each table for realizing access control (exclusive control) to the logical volume will be described with reference to FIGS.

  FIG. 9 shows an index table T1, a KEY table T2, and an iSCSI_NAME table T3. The index table T1 manages, for example, whether for each LUN number, the HBA that can access the LUN is restricted. When “1” is set, it indicates that only a specific HBA (initiator) can access the LUN.

  Here, a plurality of columns of flag bits are provided for each LUN number. The position of each column corresponds to the entry position of the KEY table T2. That is, the leftmost bit string of the index table T1 corresponds to the top entry of the KEY table T2, and similarly, the entry position of the corresponding KEY table T2 is advanced as the bit string of the index table T1 moves to the right. Will go down.

  Therefore, when a flag is set on the LUN number (when set to “1”), the position of the flag directly indicates the corresponding entry in the KEY table T2. In this way, the index table T1 and the KEY table T2 are associated with each other.

  In the KEY table T2, key information (such as “KEY1” in the figure) used for reservation (exclusive use) of the LU is associated with a WWN for specifying an HBA that exclusively uses the LU. ing.

  Here, when the HBA that accesses the LU is the HBA 13A according to the fiber channel, the WWN and key information of the HBA 13A are registered at the entry position corresponding to the LU. On the other hand, if the HBA that accesses the LU is an HBA 13B that conforms to iSCSI, the pseudo WWN and key information that are indirectly associated with the HBA 13B are registered at the entry position corresponding to the LU. In the example shown in FIG. 9, WWN1 and WWN4 are pseudo WWNs.

  Pseudo WWN is defined to be WWN-like information that is set to an invalid value that does not conform to the regular WWN generation rules, although the structure of the WWN is equal to the external structure and has a data size of 8 bytes. it can. Specifically, for example, information similar to WWN is obtained by setting null data in the upper 4 bytes. The entry position in the iSCSI_NAME table T3 is set in the lower 4 bytes of the pseudo WWN. That is, in the illustrated example, a value indicating the first entry position of iSCSI_NAME T3 is set in the lower 4 bytes of the pseudo WWN1. Similarly, a value indicating the fourth entry position of the iSCSI_NAME table T3 is set in the lower 4 bytes of the pseudo WWN4. Here, as is clear from comparison between the KEY table T2 and the iSCSI_NAME table T3, the entry positions of the tables T2 and T3 correspond to each other. That is, the iSCSI_NAME corresponding to the pseudo WWN1 set at the top entry position of the KEY table T2 is registered at the top entry position of the iSCSI_NAME table T3. Similarly, iSCSI_NAME corresponding to the pseudo WWN4 set at the fourth entry position from the top of the KEY table T2 is set at the fourth entry position of the iSCSI_NAME table T3. Thus, by associating the entry positions of the KEY table T2 and the iSCSI_NAME table T3, the tables T2 and T3 are associated with each other.

  The index table T1 and the KEY table T2 are provided for controlling access to LUs in FC_SAN. In the present embodiment, an iSCSI_NAME table T3 is introduced to correspond to both FC_SAN and iSCSI_SAN, and this iSCSI_NAME table T3 is associated with the KEY table T2. As a result, it is possible to support iSCSI_SAN without substantially changing the access control configuration for FC_SAN.

  FIG. 10A shows the resource table T4. The resource table T4 is configured, for example, by associating a resource management number with an initiator that uses the resource. In the figure, iSCSI_NAME is shown as an initiator, but in the case of FC_SAN, WWN is used.

  FIG. 10B shows a table for managing the association between LUNs and LUs. This LUN-LU table T5 can be configured, for example, by associating LUN numbers with LU numbers. Each LU is a resource LU (or also called “LDEV”) on the storage apparatus 100 side. Each LU basically has a one-to-one correspondence with a LUN. A virtual LU can also be constructed by connecting a plurality of LUs.

  FIG. 11 is a table for managing access modes. FIG. 11A shows an access type management table T6A for FC_SAN. In this table T6A, for example, for each WWN that specifies each initiator, the access type permitted for each initiator is associated. As the access type, for example, both read / write and read only can be mentioned. Similarly, FIG. 11B shows an access type management table T6B for iSCSI_SAN. In this table T6B, iSCSI_NAME is used instead of WWN.

  FIG. 12 shows a configuration table T7 of control information (LDCB) for managing the LU status. The LDCB (T7) includes, for example, a serial number, a port number, a LUN number, an LU number, a flag indicating a reserved state, path information for which a reservation is set, a flag indicating a persistent reserved state, Corresponding path information for which persistent reserve is set, ACA (Automatic Contingent Allegiance) status flag, ACA status path information, and UA (Unit Attention) status flag Can be configured.

  Here, the ACA state is a state set in order to prohibit the use of an access path when a failure occurs in the access path, for example. The ACA state can be released by a release command from the host that has set the ACA state. The UA state is a state set immediately after the storage apparatus 100 is started, for example.

  The persistent reserved state is a state in which one or a plurality of LUs are reserved through a plurality of preset access paths. In a normal reserved state, a predetermined LU can be accessed from only one access path. In the persistent reserve state, a predetermined LU can be accessed from each of a plurality of access paths. For example, the hosts 10A and 10B can use all or part of the plurality of HBAs 13A and 13B included in each of them as a persistent group. The HBAs constituting the group can access a common LU via different paths. As a result, failover and load balancing within the host can be realized. The path management programs 12A and 12B execute the persistent reserve setting (issue of the PGR command).

  Thus, when persistent reserve is set, the LU can be shared by a plurality of predetermined initiator groups. A command for setting persistent reserve is referred to herein as a persistent group reserve command (PGR command). An area related to persistent reserve in the LDCB is called a PGR area 400.

  FIG. 13 is an explanatory diagram showing the data structure of the PGR area 400 for iSCSI_SAN. The PGR area 400 has a width of 4 bytes, and includes, for example, a generation counter 410, a laser navigation key 420, a type code 431, an APTPL bit 432, a CHI bit 433, a reserved field 434, and a sleep flag 440. , A reserved field 450, an absolute port number 461, an entry number 462, and an ACA release flag 470.

  Here, the CHI bit 433 is information newly provided in the former spare field in order to support iSCSI_SAN. When “1” is set in the CHI bit 433 (when ON), the fields 461 and 462 indicate that information indicating the reference position to the LDCB iSCSI_NAME table T8 shown in FIG. 15 is stored. . When the CHI bit 433 is in the OFF state, it indicates that the WWN as usual is set in the fields 450, 461, and 462.

  Reference is first made to FIG. FIG. 14 shows a PGR area 400 for FC_SAN. As is clear from comparison between FIG. 13 and FIG. 14, the FC_SAN PGR area and the iSCSI_SAN PGR area differ in the following three points. The first difference is that in the FC_SAN PGR area, the CHI bit 433 is set to an off state. The second difference is that in the FC_SAN PGR area, the upper 4 bytes of the WWN are set in the reserved field 450 in FIG. The third difference is that in the FC_SAN PGR area, the lower 4 bytes of the WWN are set instead of the absolute port number 461 and the entry number 462 in FIG.

  In other words, the structure shown in FIG. 14 is a normal form. In this embodiment, in order to support iSCSI_SAN, a CHI bit 433 is newly added to the unused field, and the upper 4 bytes of the WWN are reserved fields. 4 bytes of null data is set, and pointer information (2-byte absolute port number and 2-byte entry number) to the iSCSI_NAME table T8 for LDCB is stored using the lower 4 bytes of the WWN.

  When CHI 110B sets information in the PGR area, null data is set in the upper 4 bytes of WWN. Therefore, when the CHF 110A refers to this PGR area, it is recognized as an invalid WWN.

  FIG. 15 shows an iSCSI_NAME table T8 for LDCB. In this table T8, for example, every iSCSI_NAME that may belong to the port is associated with each absolute port number. Therefore, it is possible to understand where to refer to the table T8 by simply designating the absolute port number and the entry position at the port number. For example, if the HBA 13B for which the persistent reserve is set is iSCSI_NAME [2] using the absolute port number 1, “1” as the absolute port number, “3” as the entry number, and the fields 461 and 462, respectively. By setting to, the PGR area 400 can be associated with the table T8.

  The tables T1 to T8 described above are stored in the shared memory 150. In each table T1 to T8 and the like, data in a predetermined range required by each CHF 110A and each CHI 110B is copied to the local memory 113 of each CHF 110A and each CHI 110B.

  The operation of this system will be described with reference to FIGS. FIG. 16 is a flowchart showing the processing when the persistent reserve is set / changed / deleted by the CHI 110B. This process is mainly executed by the CHP 112 of the CHI 110B.

  The path management program 12B of the host 10B issues a PGR command by designating each of the HBAs constituting the persistent group, the target LU (LUN), and the key information to be used. As already described, since a LUN and an LU are associated with each other, designating a LUN means designation of an LU associated with the LUN. When the CHI 110B receives the PGR command (S11), the CHI 110B refers to the index table T1 using the designated LUN number as a search key (S12).

  The CHI 110B determines whether or not a flag bit indicating that the PGR has been set is set for the designated LUN (S13). When the flag is set for the target LUN (S13: YES), the CHI 110B refers to the corresponding location in the KEY table T2 based on the set position of the flag, and sets the key information ("KEY" ]) Is acquired (S14). Further, the CHI 110B refers to the iSCSI_NAME table T3 based on the entry position of the pseudo WWN registered in the KEY table T2, and acquires the set iSCSI_NAME (S15).

  In CHI110B, the combination of the key information specified from the PGR command issuer host and the iSCSI_NAME that identifies the command issuer matches the combination of the key information registered in the KEY table T2 and the iSCSI_NAME registered in the iSCSI_NAME table T3. It is determined whether or not to perform (S17).

  When the combination of the key information and iSCSI_NAME matches (S17: YES), the CHI 110B sets a flag in the index table T1 for the LUN designated as the PGR target (S18). The CHI 110B registers the key information designated by the host in the KEY table T2 at the entry position corresponding to the flag set position in the index table T1 (S19). Also, the CHI 110B generates a pseudo WWN, associates it with the key information, and registers it in the KEY table T2 (S19). The lower 4 bytes of the pseudo WWN indicate the reference position of the iSCSI_NAME table T3. The CHI 110B registers the iSCSI_NAME that uniquely identifies the PGR command issuer at a predetermined position in the iSCSI_NAME table (S21).

  On the other hand, if it is determined in S13 that the flag is not set in the index table T1 for the designated LUN (S13: NO), S14 to S16 are skipped, and the process proceeds to S18. In S17, if the combination of the key information and iSCSI_NAME does not match (S17: NO), the CHI 110B notifies the host that issued the command that the PGR command cannot be executed (S22).

  FIG. 17 is a flowchart showing processing when setting / changing / deleting persistent reserve by the CHF 110A. This process is mainly executed by the CHP 112 of the CHA 110A.

  In this flowchart, S31 to S34, S36 to S39, S41, and S42 correspond to S11 to S14, S16 to S19, S21, and S22 in FIG. 16 described above, respectively, and perform substantially the same processing.

  However, in S36, the combination of key information and WWN is compared instead of the combination of key information and iSCSI_NAME. In the flowchart of FIG. 17, steps corresponding to steps S25 and S40 for handling the iSCSI_NAME table T3 are omitted.

  That is, in this embodiment, common PGR processing is possible in both FC_SAN and iSCSI_SAN, but the iSCSI_NAME table T3 newly added for this purpose is exclusively used only for the CHI 110B, and is basically CHF 110A. Does not need to be referenced. Therefore, it is possible to support iSCSI without substantially changing the control logic of the CHF 110A.

  If CHI110B has previously set PGR for the same LUN, pseudo WWN is registered in KEY table T2, but this pseudo WWN is an invalid value with null data set in the upper 4 bytes. It is. Therefore, if the WWN of the host that issued the PGR command is compared with the pseudo WWN, there is always a mismatch.

  When releasing persistent reserve, a release command is issued. When the release command is issued, as described in FIGS. 16 and 17, it is determined whether or not the combination of key information and iSCSI_NAME or the combination of key information and WWN matches. The part related to PGR is deleted from the table. Specifically, in the case of iSCSI, from each table T1 to T3, T7 (PGR area 400), PGR set flag (T1), key information and pseudo WWN (T2), iSCSI_NAME (T3), CHI bit, absolute port number And the entry number (PGR area 400) are deleted. On the other hand, in the case of Fiber Channel, the PGR set flag (T1), key information, WWN (T3), and WWN (PGR area 400) are deleted from each table T1, T2, T7 (PGR area 400). .

  Whether or not an access request can be made to an LU for which a PGR is set is determined whether or not the combination of key information and iSCSI_NAME, or the combination of key information and WWN matches, as described in FIGS. Access is allowed only when Then, the access to the LU is processed against the access type such as read only.

  FIG. 18 is a flowchart of access processing executed by the CHI 110B. S51 to S57 in FIG. 18 correspond to S11 to S17 in FIG.

  When it is determined that the combination of the key information and iSCSI_NAME matches (S57: YES), the CHI 110B refers to the access type management table T6B (S58) and determines whether or not the requested command processing is permitted. (S59). In the case of an authorized command (S59: YES), the CHI 110B executes processing corresponding to the command and responds with completion of command processing (S60).

  On the other hand, if the command is not allowed (S59: NO), or if the combination of key information and iSCSI_NAME does not match those already registered (S57: NO), the requested command cannot be processed. The host is notified (S61).

  FIG. 19 is a flowchart of access processing executed by the CHF 110A. In this flowchart, S71 to S74 and S76 to S81 correspond to S51 to S54 and S56 to S61 in FIG. In the process shown in FIG. 19, the step corresponding to S55 is unnecessary and is abolished, and the combination of key information and WWN is determined (S76).

  Since the present embodiment is configured as described above, the following effects can be obtained. In this embodiment, the access management information T1 to T3 is generated by associating the HBAs 13A and 13B of the hosts 10A and 10B with different protocols to be used with the LUN numbers for specifying the LUs, and the access management information T1 to T3. In order to provide an access control unit (a microprogram executed by the CHP 112) for determining whether or not each HBA 13A, 13B can access a desired LU, the existing system for FC is significantly changed. It can also support iSCSI.

  In the present embodiment, the pseudo WWN which is the pseudo first port identification information is generated as a unique value based on a rule different from the normal WWN generation rule so that it is handled as an invalid WWN, and is stored in the KEY table T2. sign up. Accordingly, when comparing the command issuing source WWN with the pseudo WWN registered in the table T2, the CHF 110A always determines that they do not match. This makes it possible to support iSCSI without substantially changing the control logic of the CHF 110A.

  A second embodiment as a modification of the first embodiment will be described with reference to FIGS. In this embodiment, it is determined whether the initiator is an FC initiator or an iSCSI initiator based on information registered in the KEY table T2 or the LDCB (T7).

  For example, as shown in FIG. 20, either or both of CHF 110A and CHI 110B (hereinafter referred to as “CHF etc.”) obtain WWN by referring to KEY table T2 (S91), and this WWN is a valid value. It is determined whether or not (S92). CHF or the like is determined to be FC when it is a valid WWN (S92: YES) (S93), and is determined to be iSCSI when it is an invalid WWN (S92: NO) (S94).

  For example, as shown in the flowchart of FIG. 21, the CHF or the like refers to the LDCB to acquire the CHI bit (S101), and determines whether the CHI bit is set to the on state (S102). When the CHI bit is on (S102: YES), CHF and the like are determined to be iSCSI (S103), and when the CHI bit is off (S102: NO), it is determined to be FC (S104).

  In this manner, the type of initiator can be determined by including determination information for indicating whether or not iSCSI is included in the KEY table T2 or LDCB (T7).

  22 and 23 show an embodiment in which a cluster system is configured by the hosts 10A and 10B.

  As shown in the block diagram of FIG. 22, the path management programs 12A and 12B of the hosts 10A and 10B are respectively provided with cluster controllers 14A and 14B (hereinafter referred to as “cluster controller 14” unless otherwise distinguished). ing. Each of these cluster control units 14 performs heartbeat communication for monitoring the other party between the hosts 10A and 10B. Each of the hosts 10A and 10B is connected to a common LU 132 via a plurality of HBAs.

  FIG. 23 is a schematic flowchart for executing failover across hosts based on a heterogeneous protocol. First, the cluster control unit 14 monitors whether or not file over is executed, and determines whether or not the failover execution time has come (S111).

  For example, failover is executed when heartbeat communication with the counterpart node is interrupted for a predetermined time or more, or when execution of file over is explicitly requested from the counterpart node (planned stop).

  The failover destination cluster control unit inherits the network setting information of the counterpart node (failover source node) (S112), and impersonates the counterpart node. In addition, the cluster control unit 14 secures the right to use the shared volume 132 (S113) and exclusively uses the volume 132 (S114). Then, the file over destination node (either 10A or 10B) resumes the provision of the information processing service (S115).

  According to this embodiment, in addition to failover between the HBAs in the hosts 10A and 10B, failover to a host using a different type of protocol can be realized.

  A fourth embodiment will be described with reference to FIG. In the previous embodiment, the failover across the hosts has been described. In contrast, in this embodiment, a case will be described in which HBAs based on different types of protocols are provided in the same host, and failover is performed between HBAs that support these different protocols.

  FIG. 24 is a schematic block diagram showing the overall configuration of the storage system. Of each HBA 13A of one host 10A, the first to third HBAs are connected to FC_SAN, and the fourth HBA is connected to iSCSI_SAN. Similarly, in the other host 10B, among the HBAs 13B, the second HBA to the fourth HBA are connected to iSCSI_SAN, and the first HBA is connected to FC_SAN. Which HBA is connected to which SAN is arbitrary. Further, although an example in which different types of HBAs are mounted in the respective hosts 10A and 10B has been shown, a plurality of FC HBAs and a plurality of iSCSI HBAs can be provided in the same host.

  As illustrated, the path management program 12A includes a cluster control unit 15A and a path control unit 16A. Similarly, the path management program 12B includes a cluster control unit 15B and a path control unit 16B. Each path management program 12A, 12B switches the path (HBA) used for data transfer when, for example, a failure or an overload condition occurs in any of the paths under its management. This path switching is executed by the path controllers 16A and 16B.

  For example, when the host 10A transmits / receives data to / from the storage apparatus 100 via the FC_SAN from the first HBA, it is assumed that some failure or the like has occurred in this path. In this case, the path management program 12A of the host 10A can switch to any one or a plurality of paths of the second to fourth HBAs. For example, when the path is switched from the first HBA to the fourth HBA, the host 10A and the storage apparatus 100 perform data transfer via iSCSI_SAN instead of FC_SAN. The same applies to the host 10B.

  Failover across hosts can be implemented based on the interruption of heartbeat communication as described above, but in the case of failover between paths within the same host, the path management programs 12A and 12B send a response at the SCSI layer. By checking, occurrence of a failure or the like can be detected, and path switching can be performed. This is because the hosts 10A and 10B are recognized as SCSI buses regardless of whether the host-storage protocol is FC or iSCSI.

  In this way, initiators based on different protocols can be mixed in the same host, and failover between initiators based on these different protocols can be executed manually or automatically.

  The present invention is not limited to the above-described embodiment. A person skilled in the art can make various additions and changes within the scope of the present invention.

It is explanatory drawing which shows the whole concept of embodiment of this invention. 1 is a block diagram showing an overall configuration of a storage system. It is a block diagram which shows schematic structure of the channel adapter for iSCSI. It is a block diagram which shows schematic structure of the channel adapter for fiber channels. It is explanatory drawing which shows the layer structure of a fiber channel. It is explanatory drawing which shows the frame structure of a fiber channel. It is explanatory drawing which shows the layer structure of iSCSI. It is explanatory drawing which shows the frame structure of iSCSI. FIG. 4 is a diagram illustrating a management table for detecting an initiator that reserves an LU, in which (a) is an index table, (b) is a KEY table, and (c) is a configuration example of an iSCSI_NAME table. (A) is a resource table, (b) is explanatory drawing which shows the example of a structure of a LUN-LU corresponding | compatible management table, respectively. (A) is an explanatory view showing an access type management table of a fiber channel initiator, and (b) is an explanatory view showing an access type management table of an iSCSI initiator. It is explanatory drawing which shows the structural example of LDCB. It is explanatory drawing which shows the mode of LDCB when a persistent reserve is set by the iSCSI initiator. It is explanatory drawing which shows the mode of LDCB when a persistent reserve is set by the fiber channel initiator. It is explanatory drawing which shows the iSCSI_NAME table for LDCB. It is a flowchart in the case of setting persistent reserve by the iSCSI initiator. It is a flowchart in the case of setting persistent reserve by a fiber channel initiator. It is a flowchart which shows the access process by an iSCSI initiator. It is a flowchart which shows the access process by a fiber channel initiator. It is a flowchart which discriminate | determines Fiber Channel and iSCSI. It is another flowchart which discriminate | determines Fiber Channel and iSCSI. 1 is an overall configuration diagram of a storage system that constructs a cluster system with different types of initiators. It is a flowchart which shows the outline of a failover process. 1 is an overall configuration diagram of a storage system capable of switching paths based on different types of protocols within the same host.

Explanation of symbols

1A, 1B ... Host, 1C, 1D ... Port, 2A ... FC_SAN, 2B ... iSCSI_SAN, 3 ... Storage device, 4A, 4B ... Channel adapter, 4A1, 4B1 ... Port, 5 ... Shared memory, 5A, 5B, 5C ... Table , 6 ... Cache memory, 7 ... Disk adapter, 8 ... Disk drive, 9A, 9B ... Logical volume, 10A, 10B ... Host, 11A, 11B ... Application group, 12A, 12B ... Path management program, 13A, 13B ... Host bus Adapter, 14A, 14B, 15A, 15B ... Cluster control unit, 16A, 16B ... Path control unit, 100 ... Storage device, 111A, 111B ... Port (port control unit), 112 ... Channel processor, 113 ... Local memory, 114 ... Microprogram adapter, 115 ... Data transfer adapter , 120 ... disk adapter, 130 ... disk drive, 131 ... RAID group, 132, 132A, 132B ... logical volume, 140 ... cache memory, 150 ... shared memory, 160 ... switch part, T1 ... index table, T2 ... KEY table, T3 ... iSCSI_NAME table, T4 ... resource table, T5 ... LUN-LU correspondence management table, T6A, T6B ... access type management table, T7 ... LDCB, 400 ... PGR area, T8 ... iSCSI_NAME table for LDCB

Claims (16)

A first host interface control unit that controls data exchange with at least one communication port of the first host device based on the first protocol;
A second higher-level interface controller that controls data exchange with at least one communication port of the second higher-level device based on the second protocol;
A lower interface control unit that controls data exchange with at least one storage device;
A memory unit shared by the first and second upper interface control units and the lower interface control unit;
At least one logical volume provided on the storage device;
An access management information storage unit for storing access management information configured by associating each communication port of the first and second host devices with the logical volume;
An access control unit that controls whether or not an access request to the logical volume can be executed from the first and second host devices by referring to the access management information;
A storage device with   The access management information includes first management information in which volume identification information for identifying the logical volume is associated with first port identification information for identifying the communication port of the first higher-level device, and the second higher-level device. The storage apparatus according to claim 1, wherein the storage apparatus is configured by associating with second management information storing second port identification information for identifying the communication port.   By storing pseudo first port identification information having a data structure that is externally common to the first port identification information in the first management information, the first management information and the second management information correspond to each other. The storage device according to claim 2 attached.   The pseudo first port identification information is generated as a unique value based on a rule different from the generation rule of the first port identification information so that the pseudo first port identification information is treated as the invalid first port identification information. The storage device described.   The storage apparatus according to claim 3, wherein the second port identification information is registered in the second management information in accordance with a registration position of the pseudo first port identification information in the first management information. The access management information includes a first management table, a second management table associated with the first management table, and a third management table associated with the second management table.
The first management table stores volume identification information for identifying the logical volume in association with information indicating a reference position of the second management table;
The second management table stores first port identification information for identifying the communication port of the first higher-level device in association with key information for exclusive control,
The third management table stores second port identification information for identifying the communication port of the second host device,
In accordance with the registration order of the first port identification information in the second management table, register the second port identification information in the third management table,
The storage apparatus according to claim 1, wherein the second management table includes information that can be used to determine whether to refer to the third management table. The access management information includes upper layer side access management information and lower layer side access management information,
The upper layer access management information includes a first management table, a second management table associated with the first management table, and a third management table associated with the second management table.
The first management table stores volume identification information for identifying the logical volume in association with information indicating a reference position of the second management table;
The second management table stores first port identification information for identifying the communication port of the first higher-level device in association with key information for exclusive control,
The third management table stores second port identification information for identifying the communication port of the second host device,
In accordance with the registration order of the first port identification information in the second management table, the previous second port identification information is registered in the third management table,
And when referring to the third management table, information indicating the reference position of the third management table is stored in the second management table in a form that is treated as invalid first port identification information,
The lower layer access management information includes a fourth management table and a fifth management table associated with the fourth management table.
The fourth management table includes reference determination information for designating whether or not the fifth management table should be referred to, and any one of the first port identification information or information indicating the reference position of the fifth management table. It is configured in association with
The fifth management table includes the first port identification information of the first and second host devices that can use each communication port for each communication port of the first and second host interface control units, or the first It is configured by associating either one of the 2-port identification information,
The storage apparatus according to claim 1.   The storage apparatus according to claim 1, wherein the access management information storage unit is provided in the memory unit, and the access control unit is provided in the first and second upper interface control units. The first and second upper interface control units each include a microprocessor, a protocol control unit, and a local memory.
The entire access management information storage unit is provided in the memory unit, and a predetermined part of the access management information associated with itself is stored in the local memory, respectively.
The storage apparatus according to claim 1, wherein the access control unit is realized by the microprocessor executing a predetermined program code.   The storage apparatus according to claim 1, wherein each of the communication ports of the first and second host apparatuses can access a common logical volume via the access control unit.   The storage apparatus according to claim 1, wherein the first protocol and the second protocol have a common command system and are configured as mutually related protocols.   The storage apparatus according to claim 1, wherein the first protocol is a fiber channel protocol, and the second protocol is an iSCSI (Internet SCSI) protocol.   The storage apparatus according to claim 1, wherein the first and second host devices form a cluster.   2. The storage apparatus according to claim 1, wherein at least one or both of the first and second host apparatuses include a plurality of the communication ports, and the communication ports can be switched within the same apparatus. .   The first and second host devices include a plurality of communication ports, and a path management unit that manages paths between the communication ports and the communication ports of the first and second host interface control units. The storage apparatus according to claim 1, wherein the path management unit designates at least one predetermined communication port and requests exclusive use of the logical volume. A first host interface control unit that controls data exchange with at least one communication port of the first host device based on the first protocol;
A second higher-level interface controller that controls data exchange with at least one communication port of the second higher-level device based on the second protocol;
A lower interface control unit that controls data exchange with at least one storage device;
A memory unit shared by the first and second upper interface control units and the lower interface control unit;
A method for performing exclusive control of the logical volume in a storage device comprising at least one logical volume provided on the storage device,
Holding common access management information configured by associating port identification information for specifying the communication ports of the first and second host devices with volume identification information for specifying the logical volume When,
Determining whether or not to execute an access request from the first and second host devices to the logical volume by referring to the held common access management information;
If execution of the access request is permitted, executing a process according to the content of the access request;
Storage device exclusive control method.

Images