JP2008097112A - Storage system, and logical volume management method applied to the system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the information amount that each node must hold. <P>SOLUTION: An LPMT 122-i (i=1 to 3) of the node 10-i holds IP addresses assigned to logical extents out of a plurality of logical extents constructing a logical volume which are stored in physical extents in physical volumes of the node 10-i, in association with the physical extents. According to the location in the logical volume of a logical address representing the destination of access requested by a host, an IO manager 112-i of the node 10-i identifies the IP address assigned to the corresponding logical extent. If the identified IP address is not in the LPMT 122-i, a TCP/IP module 113-i of the node 10-i broadcasts an ARP request to acquire, from another node containing the LPMT in which the IP address specified in the ARP request is, a physical address unique to the node. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a storage system that provides a host with a logical volume composed of a plurality of logical extents, and a logical volume management method applied to the system.

  In recent years, a storage system having a plurality of storage devices (for example, disk storage devices) connected to each other via a network has been developed. A storage cluster (storage cluster system) is known as one of this type of storage system.

  A feature of storage clusters is that storage areas (hereinafter referred to as physical volumes) that exist in each of a plurality of storage devices (hereinafter referred to as nodes) are integrated (bundled) across the nodes, so that Another object is to provide a host with at least one cross-sectional / virtual storage area (hereinafter referred to as a logical volume). The volume here is assumed to be a block volume accessed in block units.

  The nodes constituting the storage cluster are always-on storage devices. That is, the participation / removal of a node is a degree due to planned participation / removal during operation or a sudden retirement due to a failure. For this reason, in the storage cluster, there is no frequent participation / detachment of nodes, for example, due to daily power ON / OFF.

  In a storage cluster, data (fragments) is frequently moved / copied appropriately (whether automatically or manually) within a physical volume or across a physical volume of another node. This guarantees access transparency to the logical volume from the host, makes the data redundant in case of a sudden stop of the nodes constituting the storage cluster, disk failure, etc., and improves access performance and reliability. Depending on purpose.

  On the other hand, the host can access the entire logical volume configured across a plurality of nodes only by establishing a path and accessing any one node in the storage cluster. Here, it is assumed that a failure has occurred in a node where the host has a path, and the node in which the failure has occurred (failed node) has stopped. Even in this case, if the data stored in the failed node is made redundant by another node (surviving node) in the storage cluster, the host can re-establish the path to any surviving node, and the logical volume Can continue to access.

  In order to construct such a storage cluster, it is necessary to have a mechanism for managing data on a logical volume at which position on which physical volume of which node.

  In general, a node of a storage cluster is functionally composed of a controller unit and a storage unit. The controller unit manages the path with the host and accepts access. The host can access the same logical volume regardless of which controller unit has a path. The storage unit has a physical volume for storing data.

  A volume (block volume) is composed of a set of fixed-length blocks called extents. The storage unit of each node manages at least one physical volume. On the other hand, the controller unit of each node manages the physical extents constituting the physical volume and the logical extents of the logical volume (arbitrary length areas constituting the logical volume) in association with each other using the mapping table. Thereby, a logical volume (logical volume space) is configured. The association between a physical extent and a logical extent using such a mapping table is also described in Patent Document 1, for example.

  The above mapping table is called a logical extent-physical extent mapping table (LPMT). The LPMT is used to manage the correspondence between which physical extent of a physical volume of which node (storage) the logical extent of the logical volume is composed of.

  The controller units of all the nodes in the storage cluster each hold an LPMT having the same content. Each node manages all logical volumes by LPMT. That is, the controller unit of each node manages the LPMT having entries for the physical volumes / physical extents of all nodes. The reason for this is that when the path from the host to the controller unit is switched to another controller unit, it is possible to immediately continue access to the same logical volume. When creating a logical volume, the configuration management unit, which is a means different from the storage unit and controller unit of each node, determines which physical volume of which node and which physical volume constitutes the logical volume. Then, LPMT is set on each controller unit.

  In the storage cluster (storage system), the first stored in (arranged), for example, the first physical extent of the first physical volume of a certain node in the cluster (hereinafter referred to as the first node). For example, the second physical extent of the second physical volume of another node (hereinafter referred to as the second node) is migrated. Such movement is performed for the purpose of reducing the storage load of the first node. Further, as described in Patent Document 1, in order to improve the access performance to the storage of a certain node, the first logical extent stored in a certain physical extent of the node is changed to another node of the node. Sometimes moved to a physical extent. This migration can improve the access performance and reliability of the storage cluster.

That is, in a storage cluster, a logical extent is moved / copied to another physical extent in the same storage and to another physical extent in order to improve the access performance and reliability of the storage cluster. In this case, according to the conventional technique, the controller units of the storage cluster synchronize with each other and update the LPMTs managed by them simultaneously (simultaneously). For example, as described above, the first logical extent stored in the first physical extent of the first physical volume in the first node is changed to the second logical volume of the second physical volume in the second node. If moving to a physical extent, the first node / first physical volume / first physical extent entry and second node / second physical volume / second of the LPMT of all nodes The physical extent entry is updated.
JP-A-9-62452

  As described above, in a conventional storage cluster (storage system), the controller unit of each node (storage device) that constitutes the cluster is configured with which logical extent of a logical volume is composed of which physical extent of which node's physical volume. LMPT (logical extent-physical extent mapping table) indicating whether or not This LMPT has entries for physical volumes / physical extents of all nodes in the storage cluster. For this reason, the size of the LPMT increases in proportion to the number of physical volumes in all the nodes constituting the storage cluster and the number of extents constituting the physical volume. That is, the amount of information that each node must hold on the memory increases due to LPMT.

  Further, in the conventional storage cluster, when an extent move / copy occurs within a node or between nodes, the controller units of all nodes need to update the LPMTs managed by the nodes all at once (simultaneously). . However, simultaneously updating the controller units so that the LPMTs are the same causes an increase in update processing and adversely affects access performance.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a storage system that can reduce the amount of information that each node must hold and a logical volume management method applied to the system.

  According to one aspect of the present invention, a storage system that provides a host with a logical volume composed of a plurality of logical extents is provided. This system is a plurality of nodes having physical volumes that are managed by dividing the logical extent into a plurality of physical extents that can be stored, have a unique physical address, and set TCP / IP as an upper layer communication protocol. A plurality of nodes interconnected by a network to be applied; and configuration management means for managing the configuration of the logical volume, wherein the configuration management means allocates a network address unique to the logical volume as a logical volume network address. . Each of the plurality of nodes includes an IP address table, an address resolution protocol module, an input / output manager, and a TCP / IP module. The IP address table is an IP address assigned to a logical extent stored in a physical extent in the physical volume of a node (local node) including the IP address table among a plurality of logical extents constituting the logical volume. The IP address in the network address space designated by the logical volume network address corresponding to the position in the logical volume of the logical extent is stored in association with the physical extent. The address resolution protocol module receives an address resolution protocol request that is transmitted via the network and inquires about a resolution to a physical address associated with a specified IP address, and the specified IP address is the IP address of the own node. If it is held in the table, a physical address unique to the node is returned to the requesting node of the address resolution protocol request. The input / output manager specifies an IP address assigned to a logical extent in the logical volume to which the logical address belongs based on a position in the logical volume of a logical address indicating an access destination requested from the host. Then, it is determined whether or not the specified IP address is held in the IP address table of the own node. When the specified IP address is not held in the IP address table of the own node, the TCP / IP module sends an address resolution protocol request for inquiring about a resolution to a physical address associated with the specified IP address. Broadcast on the network, obtain a physical address from the address resolution protocol module of the node having the IP address table that holds the identified IP address among the plurality of nodes, and obtain the physical address The node specified in (1) is requested to access the logical extent to which the specified IP address is assigned.

  According to the present invention, an IP address table (that is, an IP address table corresponding to a logical extent-physical extent mapping table in the prior art) possessed by each node in a storage cluster is a physical volume corresponding to a logical extent managed by that node. / Since only the entries for the physical extent need be held, the amount of information that each node must hold can be reduced. As a result, it is possible to suppress the increase in the amount of information according to the number of nodes constituting the storage cluster, and to reduce the amount of update work of information managed by each node when moving / copying logical extents. Access performance can be improved.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing the configuration of a storage system having a storage cluster configuration according to an embodiment of the present invention. The storage system of FIG. 1 (hereinafter referred to as a storage cluster) has a plurality of, for example, three storage apparatuses (hereinafter referred to as nodes) 10-1, 10-2 and 10-3 (# 1, # 2 and # 3). ). The nodes 10-1, 10-2, and 10-3 are hosts (host computers) that use a storage cluster through an interface 20 such as Fiber Channel, Ethernet (registered trademark), SCSI (Small Computer System Interface), or iSCSI. Connected with.

  The nodes 10-1, 10-2, and 10-3 are composed of controller units 11-1, 11-2, and 11-3 and storage units 12-1, 12-2, and 12-3, respectively. The storage unit 12-i (i = 1, 2, 3) stores the logical extent on the physical volume. That is, the node 10-i including the storage unit 12-i owns a logical extent.

  The controller unit 11-i (i = 1, 2, 3) provides the logical volume to the host by integrating the physical volumes of the storage units 12-1, 12-2, and 12-3. That is, each node 10-i including the controller unit 11-i provides a logical volume composed of logical extents to the host. In other words, each node 10-i constitutes a logical volume. The controller unit 11-i receives an access request from the host. The access request includes a logical address indicating a position in the logical volume to be accessed, for example, a logical block address (LBA) indicating a logical block in the logical volume to be accessed. The controller unit 11-i searches for a node (having a physical volume) in which a logical extent including the requested logical block is stored, and accesses the storage of the node.

  The storage cluster in FIG. 1 plans and manages the configuration of the storage cluster (the configuration of the logical volume provided to the host by the storage cluster) separately from the nodes 10-1 to 10-3 configuring the storage cluster. A configuration management unit 30 is included. The configuration management unit 30 receives a request for creating / deleting a logical volume from a user, and monitors the status of the entire cluster. For example, in order to improve performance, the configuration management unit 30 changes a node that manages a certain logical extent to another node, that is, when moving a certain logical extent to another node. Selection of a destination node and a movement instruction to a movement source node and a movement destination node are performed. The configuration management unit 30 preferably operates on a management terminal implemented using a personal computer (PC) or the like, or any one of the nodes 10-1 to 10-3. In the example of FIG. 1, the configuration management unit 30 operates on the node 10-1. If the node 10-1 fails, the configuration management unit 30 moves to the node 10-2 or the node 10-3 and operates by well-known cluster control.

  The controller units 11-1, 11-2, and 11-3 include target interfaces 111-1, 111-2, and 111-3, IO (input / output) managers 112-1, 112-2, and 112-3, respectively. Processing units of TCP / IP (Transmission Control Protocol / Internet Protocol) modules 113-1, 113-2, and 113-3 and Ethernet controllers 114-1, 114-2, and 114-3, and a logical volume management table (LVMT: Logical Volume Management Table) 115-1, 115-2, 115-3, Ethernet Interface Management Table (EIMT: Ethernet (registered trademark) Interface Management Table) 116-1, 116-2, 116-3, and Address Resolution Protocol ( Each table includes an ARP (Address Resolution Protocol) table 117-1, 117-2, and 117-3.

  Nodes 10-1, 10-2, and 10-3 are connected to the network 40 via Ethernet controllers 114-1, 114-2, and 114-3, respectively. In the present embodiment, the network 40 is a local area network of CSMA / CD (Carrier Sense Multiple Access with Collision Detection) called Ethernet. In the network 40, Ethernet is used as a lower layer protocol (lower layer communication protocol), and TCP / IP is used as an upper layer protocol (upper layer communication protocol). A physical address (lower layer address) called a MAC address is used as a source address and a destination (destination) address of information (Ethernet frame) transmitted and received by the lower layer protocol (Ethernet). An upper layer address called an IP address is used as a source address and a destination (destination) address of information (TCP / IP packet) transmitted / received by an upper layer protocol (TCP / IP). The TCP / IP packet is transmitted in an Ethernet frame.

  The storage units 12-1, 12-2, 12-3 are respectively IO drivers 121-1, 121-2, 121-3, a logical extent-physical extent mapping table (LPMT) 122. -1, 122-2, 122-3 and disks (disk devices) 123-1, 123-2, 123-3. Here, it is assumed that the number of disks 123-i included in each storage unit 12-i is 2, and each disk 123-i provides a physical volume.

  Next, each table managed and referred to in the node 10-i (i = 1, 2, 3), that is, the LVMT 115-i, the EIMT 116-i, the ARP table 117-i, and the LPMT 122-i will be described.

  First, the LVMT 115-i will be described. The LVM 115-i holds basic management information (logical volume management information) of the logical volume. When creating the logical volume, the configuration management unit 30 determines the logical volume ID, the logical volume size, the logical extent size, the network address to be allocated to the logical volume, and the maximum number of nodes that can configure the logical volume in accordance with an instruction from the user. . The determined information constitutes logical volume management information. The configuration management unit 30 generates an LVM that stores logical volume management information, and uniformly distributes and sets the LVMT as the LVM 115-i to all the nodes 10-i configuring the storage cluster of FIG.

  FIG. 2 shows an example of the data structure of the LVMT 115-i. The entries of LVMT 115-i include LVID (Logical Volume ID), LVSIZE (Logical Volume Size), LESIZE (Logical Extent Size), LVNA (Logical Volume Network Address), and LVNODE. Has each item.

  In the LVID, an ID (logical volume ID) unique to the logical volume provided to the host is set. In LVSIZE, the size (information indicating) of the logical volume is set. In LESIZE, the size (information indicating) of the logical extent constituting the logical volume is set. The number of logical extents constituting the logical volume is obtained by LVSIZE / LESIZE.

  In LVNA, a network address (IP network address) and subnet mask assigned to the logical volume are set. In the present embodiment, a space (logical volume) of a host address (IP address) that can accommodate a maximum number of nodes, which will be described later, and the number of logical extents (LVSIZE / LESIZE) constituting a logical volume in a virtual network specified by a network address. Network address space). LVNODE defines the maximum number of nodes (that is, the maximum number of nodes) that can be included as the number of nodes that can own the logical extent that constitutes the logical volume. With this definition, the host addresses (IP addresses) from the beginning to the LVNODE items from the network address space (IP network address space) of the logical volume are 1 for each node constituting the storage cluster that provides the logical volume. Assigned one by one. Further, the LVNODE + 1 and subsequent host addresses (IP addresses) are allocated one by one to each logical extent constituting the logical volume.

  FIG. 3 shows an example of a logical volume managed by the LVMT 115-i. The logical volume shown in FIG. 3 has a size of 2 TB (LVSIZE = 2TB), and is managed by dividing the size into four 512 MB logical extents (LESIZE = 512 MB). This logical volume is assigned 1 (LVID = 1) as the logical volume ID. In the example of FIG. 3, the maximum number of nodes (maximum number of nodes) constituting this logical volume (which can own logical extents) is 10 nodes (LVnode) in anticipation of future increase in the number of logical extents (expansion of logical volume size). = 10). For example, a serial logical extent number as identification information is assigned to each logical extent constituting the logical volume in the arrangement order in the logical volume. Further, 192.168.1.0/24 is assigned as the network address of the logical volume (logical volume network address). Here, an IPv4 (IP version 4) address is used as the IP address. “24” in 192.168.1.0/24 indicates the number of bits of the subnet mask. An IP address other than the IPv4 address, for example, an IPv6 (IP version 6) address can be used as the IP address.

  FIG. 4 shows an example of the LVMT 115-i that manages the logical volume of FIG. Here, it is assumed that logical volumes other than the logical volume of FIG. 3 do not exist in the storage cluster of FIG.

  In the present embodiment, one unique network address space (IP network address space) must be set for one logical volume (LVNA = 192.168.1.0 / 24 in this example). As described above, of the host addresses (IP addresses) in this network address space, the maximum number of nodes (LVNODE = 10) from the top host address (in this example, 192.168.1.1/24) A host address is assigned to each node constituting a logical volume (a storage cluster that provides the logical volume). This host address (IP address) is called a “container IP address”. In this example, 192.168.1.1/24 to 192.168.1.10/24 are used as container IP addresses.

  Of the host addresses (IP addresses) in the network address space described above, host addresses that are not used as container IP addresses (in this example, 192.168.1.11/24 to 192.168.1.254/ 24) are allocated one by one in order from the top of the logical extent. This host address (IP address) is called a “logical extent IP address”. For example, the logical extent 1 located at the head of the logical volume shown in FIG. 3, that is, the logical extent with the logical extent number 1, has the logical extent IP address 192.168.1.11/24.

When the above is formulated, the logical extent IP address is expressed by the following formula: logical extent IP address = {network address, host address}
= {Logical volume network address, maximum number of nodes + logical extent number}
However, logical extent number = (LBA / LESIZE) +1
(1)
Represented by

  For example, when accessing the logical block address (LBA) 256 MB in the logical volume 1 (LVID = 1) shown in FIG. 3, the logical extent size of this logical volume 1 is 512 MB. Become. The logical extent IP address of the logical extent 1 is 192.168.1.11/24 as described above.

  Next, EIMT 116-i (i = 1, 2, 3) will be described. The EIMT 116-i is created for each node 10-i. That is, the EIMT 116-i is a table unique to each node 10-i. The EIMT 116-i of the node 10-i holds the IP address assigned to each logical extent (logical extent constituting the logical volume) managed by the node 10-i in association with the logical extent.

  As described above, when creating a logical volume, the configuration management unit 30 generates an LVM common to each node 10-i and distributes the LVM to each node 10-i as an LVM 115-i. Thereafter, the configuration management unit 30 determines which node 10-i manages each logical extent constituting the created logical volume. Based on the determination result, the configuration management unit 30 generates a unique EIMT 116-i for each node 10-i and distributes it to the node 10-i.

  FIG. 5 shows an example of the data structure of the EIMT 116-i. Each entry of the EIMT 116-i includes LEIP (Logical Extent IP-Address) / CIP (Container IP-Address: Container IP address) and LEMAC (Logical Extent MAC-Address: logical extent MAC address) / CMAC ( (Container MAC-Address: Container MAC address).

  In LEIP / CIP, an IP address (logical extent IP address) assigned to a logical extent managed by the corresponding node 10-i or an IP address (container IP address) assigned to the node 10-i is set. The LEMAC / CMAC has a physical address assigned to the Ethernet controller unit 114-i of the corresponding node 10-i in association with the IP address set in the LEIP / CIP paired with the LEMAC / CMAC. A MAC (Media Access Control) address is set. In the following description, “an IP address assigned to a logical extent” may be referred to as “an IP address that a logical extent has (or has)”.

  FIG. 6 shows the nodes 10-1 to 10-3 constituting the logical volume having the LVID (logical volume ID) 1 of FIG. 3 and the relevant nodes, which are prerequisites for explaining specific examples of the EIMTs 116-1 to 116-3. The relationship with the logical extent in the logical volume managed by 10-1 to 10-3 is shown. In the example of FIG. 6, the nodes 10-1, 10-2, 10-3 (the Ethernet controllers 114-1, 114-2, 114-3) have MAC addresses MAC1, MAC2, and MAC3, respectively.

  In FIG. 6, the node 10-1, that is, the node 10-1 having the MAC address MAC1 has the container IP address 192.168.1.1/24 and the logical extent IP address 192.168.1.11/24. And logical extent 4 having logical extent IP address 192.168.1.14/24. In this case, the MAC address MAC1 of the node 10-1 is changed to the MAC address MAC1 of the container IP address 192.168.1.1/24 or the logical extent IP address 192.168.1.11/24 (with logical extent IP address 192.168.1.11/24). 1) MAC address MAC1 or logical extent IP address 192.168.1.14/24 (logical extent 4 having MAC address MAC1).

  The node 10-2, that is, the node 10-2 having the MAC address MAC2, owns the container IP address 192.168.1.2/24 and has the logical extent IP address 192.168.1.12/24. 2 is managed. In this case, the MAC address MAC2 of the node 10-2 is changed to the MAC address MAC2 of the container IP address 192.168.1.2/24 or the logical extent having the logical extent IP address 192.168.1.12/24 ( 1) may be referred to as MAC address MAC2.

  The node 10-3, that is, the node 10-3 having the MAC address MAC3 owns the container IP address 192.168.1.3/24 and has the logical extent IP address 192.168.1.13/24. 3 is managed. In this case, the MAC address MAC3 of the node 10-3 is changed to the MAC address MAC3 of the container IP address 192.168.1.3/24 or the logical extent having the logical extent IP address 192.168.1.13/24 ( 1) may be called MAC address MAC3.

  FIG. 7 shows an example of the EIMTs 116-1, 116-2, and 116-3 managed by the nodes 10-1, 10-2, and 10-3 in the state of FIG.

  Next, the ARP table 117-i will be described. The ARP table 117-i is managed by the TCP / IP module 113-i included in the controller unit 11-i of the node 10-i. The ARP table 117-i holds conventionally known ARP resolved address information, that is, a pair of an IP address and a MAC address.

  Next, the LPMT 122-i will be described. The LPMT 122-i is a table unique to each node 10-i, unlike the LPMT applied in the prior art. That is, each node 10-i holds and manages its own LPMT 122-i. The LPMT 122-i is used to indicate in which physical extent of which physical volume the node 10-i has the logical extent managed by the node 10-i that holds the LPMT 122-i. In this embodiment, the configuration management unit 30 determines which node 10-i manages the logical extents that make up the logical volume. However, even if the configuration management unit 30 determines which physical extent of which physical volume the node 10-i has to store the logical extent determined to be managed by the node 10-i, the node 10-i The storage unit 12-i may be determined autonomously. The configuration management unit 30 or the node 10-i generates the LPMT 122-i based on the above determination, and sets the LPMT 122-i to the node 10-i.

  FIG. 8 shows an example of the data structure of the LPMT 122-i. Each entry of the LPMT 122-i has items of LEIP (Logical Extent IP-Address), PVID (Physical Volume ID), and PEID (Physical Extent ID).

  In the LEIP, an IP address (logical extent IP address) assigned to a logical extent managed by the corresponding node 10-i is set as in the LEIP in the EIMT 116-i. In the PVID, an ID (physical volume ID) of a physical volume associated with a logical extent IP address indicated by LEIP paired with the PVID is set. In the PEID, an ID (physical extent ID) of a physical extent associated with a logical extent IP address indicated by LEIP paired with the PEID is set. A logical extent of a logical extent IP address indicated by LEIP is stored in a physical extent in a physical volume indicated by PVID and PEID paired with the LEIP.

  FIG. 9 shows a logical extent in a logical volume managed by the nodes 10-1 to 10-3 and a physical volume in which the logical extent is stored, which is a premise for explaining a specific example of the LPMTs 122-1 to 122-3. The relationship with the physical extent in

  In the example of FIG. 9, each of the nodes 10-1 to 10-3 has physical volumes with PVIDs (physical volume IDs) 1 and 2. Each physical volume is composed of physical extents with PEIDs (physical extent IDs) 1, 2, and 3. The physical volumes having the PVIDs 1 and 2 of the node 10-i are provided by, for example, two disks (disk devices) 123-i included in the node 10-i.

  In FIG. 9, the logical extent 1 having the logical extent IP address 192.168.1.11/24 and the logical extent 4 having the logical extent IP address 192.168.1.14/24 are respectively connected to the node 10-1. The physical extent with PEID 3 in the physical volume with PVID 1 and the PEID 2 with physical ID in the physical volume with PVID 2 are included. The logical extent 2 having the logical extent IP address 192.168.1.12/24 is stored in the physical extent having the PEID 2 in the physical volume having the PVID 1 included in the node 10-2. The logical extent 3 having the logical extent IP address 192.168.1.13/24 is stored in the physical extent of PEID 3 in the physical volume of PVID 2 included in the node 10-3.

  FIG. 10 shows an example of the LPMTs 122-1, 122-2, and 122-3 managed by the nodes 10-1, 10-2, and 10-3 in the state of FIG.

Next, the operation of the node 10-1 when an access request is given from the host to the node 10-1 will be described with reference to the flowcharts of FIGS. 11A and 11B.
Now, it is assumed that an access request is given from the host to the node 10-1 in a state where a path is established from, for example, the node 10-1 among the nodes 10-1 to 10-3. To do. Then, the target interface 111-1 of the node 10-1 receives an access request from the host by an arbitrary medium / protocol such as FC / SCSI or Ethernet / iSCSI (step S1). The access request includes a logical block address indicating a logical block in the logical volume to be accessed. Here, it is assumed that an access request requests access to a logical block address LBA (logical block specified by) of a logical volume whose logical volume ID is LVID (hereinafter referred to as logical volume LVID). The target interface 111-1 requests access to the IO manager 112-1 by passing this access request to the IO manager 112-1.

  The IO manager 112-1 performs processing for specifying the logical extent to which the logical block address LBA of the logical volume LVID specified by the access request passed from the target interface 111-1 belongs. Here, the IO manager 112-1 functions as a logical extent IP address specifying means, refers to the LVM 115-1, and obtains the logical extent IP address LEIP assigned to the logical extent corresponding to the LVID and LBA (step S2). .

  Specifically, the IO manager 112-1 first refers to the LVMT 115-1, and refers to the LVNA (logical volume network address), LVSIZE (logical volume size), LESIZE (logical extent size) associated with the LVID, and LVNODE (maximum number of nodes) is acquired. Next, the IO manager 112-1 calculates “logical extent number = (LBA / LESIZE) +1” based on the LBA (logical block address) specified in the access request and the acquired LESIZE (logical extent size). To obtain the logical extent number of the logical extent in the logical volume to be accessed. Then, the IO manager 112-1 performs an operation according to the equation (1) based on the acquired LVNA (logical volume network address) and LVNODE (maximum number of nodes) and the logical extent number obtained by the operation. Thus, the logical extent IP address LEIP assigned to the logical extent in the logical volume to be accessed is obtained (specified).

  Next, the IO manager 112-1 functions as a determination unit, and a logical extent to which the logical extent IP address LEIP obtained in step S2 is assigned (hereinafter referred to as a logical extent LEIP) is designated as the IO manager 112-1. It is determined whether or not the node 10-1 (own node) including the node is managing (step S3). This determination is performed by referring to the LPMT 122-1, and checking whether the logical extent IP address LEIP obtained in step S2 is stored in the LPMT 122-1.

  First, in step S3, it is determined that the logical extent LEIP is managed by the node 10-1 (own node). In this case, the IO manager 112-1 passes an access request (access request to the logical extent LEIP) designating access to the logical extent LEIP to the IO driver 121-1 in the node 10-1 (step S4).

In contrast, in step S3, it is determined that the node 10-1 (own node) does not manage the logical extent LEIP. In this case, the IO manager 112-1 uses the logical extent IP address LEIP assigned to the logical extent LEIP to access the logical extent LEIP managed by a node (other node) other than the node 10-1. Requesting access to the access destination (that is, access to the logical extent LEIP) is made to the TCP / IP module 113-1 in the node 10-1 (step S5).
The TCP / IP module 113-1 performs TCP / IP processing for constructing a highly reliable communication path using Ethernet (lower layer) with other nodes, and controls the Ethernet controller 114-1 to control TCP / IP. Send and receive Ethernet frames that wrap packets. In this embodiment, when access to the logical extent LEIP managed by another node is requested from the IO manager 112-1, the TCP / IP module 113-1 refers to the EIMT 116-1 and refers to the node 10-1 (self The container IP address CIP owned by the Ethernet controller 114-1) of the node is acquired (step S6). In this step S6, the TCP / IP module 113-1 functions as a TCP / IP packet generation means, and issues an access request with the acquired container IP address CIP as the transmission source and the logical extent IP address LEIP as the transmission destination. Generate a TCP / IP packet containing it.

  Next, the TCP / IP module 113-1 functions as an address resolution process, and the MAC address LEMAC corresponding to the logical extent IP address LEIP of the transmission destination necessary to wrap the generated TCP / IP packet in the Ethernet frame is obtained. Address resolution processing for solving is performed (step S7). Details of this address resolution processing will be described later.

  When the TCP / IP module 113-1 resolves (determines) the MAC address LEMAC corresponding to the logical extent IP address LEIP of the transmission destination by the address resolution processing in step S7, the TCP / IP module 113-1 refers to the EIMT 116-1 and refers to the transmission source container IP. The MAC address CMAC corresponding to the address (container IP address of the own node) is acquired (step S8). Next, the TCP / IP module 113-1 is an Ethernet frame that wraps the previously generated TCP / IP packet, and uses the acquired MAC address CMAC as a source MAC address and the resolved MAC address LEMAC ( That is, an Ethernet frame having the transmission destination MAC address as the MAC address LEMAC corresponding to the transmission destination logical extent IP address LEIP is generated, and the Ethernet frame is transmitted to the network 40 via the Ethernet controller 114-1 (step S9). . That is, the TCP / IP module 113-1 transmits an Ethernet frame that wraps a TCP / IP packet including an access request. Here, the TCP / IP module 113-1 performs well-known processes such as dividing an access request into a plurality of packets, packet arrival order control, and packet retransmission control. The Ethernet controller 114-1 has a MAC address and performs transmission / reception control of the Ethernet frame. The same applies to the Ethernet controllers 114-2 and 114-3 of the nodes 10-2 and 10-3.

  Now, the destination MAC address (LEMAC) of the Ethernet frame transmitted from the TCP / IP module 113-1 (the TCP / IP module 113-1 of the node 10-1) to the network 40 via the Ethernet controller 114-1 is: For example, it is assumed that the node 10-2 (the Ethernet controller 114-2) is assigned. In this case, the Ethernet frame is received by the Ethernet controller 114-2 of the node 10-2.

  When the Ethernet controller 114-2 receives the Ethernet frame, that is, receives the Ethernet frame that wraps the TCP / IP packet including the access request, the Ethernet controller 114-2 sends the Ethernet frame to the TCP / IP module 113-2 of the node 10-2. The process is handed over to request processing (step S10). In step S10, the TCP / IP module 113-2 is requested by extracting the access request from the TCP / IP packet wrapped in the Ethernet frame and sending the access request to the IO driver 121-2. Request access processing.

  The IO driver 121-2 receives an access request for a logical extent from the IO manager 112-2 (at the time of an access request from its own node) or the TCP / IP module 113-2 (at the time of an access request from another node). Here, it is assumed that an access request is sent from the TCP / IP module 113-2 to the IO driver 121-2 (step S10). In this case, the IO driver 121-2 receives the access request sent from the TCP / IP module 113-2, and performs access processing for the logical extent LEIP specified by the access request (step S11). Details of the access processing by the IO driver 121-2 will be described later.

  When the IO driver 121-2 executes an access process specified by the access request sent from the TCP / IP module 113-2, the IO driver 121-2 returns a response to the access request to the TCP / IP module 113-2. When receiving a response to the access request from the IO driver 121-2, the TCP / IP module 113-2 responds to the TCP / IP module 113-1 of the access request source node 10-1 via the Ethernet controller 114-2. Is returned (step S12).

  When receiving the response from the TCP / IP module 113-2 of the access request destination node 10-2 via the Ethernet controller 114-1, the TCP / IP module 113-1 of the access request source node 10-1 receives the IO A response is returned to the manager 112-1 (step S13). Then, the IO manager 112-1 receives a response (that is, a response to the access request) from the TCP / IP module 113-1 (step S14). When the IO manager 112-1 receives a response to the access request, the target interface 111-1 returns a response (access response) to the host (step S15).

  On the other hand, when it is determined in step S3 that the logical extent LEIP is managed by the node 10-1 (own node), an access request is sent from the IO manager 112-1 to the IO driver 121-1 in the node 10-1. If it is sent (step S4), the IO driver 121-1 performs access processing to the logical extent LEIP (step S16). Details of the access processing by the IO driver 121-1 will be described later.

  When the IO driver 121-1 executes the access process specified by the access request sent from the IO manager 112-1, the IO driver 121-1 returns a response to the access request to the IO manager 112-1. Then, the IO manager 112-1 receives a response (that is, a response to the access request) from the IO driver 121-1 (step S17). When the IO manager 112-1 receives a response to the access request, the target interface 111-1 returns a response (access response) to the host (step S15).

  Next, details of the address resolution process (step S7 in FIG. 11A) for resolving the MAC address LEMAC corresponding to the logical extent IP address LEIP by the TCP / IP module 113-1 will be described with reference to the flowchart of FIG. .

  First, the TCP / IP module 113-1 functions as an ARP table reference unit and refers to the ARP table 117-1 so that an ARP-resolved entry for the logical extent IP address LEIP already exists in the ARP table 117-1. It is determined whether it is cached (step S21). If it exists, the TCP / IP module 113-1 acquires the MAC address LEMAC corresponding to the logical extent IP address LEIP from the corresponding entry of the ARP table 117-1 (step S22). The TCP / IP module 113-1 sets the MAC address LEMAC as the MAC address LEMAC of the logical extent LEIP (step S23).

  On the other hand, if there is no ARP-resolved entry for the logical extent IP address LEIP in the ARP table 117-1 (step S21), the TCP / IP module 113-1 functions as an ARP request unit, and the EIMT 116-1 Is obtained, the container IP address CIP and the MAC address CMAC possessed by the node 10-1 (own node) are acquired (step S24). In this step S24, the TCP / IP module 113-1 generates an ARP request packet having the acquired container IP address CIP and MAC address CMAC as the transmission source and the logical extent IP address LEIP as the transmission destination, and the ARP request In order to broadcast the packet on the network 40 by the Ethernet controller 114-1, the ARP request packet is sent to the Ethernet controller 114-1.

  When receiving the ARP request packet from the TCP / IP module 113-1, the Ethernet controller 114-1 broadcasts the ARP request packet (ARP request frame) on the network 40 (step S25). Then, the Ethernet controller 114-1 waits for a response to the ARP request.

  The ARP request packet broadcast on the network 40 from the Ethernet controller 114-1 of the node 10-1 is the Ethernet controller of another node in the storage cluster, that is, the Ethernet controllers 114-2 of the nodes 10-2 and 10-3. 114-3. Then, the nodes 10-2 and 10-3 perform processing for the ARP request, and the nodes 10-2 and 10-3 manage the logical extent specified by the IP address LEIP of the destination of the ARP request packet. For example, the node 10-2 returns an ARP response packet (ARP response frame) to the transmission source of the ARP request packet. The ARP response packet includes a logical extent IP address LEIP and a MAC address LEMAC held in the EIMT in association with the IP address LEIP as a transmission source of the ARP response packet (response source to the ARP request packet). Details of processing for an ARP request will be described later.

  The Ethernet controller 114-1 of the node 10-1 receives the ARP response packet returned from the Ethernet controller of the node that manages the logical extent LEIP (here, the Ethernet controller 114-2 of the node 10-2) (step S26). . The ARP response packet received by the Ethernet controller 114-1 is passed to the TCP / IP module 113-1. When the TCP / IP module 113-1 receives the ARP response packet from the Ethernet controller 114-1, the TCP / IP module 113-1 functions as an ARP table registration unit, and the logical extent IP address LEIP and MAC address LMAC of the transmission source of the ARP response packet, that is, access The logical extent IP address LEIP to be performed and the MAC address LMAC of the node that manages the logical extent LEIP specified by the IP address LEIP (the MAC address LMAC of the logical extent LEIP) are added (registered) to the ARP table 117-1. (Step S27). In this way, the TCP / IP module 113-1 acquires the MAC address LMAC of the node that manages the logical extent LEIP as the MAC address LEMAC of the logical extent LEIP (step S23).

Next, details of the processing for the ARP request (ARP request processing) will be described with reference to the flowchart of FIG. 13 by taking as an example the case where the processing is performed in the node 10-2.
In step S25 (see FIG. 12), when the ARP request packet is broadcast from the Ethernet controller 114-1 of the node 10-1, the Ethernet controller 114-2 of the node 10-2 receives the ARP request packet (step). S31). This ARP request packet is also received by the Ethernet controller 114-3 of the node 10-3.

  The ARP request packet received by the Ethernet controller 114-2 is passed to the TCP / IP module 113-2 of the node 10-2. As a result, the Ethernet controller 114-2 requests the TCP / IP module 113-2 to respond to the ARP request. Then, the TCP / IP module 113-2 functions as an ARP module (ARP request processing module).

  First, when the entry of the container IP address CIP of the transmission source of the ARP request packet already exists in the ARP table 117-2, the TCP / IP module 113-2 uses the MAC address CMAC of the transmission source of the ARP request packet as the ARP table. 117-2 is updated (step S32).

  Next, the TCP / IP module 113-2 determines whether the node 10-2 (own node) has the logical extent IP address LEIP depending on whether the entry of the logical extent IP address LEIP of the transmission destination of the ARP request packet exists in the EIMT 116-2. It is determined whether the logical extent LEIP specified in (1) is managed (step S33). If the entry of the logical extent IP address LEIP exists in the EIMT 116-2, that is, if the node 10-2 (local node) manages the logical extent LEIP, the TCP / IP module 113-2 A pair of the source IP address (container IP address) CIP of the request packet and the source MAC address CMAC is registered in the ARP table 117-2 (step S34).

  When an entry for the logical extent IP address LEIP exists in the EIMT 116-2, an entry for the logical extent IP address LEIP also exists in the LPMT 122-2. Therefore, the determination in step S33 is equivalent to determining whether the node 10-2 (own node) manages the logical extent LEIP specified by the logical extent IP address LEIP by referring to the LPMT 122-2. is there.

  When executing step S34, the TCP / IP module 113-2 refers to the EIMT 116-2 and acquires the IP address of the transmission destination of the ARP request packet, that is, the MAC address LEMAC associated with the logical extent IP address LEIP. (Step S35). In step S35, the TCP / IP module 113-2 responds to the ARP request with the MAC address of the logical extent IP address LEIP (that is, the acquired MAC address LEMAC), and then the logical extent IP address LEIP and the MAC address LEMAC. ARP response packet having the IP address CIP and MAC address CMAC as the transmission destination is assembled, and the ARP response packet is passed to the Ethernet controller 114-2. Then, the Ethernet controller 114-2 transmits the ARP response packet passed from the TCP / IP module 113-2 to the network 40 (step S36).

  On the other hand, if the entry of the logical extent IP address LEIP does not exist in the EIMT 116-2 (LPMT 122-2), that is, if the node 10-2 (own node) does not manage the logical extent LEIP (step S33), the TCP The / IP module 113-2 ignores the ARP request and notifies the Ethernet controller 114-2 that no response is required (step S37). In this case, no response is returned from the Ethernet controller 114-2 to the transmission source of the ARP request packet.

Next, details of the access processing (step S11 or S16 in FIG. 11B) by the IO driver will be described with reference to the flowchart of FIG.
Now, the TCP / IP module 113-2 of the node 10-2 gives an access request to the logical extent LEIP to the IO driver 121-2 (access designating access to the logical extent LEIP to which the logical extent IP address LEIP is assigned. Request) is passed (step S10 in FIG. 11B). Then, the IO driver 121-2 performs the process (access process) of step S11 of FIG. 11B as follows according to the flowchart of FIG.

  First, the IO driver 121-2 receives an access request to the logical extent LEIP passed from the TCP / IP module 113-2 (step S41). Next, the IO driver 121-2 searches the LPMT 122-2 using the logical extent IP address LEIP as a key, and thereby associates the physical volume ID (PVID) and physical extent ID (PEID) associated with the IP address LEIP. Is acquired (step S42).

  The logical extent LEIP to which the logical extent IP address LEIP is assigned is stored in the physical extent specified by the acquired PEID of the physical volume specified by the acquired PVID. Therefore, the IO driver 121-2 accesses the physical extent specified by the PEID of the physical volume specified by PVID (step S43). When the access is completed, the IO driver 121-2 returns a response to the access request (access processing completion response) to the direct request source of the access request, that is, the TCP / IP module 113-2 (step S44).

  The details of the access process in step S11 of FIG. 11B have been described above. Next, the access process in step S16 of FIG. 11B will be described. Assume that an access request to the logical extent LEIP is passed from the IO manager 112-1 of the node 10-1 to the IO driver 121-1 (step S4 in FIG. 11A). In this case, the IO driver 121-1 executes the access process of FIG. 11B step S16 in the same procedure as the access process of FIG. 11B step S11 by the IO driver 121-2 (steps S41 to S44 in FIG. 14). However, the IO driver 121-1 receives an access request from the IO manager 112-1 instead of the TCP / IP module 113-1 (step S41), and a response to the access request is received from the IO / not the TCP / IP module 113-1. It returns to the manager 112-1 (step S44).

  FIG. 15 shows that in the storage cluster of FIG. 1, LVMTs 115-1 to 115-3, EIMTs 116-1 to 116-3 and LPMTs 122-1 to 122-3 are configured as shown in FIGS. 4, 7, and 10, respectively. In the state where the entries in the ARP tables 117-1 to 117-3 are empty, an access request designating access to the logical block address (LBA) 768MB of the logical volume 1 is given from the host to the node 10-1. The flow of the operation when Hereinafter, the operation flow of FIG. 15 will be described with reference to the flowcharts of FIGS. 11A and 11B.

  As shown by an arrow 501 in FIG. 15, the logical block address (LBA) “768 MB” (LBA = 768 MB) of the logical volume 1 (LVID = 1) from the host to the node 10-1 in the storage cluster. Assume that an access request specifying access to is given.

  The target interface 111-1 of the node 10-1 receives the access request from the host, and requests the IO manager 112-1 of the node 10-1 to take action for the access request as indicated by an arrow 502. (FIG. 11A step S1).

  The IO manager 112-1 uses the logical block address (LBA) “768MB” of the logical volume 1 as the logical extent IP address (LEIP) “192.168.1.12/24, based on the LVM 115-1 shown in FIG. Is located in a logical extent having "" (step S2 in FIG. 11A).

  Next, the IO manager 112-1 refers to the LPMT 122-1 shown in FIG. 10, and assigns the logical extent IP address (LEIP) “192.168.1.12/24” (the logical extent LEIP) having the node 10. -1 (own node) is managed (step S3 in FIG. 11A). In the example of the LPMT 122-1 shown in FIG. 10, "192.168.1.12/24" is not managed by the node 10-1 (own node) (managed by another node). 112-1, as indicated by an arrow 503, requests the TCP / IP module 113-1 of the node 10-1 to process the access request (step S5 in FIG. 11A).

  The TCP / IP module 113-1 refers to the EIMT 116-1 shown in FIG. 7 and determines the container IP address (CIP) “192.168.1.1/24” owned by the node 10-1 (own node). The TCP / IP packet of the access request having the container IP address CIP = 192.168.1.1 / 24 as the transmission source and the logical extent IP address LEIP = 192.168.1.12 as the transmission destination is acquired. (Step S6 in FIG. 11A).

  Next, the TCP / IP module 113-1 performs address resolution processing for resolving the MAC address of the destination logical extent IP address LEIP = 192.168.1.12 / 24 (step S7 in FIG. 11A). In this address resolution process, the TCP / IP module 113-1 refers to the ARP table 117-1. However, the MAC address “192.168.1.12/24” is not cached in the ARP table 117-1 (step S21 in FIG. 12).

  Therefore, the TCP / IP module 113-1 refers to the EIMT 116-1 and transmits the container IP address CIP = 192.168.1.1 / 24 of the node 10-1 (own node) and its MAC address CMAC = MAC1. An ARP request packet is generated with the logical extent IP address LEIP = 192.168.1.12 / 24 as the transmission destination (step S24 in FIG. 12). The TCP / IP module 113-1 broadcasts the generated ARP request packet by the Ethernet controller 114-1 of the node 10-1 as indicated by an arrow 504a (step S25 in FIG. 12).

  Then, among the nodes 10-1 to 10-3 constituting the logical volume 1 (that is, participating in the 192.168.1.0/24 network) in the storage cluster of FIG. The nodes 10-2 and 10-3 other than 10-1 receive the ARP request packet from the node 10-1 (step S31 in FIG. 13). The ARP packets received by the nodes 10-2 and 10-3 are respectively transmitted from the Ethernet controllers 114-2 and 114-3 of the nodes 10-2 and 10-3 to the TCP / IP of the nodes 10-2 and 10-3. Modules 113-2 and 113-3 are sent as indicated by arrow 504b.

  The TCP / IP module 113-3 of the node 10-3 refers to the ARP table 117-3. At this time, the source IP address CIP = 192.168.1.1 / 24 does not yet exist in the ARP table 117-3. Therefore, the TCP / IP module 113-3 does not update the ARP table 117-3 (step S32 in FIG. 13). In addition, the entry of the IP address (logical extent IP address) LEIP = 192.168.1.12 / 24 of the transmission destination does not exist in the EIMT 116-3 (see FIG. 7). For this reason, the TCP / IP module 113-3 does not respond to the ARP request (step S37 in FIG. 12).

  On the other hand, the TCP / IP module 113-2 of the node 10-2 refers to the ARP table 117-2. At this time, the source IP address CIP = 192.168.1.1 / 24 does not yet exist in the ARP table 117-2. Therefore, the TCP / IP module 113-2 does not update the ARP table 117-3 (step S32 in FIG. 13). Next, the TCP / IP module 113-2 refers to the EIMT 116-2, and an entry of the destination IP address (logical extent IP address) LEIP = 192.168.1.12 / 24 exists in the EIMT 116-2. (Step S33 in FIG. 13). Here, the entry of the logical extent IP address LEIP = 192.168.1.12 / 24 exists in the EIMT 116-2 (see FIG. 7). In this case, the TCP / IP module 113-2 registers the source IP address CIP = 192.168.1.1 / 24 and the MAC address CMAC = MAC1 in the ARP table 117-2 (step S34 in FIG. 13). The TCP / IP module 113-2 refers to the EIMT 116-2, acquires the MAC address LEMAC = MAC2 associated with the logical extent LEIP = 192.168.1.12 / 24, and obtains the logical extent. IP address LEIP = 192.168.1.12 / 24 and MAC address LEMAC = MAC2 are set as a transmission source (response source), IP address (container IP address) CIP = 192.168.1.1 / 24, and MAC address CMAC = The ARP response packet having MAC1 as the transmission destination (response destination) is generated (step S35 in FIG. 13). The TCP / IP module 113-2 returns the generated ARP response packet to the transmission source of the ARP request packet as indicated by the arrow 505a by the Ethernet controller 114-2 of the node 10-2 (step S36 in FIG. 13).

  The ARP response packet returned from the node 10-2 is received by the Ethernet controller 114-1 of the node 10-1 (step S26 in FIG. 12). The Ethernet controller 114-1 requests the TCP / IP module 113-1 to analyze the received ARP response packet as indicated by an arrow 505b. In response, the TCP / IP module 113-1 sets the IP address (logical extent IP address) LEIP = 192.168.1.12 / 24 and the MAC address LEMAC = MAC2 of the transmission source (response source) of the ARP response packet. It is registered in the ARP table 117-1 (step S27 in FIG. 12), and the MAC address LEMAC = MAC2 of the logical extent LEIP specified by the IP address (logical extent IP address) LEIP is obtained (step S23 in FIG. 12). Thereby, the address resolution process (step S7 in FIG. 11A) in the TCP / IP module 113-1 is completed.

  Next, the TCP / IP module 113-1 transmits a TCP / IP packet including an access request for the logical extent LEIP = 192.168.1.12 / 24 with CMAC = MAC1 as the source MAC address and LEMAC = MAC2. Wrap it in an Ethernet frame with the destination MAC address. The TCP / IP module 113-1 transmits this Ethernet frame via the Ethernet controller 114-1 of the node 10-1 as indicated by an arrow 506a (step S9 in FIG. 12).

  The Ethernet frame transmitted from the node 10-1 is received by the Ethernet controller 114-2 of the node 10-2. The Ethernet controller 114-2 sends the received Ethernet frame to the TCP / IP module 113-2 of the node 10-2 as indicated by an arrow 506b.

  The TCP / IP module 113-2 disassembles the Ethernet frame and sends the access request contained in the TCP / IP packet in the frame as indicated by the arrow 506c in the IO driver 121- of the node 10-2. 2 is requested to the IO driver 121-2 for access processing specified by the access request (step S10 in FIG. 12). When receiving the access request from the TCP / IP module 113-2, the IO driver 121-2 refers to the LPMT 122-2 shown in FIG. 10, and the logical extent LEIP = 192.168. 1. Physical volume ID (PVID) = 1 and physical extent ID (PEID) = 2 associated with 12/24 are acquired (steps S41 and S42 in FIG. 14). As a result, the IO driver 121-2 accesses the physical extent of PEID = 2 in the physical volume of PVID = 1 reserved in the disk 123-2 as indicated by the arrow 508 (step S43 in FIG. 14).

  Next, the IO driver 121-2 uses the access result as a return value, and returns a response to the access request to the TCP / IP module 113-2 of the node 10-2 as indicated by an arrow 509a (step S44 in FIG. 14). Then, the TCP / IP module 113-2 returns a response to the TCP / IP module 113-1 of the access request source node 10-1 via the Ethernet controller 114-2 as indicated by an arrow 509b (step B in FIG. 11B). S12).

  When the TCP / IP module 113-1 of the node 10-1 receives the response from the TCP / IP module 113-2 of the node 10-2 via the Ethernet controller 114-1, the IO manager 112- A response is returned to 1 (step S13 in FIG. 11B). In response to this, the IO manager 112-1 returns a response to the target interface 111-1 as indicated by an arrow 509d (step S14 in FIG. 11B). Then, the target interface 111-1 returns a response to the host as indicated by an arrow 509e (step S15 in FIG. 11B).

  As described above, in the present embodiment, while applying a configuration in which the LPMT 122-i has only entries for the physical volume / physical extent of the node 10-i (i = 1, 2, 3), the node 10- Even when a logical extent to be accessed according to an access request from the host for i is managed by another node, the physical volume / physical extent in which the logical extent is stored can be accessed. Accordingly, each of the nodes 10-1 to 10-3 constituting the storage cluster manages the LPMT having entries corresponding to the physical volumes / physical extents of all the nodes 10-1 to 10-3, unlike the conventional technology. There is no need to do. As a result, the amount of information that must be managed by each of the nodes 10-1 to 10-3 constituting the storage cluster can be reduced as compared with the prior art. Further, when moving (or copying) a logical extent described below, it is possible to improve the access performance by reducing the amount of update work of information managed by each node.

Next, logical extent migration (migration) will be described.
In a storage cluster, in order to improve the performance of the storage cluster and solve operational problems, the logical extent of one node is combined with the physical volume / volume of another node that constitutes the logical volume containing the logical extent (along with the node). Moving to a physical extent is done. At this time, it is desirable to minimize the amount of update information of the nodes in the storage cluster necessary for migration in order to realize migration at high speed and safely while continuing access from the host. In this embodiment, an unused host address (IP address) is selected from the network address space of the logical volume including the migration target logical extent (migration source logical extent), and this is temporarily used as a temporary logical extent IP address. By using this, logical extent migration is realized.

  Hereinafter, the movement of the logical extent will be described with reference to FIGS. 16A to 16C and FIG. 16A to 16C are flowcharts showing the procedure of logical extent migration processing, and FIG. 17 is a diagram for explaining logical extent migration.

  First, in the storage cluster of FIG. 1, LVMTs 115-1 to 115-3, EIMTs 116-1 to 116-3, and LPMTs 122-1 to 122-3 are configured as shown in FIGS. 4, 7, and 10, respectively. Shall. In this state, the configuration management unit 30 determines a certain logical extent LEIP (that is, a logical extent to which the logical extent IP address LEIP is assigned) as a movement target (movement source), and the movement destination of the logical extent LEIP. It is assumed that the node, the physical volume TPVID, and the physical extent TPEID are determined (step S51). “T” in the TPVID and the TPEID is a symbol representing the migration destination, and the physical volume TPVID and the physical extent TPEID are respectively a physical volume whose physical volume ID is TPVID (PVID) and a physical extent whose physical extent ID is TPEID (PEID). Point to.

  In this embodiment, the logical extent 4 (logical extent IP address LEIP = 192.168.1.14) managed by the node 10-1 is changed to the physical volume 2 (TPVID = 2) / physical extent of the node 10-3. Assume that it is determined to move to 1 (TPEID = 1). In this embodiment, the logical extent 4 is stored (arranged) in the physical volume 2 (PVID = 2) / physical extent 2 (PEID = 2) of the node 10-1. That is, the logical extent 4 (logical extent IP address LEIP = 192.168.1.14) stored in the physical volume 2 (PVID = 2) / physical extent 2 (PEID = 2) of the node 10-1 is shown in FIG. As indicated by an arrow 700 in FIG. 17, it is assumed that the node 10-3 moves to physical volume 2 (TPVID = 2) / physical extent 1 (TPEID = 1).

  In step S51, the configuration management unit 30 selects an unused host address from the network address space of the logical volume LVID including the migration target (source) logical extent as the temporary logical extent IP address TLEIP used for migration. And assign. In the example in which the migration target logical extent is logical extent 4 (LEIP = 192.168.1.14 / 24), the network address space 192.168.1.0/24 of logical volume 1 (LVID = 1) is not yet stored. The host address in use is selected. Here, it is assumed that TLEIP (LEIP) = 192.168.1.101 / 24 is selected as the temporary logical extent IP address. If there is no unused host address, it is determined that the logical extent cannot be moved.

  Next, the configuration management unit 30 searches for the entry of the physical volume 2 (TPVID = 2) / physical extent 1 (TPEID = 1) as the migration destination from the LPMT 122-3 of the migration destination node 10-3. As the logical extent IP address LEIP of volume 2 (TPVID = 2) / physical extent 1 (TPEID = 1), the temporary logical extent IP address “192.168.1.101/24” (TLEIP = 192.168.1.101). / 24) is assigned (step S52). The state of the LPMTs 122-1 to 122-3 at this time is shown in FIG. In FIG. 18, the hatched portion indicates a portion to which the temporary logical extent IP address is assigned.

  Next, the configuration management unit 30 adds the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 to the EIMT 116-1 of the source node 10-1 and the (Ethernet controller) of the node 10-3 bound thereto. An entry including MAC address TLEMAC = MAC3 (which 114-3 has) is additionally set (step S53). The state of EIMTs 116-1 to 116-3 at this time is shown in FIG. In FIG. 19, the hatched portion indicates the additionally set entry.

  Next, the configuration management unit 30 uses the TCP / IP module 113-3 of the destination node 10-3 to create a temporary logical extent IP address TLEIP = 192.168.1.101 / 24 and its MAC address TLEMAC = MAC3. The ARP request is broadcast via the Ethernet controller 114-3 (step S54). Graftious ARP (hereinafter referred to as G-ARP) request is a special ARP used to notify other nodes that the IP address has been changed, as is well known.

  By sending a G-ARP request by the TCP / IP module 113-3 of the node 10-3, the ARP table 117- of all nodes of the storage cluster excluding the node 10-3, that is, the nodes 10-1 and 10-2. Corresponding entries 1 and 117-2 are initialized. That is, when an entry including the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 already exists in the ARP tables 117-1 and 117-2, the MAC address in the entry is updated to TLEMAC = MAC3. The

  Next, the configuration management unit 30 transfers the data of the migration source logical extent LEIP = 192.168.1.14 / 24 to the IO driver 121-1 of the migration source node 10-1, and the temporary logical extent TLEIP = 192. .168.1.101 / 24 is requested to be copied (step S55). In response to this, the IO driver 121-1 reads the data of the logical extent LEIP = 192.168.1.14 / 24, and sends the data to the TCP / IP module 113-1 of the node 10-1 as temporary logical. A request is made to write to extent TLEIP = 192.168.1.101 / 24 (step S56).

  Then, the TCP / IP module 113-1 has the container IP address CIP = 192.168.1.1 / of the node 10-1 (own node) that manages the migration source logical extent LEIP = 192.168.1.14 / 24. A TCP / IP packet having 24 as a transmission source and a destination temporary logical extent IP address TLEIP = 192.168.1.101 / 24 as a transmission destination is generated (step S57). This TCP / IP packet includes a write request for writing the data of the migration source logical extent LEIP = 192.168.1.14 / 24 into the temporary logical extent TLEIP.

  Next, since the TCP / IP module 113-1 wraps the generated TCP / IP packet in an Ethernet frame, the TCP / IP module 113-1 resolves the MAC address TLEMAC of the temporary logical extent IP address TLEIP = 192.168.1.101 / 24. Address resolution processing is performed (step S58). This address resolution process is performed in the same procedure as in the flowchart of FIG. That is, if there is an entry (target entry) of the temporary logical extent TLEIP in the ARP table 117-1 of the node 10-1, the TCP / IP module 113-1 acquires the MAC address TLEMAC from the entry. On the other hand, if the target entry does not exist in the ARP table 117-1, the TCP / IP module 113-1 sends an ARP request to the node 10-3 (movement destination node 10-3) that manages the temporary logical extent TLEIP and The MAC address TLEMAC = MAC3 of the temporary logical extent TLEIP is acquired by the ARP response from the node 10-3. In this case, the TCP / IP module 113-1 registers the pair of the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 and the MAC address TLEMAC = MAC3 in the ARP table 117-1.

  Next, the TCP / IP module 113-1 sets the MAC address of the transmission source (movement source) MAC address of the container IP address CIP = 102.168.1.1 / 24 (having the Ethernet controller 114-1 of the node 10-1). The address CMAC = MAC1 and the MAC address TLEMAC of the destination (move destination) MAC address as the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 (the Ethernet controller 114-3 of the node 10-3 that manages) = MAC3 (that is, the destination is the node 10-3), an Ethernet frame that wraps the TCP / IP packet is generated (step S59). In step S59, the TCP / IP module 113-1 transmits the generated Ethernet frame to the destination node, that is, the node 10-3 via the Ethernet controller 114-1. This Ethernet frame is received by the Ethernet controller 114-3 of the destination node 10-3, and passed to the TCP / IP module 113-3 of the node 10-3.

  The TCP / IP module 113-3 of the movement destination node 10-3 is accessed from the TCP / IP packet included in the Ethernet frame sent from the movement source node 10-1 in the same manner as in the normal IO processing. The request is extracted and the requested access processing is requested by sending the access request to the IO driver 121-3 of the node 10-3 (step S60). Then, the IO driver 121-3 refers to the LPMT 122-3 of the node 10-3 (own node), and the temporary logical extent TLEIP = 192.168.1.101 / 24 as the write destination is arranged (stored). The physical volume 2 / physical extent 1 physical volume ID (TPVID = 2) / physical extent ID (TPEID = 1) is acquired (step S61). In step S61, the IO driver 121-3 determines the physical extent 1 in the physical volume 2 (TPVID = 2) designated by the acquired physical volume ID (TPVID = 2) / physical extent ID (TPEID = 1). Write data to (TPEID = 1).

  When the data writing (access process) specified by the access request is completed, the IO driver 121-3 responds to the TCP / IP module 113-3 of the node 10-3 with the process completion (step S62). In step S62, the TCP / IP module 113-3 determines that the processing is completed, the Ethernet controller 114-3 of the node 10-3, and the Ethernet controller 114-1 and the TCP / IP module 113-1 of the source node 10-1. To the IO driver 121-1 of the source node 10-1. In response, the IO driver 121-1 notifies the configuration management unit 30 of the completion of processing (copy completion).

  Upon receiving the copy completion notification from the IO driver 121-1, the configuration management unit 30 now issues an access request from the host to the logical volume LVID = 1 including the migration target logical extent LEIP = 192.168.1.14 / 24. The accepted target interface, that is, the target interface 111-1 of the node 10-1 is set in a temporary waiting state for accepting an access request for the logical extent LEIP = 192.168.1.14 / 24 (step S63). . An access request for the logical extent LEIP = 192.168.1.14 / 24 given from the host during this temporary waiting state is not accepted by the target interface 111-1, but is waited for.

  Next, the configuration management unit 30 uses the TCP / IP module 113-3 of the destination node 10-3 to set the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 and its MAC address to NULL. -Broadcast the ARP request via the Ethernet controller 114-3 (step S64). By sending a G-ARP request by the TCP / IP module 113-3 of the node 10-3, the node 10-3 already has an ARP table 117-i including an entry of the temporary logical extent IP address TLEIP = 192.168.1.101 / 24. All nodes 10-i (in the storage cluster of FIG. 1) initialize the entry.

  Next, the configuration management unit 30 holds a pair of the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 and its MAC address TLEMAC = MAC3 from the EIMT 116-3 of the destination node 10-3. The IP address “192.168.1.101/24” in the entry, that is, the temporary logical extent IP address (192.168.1.101/24), is moved to the logical extent IP address to be moved. Rewrite with LEIP = 192.168.1.14 / 24 (step S65). The state of EIMTs 116-1 to 116-3 at this time is shown in FIG. In FIG. 20, in the hatched portion, the temporary logical extent IP address (192.168.1.101/24) is rewritten with the migration target logical extent IP address LEIP = 192.168.1.14 / 24. Indicates an entry.

  Next, the configuration management unit 30 provisional logical extent IP address TLEIP = 192 assigned to the physical volume 2 (TPVID = 2) / physical extent 1 (TPEID = 1) from the LPMT 122-3 of the migration destination node 10-3. .168.1.101 / 24 is searched and the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 is replaced with the migration target logical extent IP address LEIP = 192.168.1.14 / 24. (Step S66). As a result, the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 is released as unused. The states of the LPMTs 122-1 to 122-3 at this time are shown in FIG. In FIG. 21, in the hatched portion, the temporary logical extent IP address TLEIP = 192.168.1.101 / 24 is replaced with the migration target logical extent IP address LEIP = 192.168.1.14 / 24. Indicates the location.

  Next, the configuration management unit 30 searches the entry including the logical extent IP address LEIP = 192.168.1.14 / 24 and its MAC address LEMAC = MAC1 from the EIMT 116-1 of the source node 10-1. Then, the entry is deleted (step S67). The state of EIMTs 116-1 to 116-3 at this time is shown in FIG. In FIG. 22, hatched portions indicate deleted entries.

  Next, the configuration management unit 30 searches the LPMT 122-1 of the source node 10-1 for an entry including the logical extent IP address LEIP = 192.168.1.14 / 24, and sets the logical extent IP address LEIP to the logical extent IP address LEIP. By setting to NULL, the entry is logically deleted (step S68). As a result, the physical extent 2 (PEID = 2) of the physical volume 2 (PVID = 2) in the node 10-1 where the logical extent LEIP = 192.168.1.14 / 24 has been arranged is released. The state of the LPMTs 122-1 to 122-3 at this time is shown in FIG. In FIG. 23, hatched portions indicate locations where the logical extent IP address LEIP is set to NULL.

  Next, the configuration management unit 30 uses the TCP / IP module 113-3 of the destination node 10-3 to perform G-related processing on the logical extent IP address LEIP = 192.168.1.14 / 24 and its MAC address TLEMAC = MAC3. The ARP request is broadcast via the Ethernet controller 114-3 (step S69). As a result, the completion of the migration of the logical extent LEIP = 192.168.1.14 / 24 is notified to all the nodes of the storage cluster except the node 10-3, that is, the nodes 10-1 and 10-2. As a result, the corresponding entries in the ARP tables 117-1 and 117-2 of the nodes 10-1 and 10-2 are initialized. That is, if an entry including the logical extent IP address TLEIP = 192.168.1.14 / 24 already exists in the ARP tables 117-1 and 117-2 of the nodes 10-1 and 10-2, the MAC in the entry The address is updated to TLEMAC = MAC3.

  Next, the configuration management unit 30 sets the target interface that is temporarily waiting to accept an access request from the host for the logical extent LEIP = 192.168.1.14 / 24 that has been moved, that is, the target of the node 10-1. The temporary waiting state is canceled for the interface 111-1 (step S70). As a result, a series of logical extent movement processing is completed.

[Modification]
Next, a modification of the above embodiment will be described.
In the above embodiment, the IP address (logical extent IP address) LEIP assigned to the logical extent in the logical volume to which the logical block address (logical address) LBA indicating the access destination requested by the host belongs is expressed by the equation (1). Is determined (specified) by an operation according to

  The feature of this modification is that the logical extent IP address LEIP assigned to the logical extent in the logical volume to which the logical block address LBA belongs is obtained without performing such an operation. Therefore, in this modification, each node 10-i (i = 1, 2, 3) in the storage cluster shown in FIG. 1 has a logical extent management table (LEMT) 118 having the structure shown in FIG. -i is provided.

  Each entry of the LEMT 118-i has a logical block address LBA (logical address) range (logical block address range) LBARANGE of a logical extent constituting the logical volume and a logical extent IP address ( Logical Extent IP-Address) Each item of LEIP is included. The IO manager 112-i searches the LEMT 118-i for an entry in which the logical block address range LBARANGE including the logical block address LBA specified by the access request is set, and makes a pair with the logical block address range LBARANGE. Obtain (specify) the logical extent IP address LEIP. Here, each entry of the LVM 115-i may have the number of logical extents constituting the logical volume instead of the logical extent size (LESIZE).

  According to this modification, unlike the above-described embodiment, the logical volume is configured even if the sizes of the plurality of logical extents constituting the logical volume are not constant (that is, the sizes differ for each logical extent). The logical extent IP address LEIP assigned to the logical extent in the logical volume to which the logical block address LBA belongs is converted into the LEMT 118 even if consecutive logical extent IP addresses (host addresses) are not assigned in the arrangement order of the plurality of logical extents. -i can be acquired promptly from i. However, the number of entries in the LEMT 118-i is proportional to the number of logical extents constituting the logical volume.

  In addition, this invention is not limited to the said embodiment or its modification example as it is, A component can be deform | transformed and embodied in the range which does not deviate from the summary in an implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment or the modification thereof. For example, you may delete a some component from all the components shown by embodiment or its modification.

The block diagram which shows the structure of the storage cluster which concerns on one Embodiment of this invention. The figure which shows the example of a data structure of the logical volume management table (LVMT) shown by FIG. The figure which shows the example of the logical volume whose logical volume ID (LVID) is 1. FIG. 4 is a diagram showing an example of an LVM that manages the logical volume of FIG. 3. The figure which shows the example of a data structure of the Ethernet interface management table (EIMT) shown by FIG. The figure which shows the relationship between each node which comprises the logical volume of FIG. 3, and the logical extent in the said logical volume managed by the said node used as the premise at the time of explaining the specific example of EIMT for every node. The figure which shows the specific example of EIMT of each node in the case of FIG. The figure which shows the data structure example of the logical extent-physical extent mapping table (LPMT) shown by FIG. The figure which shows the relationship between the several logical extent which comprises a logical volume, and the physical extent in the physical volume in which the said logical extent is stored used as the premise at the time of explaining the specific example of LPMT for every node. The figure which shows the specific example of LPMT of each node in the case of FIG. The flowchart which shows the procedure of the access process with respect to the access request from a host. The flowchart which shows the procedure of the access process with respect to the access request from a host. The flowchart which shows the procedure of the address resolution process for resolving the MAC address corresponding to a logical extent IP address. The flowchart which shows the procedure of an ARP request | requirement process. The flowchart which shows the procedure of the access process by IO driver. The figure which shows the flow of operation | movement when an access request | requirement is given to a certain node in the storage cluster of FIG. 1 from a host. The flowchart which shows the procedure of a logical extent movement process. The flowchart which shows the procedure of a logical extent movement process. The flowchart which shows the procedure of a logical extent movement process. The figure for demonstrating a logical extent movement. FIG. 16A is a diagram showing an example of the state of each LPMT after step S52 is executed. FIG. 16A is a diagram showing a state example of each EIMT after step S53 is executed. FIG. 16C is a diagram showing a state example of each EIMT after step S65 is executed. FIG. 16C is a diagram showing an example of the state of each LPMT after step S66 is executed. FIG. 16C is a diagram showing a state example of each EIMT after step S67 is executed. FIG. 16C is a diagram showing an example of the state of each LPMT after step S68 is executed. The figure which shows the example of a data structure of the logical extent management table (LEMT) applied in the modification of the said embodiment.

Explanation of symbols

  10-1 to 10-3 ... node (storage device), 11-1 to 11-3 ... controller unit, 12-1 to 12-3 ... storage unit, 30 ... configuration management unit, 40 ... network, 111-1 ... 111-3 ... Target interface, 112-1 to 112-3 ... IO manager, 113-1 to 113-3 ... TCP / IP module (address resolution protocol module), 114-1 to 114-3 ... Ethernet controller, 115 -1 to 115-3, 115-i ... Logical volume management table (LVMT), 116-1 to 116-3, 116-i ... Ethernet interface management table (EIMT), 117-1 to 117-3 ... Address resolution Protocol (ARP) table, 118-i ... logical extent management table (LEMT), 121-1 to 121-3 ... IO driver, 122-1 to 122-3, 122-i ... logical error Stent - physical extent mapping table (LPMT, IP address table), 123-1～123-3 ... disk.

Claims (9)

In a storage system that provides a host with a logical volume consisting of multiple logical extents,
A plurality of nodes having physical volumes divided and managed into a plurality of physical extents capable of storing the logical extent, having a unique physical address, and applying a TCP / IP to an upper layer communication protocol A plurality of interconnected nodes;
Configuration management means for managing the configuration of the logical volume, comprising configuration management means for assigning a network address unique to the logical volume as a logical volume network address,
Each of the plurality of nodes is
An IP address assigned to a logical extent stored in a physical extent stored in the physical volume of the physical volume of the plurality of logical extents constituting the logical volume, wherein the logical extent is located at a position in the logical volume. An IP address table that holds the corresponding IP address in the network address space designated by the logical volume network address in association with the physical extent;
When an address resolution protocol request is transmitted that is transmitted through the network and inquires about resolution to a physical address associated with a designated IP address, and the designated IP address is held in the IP address table of the node An address resolution protocol module for returning a physical address unique to the node to the requesting node of the address resolution protocol request;
Based on the position in the logical volume of the logical address indicating the access destination requested from the host, the IP address assigned to the logical extent in the logical volume to which the logical address belongs is specified, and the specified IP An input / output manager that determines whether an address is held in the IP address table of the node;
If the specified IP address is not held in the IP address table of the node, an address resolution protocol request for inquiring resolution to a physical address associated with the specified IP address is broadcast on the network. The physical address is acquired from the address resolution protocol module of the node having the IP address table that holds the specified IP address among the plurality of nodes, and the node specified by the acquired physical address is acquired. And a TCP / IP module for requesting access to the logical extent to which the specified IP address is allocated. Each of the plurality of nodes retains logical volume management information having the logical block IP address for specifying a head IP address among consecutive IP addresses allocated to a plurality of logical extents constituting the logical volume. Logical volume management table
The configuration management means determines an IP address in a network address space designated by the logical volume network address for the plurality of logical extents constituting the logical volume from a head IP address specified by the logical volume management information. , Assigning the plurality of logical extents in the arrangement order in the logical volume,
The I / O manager is specified by the relative position in the logical block of the logical extent to which the logical address indicating the access destination requested by the host belongs and the logical volume management information in the network address space designated by the logical volume network address. 2. The storage system according to claim 1, wherein an IP address assigned to the logical extent is specified based on the first IP address to be assigned. Each of the plurality of nodes includes a logical extent management table that holds an IP address allocated to a plurality of logical extents constituting the logical volume in association with a range of logical addresses belonging to the logical extent;
The I / O manager refers to the logical extent management table, and is associated with the logical address range including the logical address indicating the access destination requested from the host, and the logical extent in the logical address range. The storage system according to claim 1, wherein an IP address assigned to the server is specified. Each of the plurality of nodes has an address resolution protocol for storing a physical address associated with an IP address assigned to a logical extent, acquired in response to an inquiry by the address resolution protocol request, in association with the logical extent. Including tables,
The TCP / IP module refers to the address resolution protocol table when the specified IP address is not held in the IP address table of the node, and the physical address associated with the specified IP address is If it is not held in the address resolution protocol table, the address resolution protocol request is broadcast on the network to acquire the physical address, and the acquired IP address is assigned to the acquired physical address. Add to the address resolution protocol table in association with the logical extent, and if the physical address associated with the specified IP address is held in the address resolution protocol table, the physical address is related to the address resolution protocol table. The storage system of claim 1, wherein the obtaining the Lotto Col table. Each of the plurality of nodes is an IP address unique to the node, and is an IP address other than an IP address in a network address space designated by the logical volume network address assigned to the plurality of logical extents constituting the logical volume. A pair of an address and a physical address unique to the node, a pair of an IP address assigned to a logical extent stored in a physical extent in a physical volume of the node and a physical address unique to the node Including an interface table that holds
When the physical address associated with the specified IP address is not held in the address resolution protocol table, the TCP / IP module refers to the interface table and refers to the IP address and physical address unique to the node. Obtaining an address, generating the address resolution protocol request with the acquired IP address and physical address as the transmission source and the specified IP address as the transmission destination, and generating the generated address resolution protocol request Broadcast on the network,
When the address resolution protocol module receives an address resolution protocol request broadcast from another node, the address resolution protocol module determines whether or not the destination IP address of the address resolution protocol request is held in the interface table. The address resolution protocol request destination IP address is registered in the address resolution protocol in association with the address resolution protocol request destination physical address and held in the interface table. A response with the IP address and physical address of the transmission destination of the address resolution protocol request as the transmission source and a response with the IP address and physical address of the transmission source of the address resolution protocol request as the transmission destination is returned to the request source. Item 4 Storage system. Each of the plurality of nodes has a physical volume associated with the identified IP address by the IP address table when the identified IP address is held in the IP address table of the node. Includes an I / O driver that identifies the physical extent and accesses the physical extent in the identified physical volume;
The configuration management means determines an arbitrary logical extent in the logical volume as a migration source, and determines a migration destination of the logical extent, a node, a physical volume, and a physical extent in the physical volume, and moves the migration extent. An unused IP address in the network address space designated by the logical volume network address is assigned to a destination physical extent as a temporary logical extent IP address, and the temporary logical extent IP address is assigned to the IP address table of the destination node. Is registered in association with the physical extent in the determined physical volume, and the pair of the temporary logical extent IP address and the physical address unique to the migration destination node is registered in the interface table of the migration destination node. The The output driver of the recording and the movement source node reads the data of the migration source logical extent,
When the temporary logical extent IP address is not held in the address resolution protocol table, the TCP / IP module of the source node inquires about resolution to a physical address associated with the temporary logical extent IP address. By broadcasting a resolution protocol request on the network and obtaining a physical address associated with the temporary logical extent IP address, the temporary logical extent is sent to the destination node specified by the physical address. Requesting that the read data be written to a logical extent to which an IP address is assigned;
The input / output driver of the migration destination node refers to the IP address table of the migration destination node, identifies a physical extent in a physical volume associated with the temporary logical extent IP address, and 6. The storage system according to claim 5, wherein the requested data is written to a physical extent in the specified physical volume.   The configuration management means uses the temporary logical extent IP address registered in the interface table of the migration destination node in response to completion of writing by the input / output driver of the migration destination node. The IP address table of the migration destination node is rewritten with the IP address of the extent, and the temporary logical extent IP address held in association with the physical extent in the migration destination physical volume is changed to the migration source logical extent. And the IP address and physical address pair of the source logical extent already registered in the interface table of the destination node before the rewriting are deleted, and before the rewriting The IP address of the destination node The storage system of claim 6, wherein the deleting the IP address of the migration source logical extents in less table has already been registered. A plurality of nodes having physical volumes that are managed by dividing a physical extent into a plurality of physical extents that can store logical extents, each having a unique physical address, and mutually connected by a network that applies TCP / IP to an upper layer communication protocol A plurality of nodes connected to each other, and configuration management means for managing the configuration of the logical volume, the configuration management means for assigning a network address unique to the logical volume as a logical volume network address, and comprising a plurality of logical extents In a logical volume management method applied to a storage system that provides a configured logical volume to a host,
When configuring the logical volume, each of the plurality of nodes is assigned an IP address assigned to a logical extent stored in a physical extent in the physical volume of the physical volume of the plurality of logical extents constituting the logical volume. An IP address table that holds an IP address in a network address space designated by the logical volume network address corresponding to a position in the logical volume of the logical extent in association with the physical extent, Steps set by the configuration management means or the node itself;
Of the plurality of nodes, a node that has received an access request from the host, based on the position in the logical volume of the logical address indicating the access destination specified in the access request, the logical address to which the logical address belongs. Identifying an IP address assigned to a logical extent in the volume;
The node receiving the access request determines whether the identified IP address is held in the IP address table of the node receiving the access request;
If the specified IP address is not held in the IP address table of the node that has received the access request, an address resolution protocol request for inquiring resolution to a physical address associated with the specified IP address is sent to the access The node receiving the request broadcasts on the network;
When the specified IP address is held in the IP address table of the node that has received the address resolution protocol request among the plurality of nodes, a physical address unique to the node that has received the address resolution protocol request is set. Returning from the node that received the address resolution protocol request to the requesting node of the address resolution protocol request;
The requesting node to which the physical address is returned requests the node specified by the acquired physical address to access the logical extent to which the specified IP address is assigned. A logical volume management method comprising:   When the identified IP address is held in the IP address table of the node that has received the access request, the node that has received the access request is associated with the identified IP address by the IP address table. 9. The logical volume management method according to claim 8, further comprising the step of specifying a physical extent in the physical volume and accessing the physical extent in the specified physical volume.

Images