JP2007042119A - Computer system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that access performance is deteriorated by overhead of a LAN and a server though TCO is reduced by integrating operation management to the server in the conventional Thin client computer system such as a NC. <P>SOLUTION: The computer system is constituted of: a plurality of client computers; a storage shared by the respective client computers; a management console for managing the storage; and a connecting device for mutually connecting the respective client computers, the storage and the management console. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a method for realizing a client computer system composed of a plurality of client computers, and in particular, a method for realizing both reduction in operation cost required for maintenance of a storage device for a client computer and improvement in access performance of the storage device. About.

  A computer system in which a plurality of users perform business processing using client computers such as a plurality of PCs is called a client computer system. Conventionally, a client computer system has been constructed using a PC or the like having a function capable of executing all the calculation processes independently. Since such a client computer is a “fat” computer equipped with a high-performance CPU, a large-capacity memory, a large-capacity hard disk device, a high-speed graphic device, etc., it is commonly referred to as a “Fat client”. A client computer system using the Fat client computer is called a Fat client system (FCS). In FCS, multiple Fat clients are connected to each other via LAN, and communication such as file transfer and mail transmission / reception is performed as necessary. The Fat client includes a disk device as storage for storing the OS, applications, user data, and the like. The OS and applications are usually installed by the user himself / herself. Also, the created user data is backed up by a backup device such as a tape device located anywhere on the LAN so that it will not be lost due to a hard disk failure.

  As another example of such a client computer system, a Network Computer system is known. A computer that constructs this system is called a Network Computer (NC), and uses a low-cost, low-functional “skinny” computer that does not have a hard disk device that stores an OS or applications. Such a computer is called a “Thin client” in contrast to a Fat client. A client computer system using this thin client is referred to as a thin client system (TCS).

  Hereinafter, an NC system system will be described as an example of TCS with reference to FIG.

  1000 (ac) is a plurality of network computers (hereinafter abbreviated as NC), 7 is a server, 1 is a storage for the server 7, and 6 is a LAN for connecting the NC 1000 and the server to each other.

  The NC1000 does not have a storage for storing the OS, applications, and user data. A part of the storage 1 of the server 7 is mounted as a remote storage for the NC1000 via the LAN 6, and all OS, applications, and user data are stored here. At startup, the OS is loaded from the storage 1 of the server 7 via the LAN 6 and executed. The same applies to the activation of the application.

JP-A-6-168198

  While the conventional FCS has the advantages of easy system construction and low initial installation costs, as described above, the OS / application must be installed and upgraded by the user himself. There is a problem that the management cost increases, and as a result, the total cost of ownership (TCO) generated to maintain the system becomes very expensive.

  In addition, TCS, which is represented by the NC system of the conventional technology, aims to reduce the TCO by consolidating the operation management to the server so that the administrator can centrally manage it, eliminating the operation management by individual users. It is a thing. On the other hand, since disk access for all users is issued via LAN and server, the overhead of LAN processing and the remote disk control processing overhead generated by both NC and server will improve the access performance to each client's internal disk. There is a problem that it is greatly reduced. In addition, since all disk access is concentrated on the server, the server is required to have extremely high CPU performance and I / O processing performance, and there is a problem that an expensive server is required for the operation of the client computer system. .

  An object of the present invention is to provide a client computer system that reduces TCO by reducing operation management costs.

  Another object of the present invention is to provide a client computer system that realizes high-speed disk access with low overhead.

  It is another object of the present invention to provide a client computer system that does not place a load on a LAN or server.

  Another object of the present invention is to provide a client computer system that realizes high-speed disk access exceeding the built-in disk access performance of the conventional FCS.

  The present invention provides a plurality of diskless client computers, a shared storage of all client computers, a management console for managing the shared storage, and an I / O for interconnecting all the diskless computer systems, the shared storage and the management console. It is a client computer system configured by a channel coupling circuit.

  Here, the diskless client computer is used to boot up the OS from the I / O channel I / F control circuit that connects to and controls the I / O channel such as the fiber channel and the storage connected to the I / O channel. Bootup control circuit. The diskless client computer does not have a built-in disk device.

  In addition, the shared storage includes an I / O channel I / F control circuit that connects to and controls I / O channels such as Fiber Channel, a dedicated logical unit (LU) for each client computer, LU definition part that creates and deletes logical units shared by clients, access prohibition / permission of the created LU to each client computer, LU attributes, and virtual LU number (LUN displayed on the client computer) : Logical Unit Number) and the LUN management table for managing the LU number for internal control of the shared storage, and the access to each LU from each client is restricted according to the storage information of the LUN management table, and if permitted Cache the data of the access control unit that performs read / write access control for the LU and at least one disk device that constructs the LU. In the event of a cache hit, a cache for realizing high-speed data transfer without reading / writing from the disk device, a cache resident control unit that controls the specified LU to reside in the cache, and shared storage are set. And a communication control circuit for controlling communication with a management console for control.

  In addition, the management console instructs the LU creation of the shared storage, sets access restrictions with each LU for each client computer, and communicates between the LU management unit that controls the setting of the virtual LUN for each LU and the shared storage. And an I / O channel I / F control circuit for connecting to and controlling an I / O channel such as a fiber channel.

  According to the present invention, it is possible to centrally manage installation and version upgrade of applications and OSs, and data backup, so that it is possible to reduce the operation management cost and realize a client computer system with a low TCO.

  In addition, according to the present invention, since a LAN or server is not used to share storage among a plurality of clients, there is an effect that a client computer system that realizes high-speed disk access with low overhead can be realized.

  In addition, according to the present invention, a LAN or server is not used to share storage among multiple clients, so high-speed server processing and network processing are realized by releasing the load for storage sharing from the LAN or server. There is an effect that can be done.

  In addition, according to the present invention, since an application or OS can be shared by a plurality of clients, the operation management cost caused by the installation or version upgrade can be reduced to a minimum, and by making the cache resident, There is an effect that high-speed access performance can be realized.

Embodiments of the present invention will be described below with reference to the drawings. The present invention is not limited to this embodiment.

(A) 1st embodiment (circle) structure 1st embodiment is described. FIG. 1 is a configuration diagram of a client computer system according to the first embodiment. In this embodiment, the client computer is assumed to be a PC (Personal Computer), and in particular, this PC will be described as a “diskless PC” that does not include a hard disk device therein. Of course, the same can be applied to a client computer system other than a PC.

  In the figure, 2 (2a to 2c) is a plurality of diskless PCs, 1 is a storage shared by all client PCs 2, 4 is a management console for managing storage, 3 is a diskless PC 2, storage 1 and management console Fiber channel connection circuits for connecting 4 to each other, and 5a to 5i are fiber channels.

  FIG. 2 is a configuration diagram of the diskless PC 2. 21 is a central control unit that controls the entire diskless PC and executes various calculation processing programs, 22 is a fiber channel I / F control circuit that connects and controls the fiber channel, and 23 is operated by the operator of this PC2 221 is a bootup control circuit for booting up the OS from the storage connected to the fiber channel. As described above, the diskless PC does not include a built-in disk device.

  FIG. 3 is a configuration diagram of the storage 1. Reference numeral 11 denotes a central control unit that controls the entire storage. In the present embodiment, the central control device 11 is a device that operates by a program, and includes the following program parts. 12 is a logical unit (LU: Logical Unit) dedicated to each diskless PC2 (2a to 2c), a LU definition unit that creates or deletes a logical unit shared by multiple diskless PC2, and 14 is each diskless LUN management table to manage the access prohibition / permission of each LU to the PC, the attributes of the LU, and the correspondence between the virtual LU number (LUN: Logical Unit Number) shown on the diskless PC and the internal LUN for internal storage management , 16 is an access control unit that performs control for restricting / permitting access from each diskless PC 2 to each LU according to the information in the LUN management table 14, and, if permitted, performs read / write access control for the LU.

  13 is a communication control circuit that controls communication with a management console that performs control such as defining LUs in the storage, 15 is a fiber channel I / F control circuit that connects to and controls the fiber channel 5d, 17 Reference numerals (170 to 176) denote logical units (LUs) that are logical volumes that are configured by a fixed storage area of at least one physical disk device and are visible to the host device.

  FIG. 4 is a configuration diagram of the management console 4. Reference numeral 41 denotes a central control device that controls the entire management console 4. In the present embodiment, the central control device 41 is assumed to be a device that operates by a program, and includes the following program parts. An LU management unit 44 instructs to create an LU of the storage 1, sets an access restriction for each LU for each diskless PC, and controls to associate each internal LUN with a virtual LUN.

42 is an input / output device for the operator to operate the management console 4, 43 is a communication control circuit for controlling the communication path 6 for communication for various settings of storage, 45 is a connection to the fiber channel 5e, A fiber channel I / F control circuit 46 that performs control is a local disk that is an internal disk that stores the OS and programs of the management console 4.
O Operation Next, the operation of this embodiment will be described.

(1) Internal LUN and virtual LUN
The LU (Logical Unit) will be described. LU is a logical volume when viewed from a host computer such as a diskless PC. The host computer recognizes one LU as one logical disk device.

  The storage 1 can build a plurality of LUs inside. This is called an internal LU. In storage 1, serial numbers are assigned with integers starting from 0 to manage internal LUs. This number is called an internal LUN (Logical Unit Number).

On the other hand, in the client computer system of the present invention, since a plurality of diskless PCs share one storage, a dedicated LU is allocated to each diskless PC. Generally, a host computer such as a PC searches a storage to be connected at the time of OS booting to detect an LU, but there are cases where there are some restrictions on the search method. that is,
(a) Search LUNs in order from 0
(b) Assuming that LUNs exist with consecutive numbers, there are two points: no search is performed after a certain number does not exist. This is a device for shortening the search time. Assuming that the diskless PC of the present invention is also a host computer having such characteristics, a diskless PC to which an LU other than internal LUN = 0 is assigned cannot detect this LU. Therefore, when allocating one or more LUs to all diskless PCs, it is necessary to allocate LUNs starting from 0 and using consecutive numbers. The LU that can be seen from the diskless PC is called a virtual LU, and its number is called a virtual LUN, and is distinguished from an internal LU and an internal LUN. The correspondence relationship between these internal LUNs and virtual LUNs is managed by a LUN management table 14 provided in the central control device 11 of the storage 1.
(2) LUN management table
An example of the LUN management table 14 is shown in FIG. The LUN management table 14 stores the owned upper computer identifier, port number, Target ID, virtual LUN, internal LUN, and attribute.

  The owned host computer identifier stores information for specifying a host computer that owns a certain LU. More specifically, when the storage and the host computer are connected by Fiber Channel as in this embodiment, the Port Name based on the World Wide Name of the Fiber Channel I / F connection circuit 22 of the diskless PC 2 and the Fiber Channel Represented by a combination of S-ID (Source-ID) determined by initialization. Although not shown in the present embodiment, the SCSI ID is stored when SCSI (Small Computer Systems Interface) is used as a connection I / F with a host computer.

  The port number stores the number of the Fiber Channel connection port provided in the storage 1. In this embodiment, since the number of ports is two, the number of the port to which each diskless PC 2 is connected is stored.

  TargetID stores the identification ID of storage 1 in the connection I / F with the host computer. When the connection I / F is Fiber Channel as in this embodiment, since it is determined that each port has only one D-ID (Destination-ID), it can be identified by the port number. May be omitted, or the D-ID determined at the time of fiber channel initialization may be stored. When the connection I / F is SCSI, a plurality of IDs can be provided for the same port, and each ID at that time is stored.

  The virtual LUN and the internal LUN indicate a correspondence relationship between the internal LUN assigned to the host computer and the virtual LUN that can be seen from each host computer. For example, two LUs with internal LUN = 0 and 1 are allocated to the diskless PC0, and the virtual LUNs are 1 and 0, respectively.

The attribute indicates the ownership attribute of each LU by the host computer. “Dedicated” indicates that the LU is exclusively used by one host computer. “Shared” indicates that the LU is shared by a plurality of host computers. “Privilege” is synonymous with “dedicated”, but is in a special management state, and in this embodiment, indicates that it can be operated only by the management console 4. The difference from “dedicated” is that the management console 4 is not an LU created for storing and using programs and the like.
(3) LU Creation Next, a procedure for creating the LU 17 by controlling the storage 1 from the management console 4 will be described.

  The administrator operates the LU management unit 44 using the input / output device 42 of the management console 4. The administrator uses the LU management unit 44 to select a physical disk device for creating an LU. If the storage 1 has a plurality of disk devices, one of them is selected. If a RAID (Redundant Array of Inexpensive Disks) is to be configured, a plurality of disk devices are selected, and these are grouped and handled as one virtual physical disk device.

  Next, the capacity of the LU to be created in the selected physical disk device is determined. Using the above setting information, the administrator issues a LU creation command to the storage 1 by controlling the LU management unit 44. Based on the command of the LU management unit 44, the central control device 41 controls the communication control circuit 43 and transfers the setting information to the storage 1 via the communication path 6.

  The communication control circuit 13 of the storage 1 receives this and notifies the central control device 11, and the device 11 creates an LU having a specified capacity in a specified physical disk device by means of the LU definition unit 12. The LU definition unit 12 registers this LU in the internal LUN of the LUN management table 14. At this point, the attribute is privileged and other settings are not set.

  When the LU creation is successful, the LU definition unit 12 reports the LU creation success to the management console 4 through the communication path 6 under the control of the communication control circuit 13.

  Next, the identifier of the host computer that owns the LU is set. Assume that the created internal LUN = 1. It is assumed that the administrator registers this LU with LUN = 0 of the diskless PC0. The administrator operates the LU management unit 44 and sets the LU owned host computer identifier = World Wide Name of the diskless PC 0, virtual LUN = 0, port number = 0, and attribute = dedicated. Thereafter, these setting values are transferred to the storage 1 via the communication path 6 as described above, and the LU definition unit 12 sets them in the LUN management table 14. As a result, only the client PC0 can use this LU thereafter, and it can be set to an LU that cannot be recognized by other diskless PCs.

  In this embodiment, since Fiber Channel is used, TargetID is unused, and 0 is stored in the same section of the LUN management table for convenience.

Subsequently, other LUs can be created and owned in the same manner as described above, and dedicated LUs for each diskless PC can be created. One diskless PC can include a plurality of dedicated LUs, and the diskless PC2 shown in FIG. 5 is an example including two dedicated LUs.
(4) Creating a shared LU In the above description, the creation of an LU dedicated to each diskless PC has been described. However, a shared LU can also be created by each diskless PC. The creation of the dedicated LU is different from setting the LU attribute to “shared”. The same internal LU is defined for each diskless PC so that each diskless PC can recognize it as one LUN. In the example of FIG. 5, the internal LUN = 0 is defined as a shared LU three times as a virtual LUN = 1 for diskless PC0, a virtual LUN = 1 for diskless PC1, and a virtual LUN = 2 for diskless PC2. It is also possible not to allow a certain upper computer to be shared. This can be realized simply by not defining a virtual LUN for the host computer in the LUN management table 14.
(5) Access Control Next, LU access control will be described. When the diskless PC 2 starts up, initialization processing of the following two steps is performed.

  The first step is a port login process that is performed when the Fiber Channel is initialized. This is a process performed to establish a connection on the fiber channel. Through this process, the Fiber Channel I / F control circuit 15 of the storage 1 acquires the World Wide Name and the S-ID as the PortName of each connectable diskless PC. The LUN definition unit 12 compares the S-ID corresponding to the World Wide Name of each client PC stored as the host computer identifier owned by the LUN management table. Thereafter, only the S-ID is used to identify the host computer.

  The second step is LU search processing that is performed when SCSI is initialized. The fiber channel I / F control circuit 22 including the bootup control circuit 221 performs an LU presence check for all storages in order to detect an LU that can be booted up. Even when the bootup control circuit 221 is not provided, the OS performs a similar search when performing initialization processing of the fiber channel I / F. In any inspection, the inspection is performed by changing the LUN in order using the inquiry command defined in the SCSI specification. The detected LU is recognized as one of the bootable LUs in the bootup control circuit 221 and as one of the usable LUs in the OS.

  When the Fiber Channel I / F control circuit 15 of the storage 1 receives the Inquiry command from one diskless PC 2, it notifies the access control unit 16 of the issuing S-ID and the virtual LUN to be inspected. The access control unit 16 refers to the LUN management table 14 and checks the relationship between the virtual LUN and the host computer. If the combination exists in the table 14, this check is valid, and information indicating that an LU exists is transmitted to the diskless PC 2 as a result of the Inquiry command. On the other hand, if this combination does not exist in the table 14, since this LU is an LU for which access is prohibited from the diskless PC, information indicating that there is no LU such as a No Device response is transmitted. As a result of this operation, only internal LUs that are allowed access are recognized as virtual LUNs by LU inspection of diskless PC2, and other internal LUs are not recognized, so exclusive control for each diskless PC for access is performed. realizable.

  Thereafter, access is performed only in accordance with the relationship defined in the table 14 in principle. However, there is a possibility that access outside the range may occur exceptionally from the host computer. In order to suppress this, it is necessary to check the relationship between the S-ID and the virtual LUN for every access, but this can be realized by the above application.

When an LU access is issued from the diskless PC under the defined correct relationship, the access control unit 16 executes the specified command processing such as Read and Write.
(6) Storing OS and APP in dedicated LU Next, the most basic usage method of the present invention will be described. As shown in the example of the LUN management table 14 in FIG. 5, a dedicated LU is created in the storage 1 for each diskless PC 2 and all the OSs, applications, and data are stored. The method for using the shared LU will be described in the second and subsequent embodiments.

  First, the OS will be explained. Each OS is stored in virtual LUN = 0 assigned to each diskless PC2. When the diskless PC 2 is powered on, the fiber channel is initialized as described in (5) above, and the fiber channel I / F control circuit 22 checks the LU to be connected. Thereby, a dedicated LU is detected. Next, the bootup control circuit 221 executes OS bootup with the detected LU as a “boot disk”. Of all the detected LUs, it is necessary to specify in advance which LU will be the boot disk. In the example of the LUN management table 14 in FIG. 5, it is assumed that virtual LUN = 0 is set as the boot disk for both the diskless PCs 0, 1, and 2.

  The boot-up control circuit 221 emulates this LU as if it were a built-in disk device. As a result, the OS can be installed in this LU, and it is possible to boot up from here. This control is called boot emulation.

Also, each application and data can be stored and operated in this LU since this LU has already been emulated as a built-in disk device.
(7) Use of Privilege As shown in FIG. 5, the attribute of “privilege” can be given to the created internal LU. If the privilege attribute is set, access from the host computer can be temporarily suppressed and operations can be performed from the management console 4. Various LU maintenance can be performed using this.

  For example, it can be used for backup processing. Access to the backup target LU from the temporary diskless PC is prohibited, and this LU is placed under the management console 4. The management console 4 can connect this LU as an external disk device using a fiber channel I / F control circuit 45 and perform backup processing using a backup device (not shown) provided in the system.

  It can also be used to install OS and applications. For example, pre-installation work can be performed from the management console 4 when newly installing an OS. With this process, each diskless PC can use a new environment with minimum settings such as hardware information unique to each diskless PC and user registration when it is next started.

It can also be used to upgrade OS and applications. If version upgrade processing is executed for this LU from the management console 4, the contents of the disk will be updated, and the new version of OS and application will be available immediately from the next diskless PC2.
(8) Benefits As described above, multiple diskless PCs 2 share one storage 1 and create their own dedicated LUs to store their own OS and applications. I / O load is possible. Each data can also be stored in the same storage.

  As a result, the OS, applications, and data can be aggregated in one storage 1, so that the maintenance of the storage can be performed intensively under one administrator and the operation management cost can be reduced. There is.

  In addition, a high-speed, large-capacity, and highly-reliable storage such as a disk array can be used as the shared storage, and performance, capacity expandability, and reliability can be improved.

  In particular, since the program can be directly I / O loaded, there is an effect that high-speed program start can be realized without the overhead of LAN and server as in the conventional NC system.

  In addition, there is no need for a server that takes over the processing of each diskless PC, and the cost of the entire client computer system can be reduced.

(B) Embodiment 2
○ Configuration A second embodiment will be described. The configuration of this embodiment is the same as that of the first embodiment. However, in this embodiment, a shared LU that has not been described in the first embodiment will be described. In the following, it is assumed that the internal LU = 0 is a shared LU according to FIG.
○ Operation
(1) Application Sharing In the first embodiment, each diskless PC 2 has been described as having a dedicated LU and storing all of the OS, application, and data. In this embodiment, an example in which an application is stored in a shared LU will be described.
(2) Shared application setting information FIG. 7 is a schematic diagram showing the relationship between the diskless PC 2 and the internal LU. The shared LU 170 appears as virtual LUN = 1 from diskless PC # 0 (2a) and diskless PC # 1 (2b) and as virtual LUN = 2 from diskless PC # 2 (2c). The application program is loaded into the memory of the diskless PC2 and executed, and since the program itself is not changed, exclusive control between multiple diskless PC2s is not required. Each diskless PC2 # 0-2 (2a-2c ) Also does not recognize that it shares internal LU0. Such a shared application is called a shared application.

When using a shared application, unique setting information for each client PC 2 is required. The setting information includes use license information, personal information, various environment information, and the like. These pieces of information are stored in a dedicated LU for each diskless PC2. By configuring the shared application in this way, it is possible to construct a different application environment for each user while sharing it among a plurality of users, and to prevent unauthorized use.
(3) Effects When applications are stored in their dedicated LUs as in the first embodiment, maintenance such as installation and version upgrades must be performed on all LUs, which increases management costs. According to the method for storing the shared application in the shared LU according to the present embodiment, it is possible to consolidate the maintenance in one place and to reduce the operation management cost required for the maintenance.

(C) Embodiment 3
○ Configuration A third embodiment will be described. The configuration of this embodiment is the same as that of the first embodiment, as in the second embodiment.

  The difference from the second embodiment is that each diskless PC OS is also shared and stored in a shared LU. In the same manner as in the second embodiment, the following description will be given according to FIG. 5 assuming that the internal LU = 0 is a shared LU.

○ Operation
(1) OS sharing In general, client computer systems often use the same OS. Therefore, it is considered that this OS is also stored in the shared LU 170 and shared by a plurality of diskless PCs 2. Such an OS is called a shared OS.

(2) Setting Up a Virtual LUN for Bootup FIG. 8 is a schematic diagram showing the relationship between the diskless PC 2 and the internal LU of this embodiment. As in the second embodiment, the shared LU 170 appears as virtual LUN = 1 from the diskless PC # 0 (2a) and diskless PC # 1 (2b), and as virtual LUN = 2 from the diskless PC # 2 (2c).

  The shared OS is stored in the shared LU 170 in the same manner as the shared application. The difference from the second embodiment is that the boot LU set in the bootup control circuit 221 of the diskless PC 2 is diskless PC # 0 (2a), the diskless PC # 1 (2b) is set to virtual LUN = 1, and the diskless PC # 2 ( In 2c), virtual LUN = 2 is set. As a result, the bootup control circuit 221 of each diskless PC 2 controls the shared LU 170 as a virtual boot disk and starts the OS.

(3) Setting information for shared OS Like the setting information for shared application in the second embodiment, the setting information for each diskless PC 2 is required. The setting information includes use license information, personal information, various environment information, and the like. These are stored as shared OS setting information in the corresponding dedicated LU for each diskless PC 2 as in the shared application. By configuring the shared application in this way, it is possible to construct a different application environment for each user while sharing it among a plurality of users, and to prevent unauthorized use.

(4) Benefits If the OS is stored in each dedicated LU as in the first and second embodiments, maintenance such as installation and version upgrades must be performed for all LUs, increasing management costs. However, according to the method of storing the shared OS of the present embodiment in the shared LU, there is an effect that maintenance can be concentrated in one place and the operation management cost required for the maintenance can be further reduced.

(D) Fourth Embodiment ○ Configuration Next, a fourth embodiment will be described. In the present invention, as shown in FIG. 9, a cache 18 and an LU cache resident control unit 19 are provided in the configuration of the storage 1 of the first embodiment.

○ Operation
(1) Issues of shared LU Explain the operation. As described in the second and third embodiments, the management cost can be reduced by storing the shared application and the shared OS in the shared LU. On the other hand, accesses from the diskless PC 2 are concentrated in the same LU, which causes a new problem that performance degradation occurs. Therefore, a method for solving this performance degradation is provided.

(2) Cache The cache 18 is generally a circuit composed of DRAM, and is located between the LU and the host computer under the control of the central control unit 11 to achieve high speed access. When a read access request is generated, if a copy of the LU data exists in the cache 18 (cache hit), the data can be transferred from the cache 18 to the host computer without performing disk access. Further, when a write access request is generated, write data can be temporarily stored and the access can be terminated without performing disk access. This write data will eventually need to be written back to the LU. This write-back process is called delayed write.

(3) Cache resident control unit The cache resident control unit 19 stores all or part of the data of the shared LU of the second and third embodiments in the cache, and once stored in the cache 18, the cache resident control unit 19 The cache 18 is controlled so as to keep holding the data. When there is an access hit / miss determination request from the access control unit 16, the cache is inspected to determine a hit / miss.

  FIG. 10 is a configuration diagram of the LUN management table 14. The difference from the LUN management table 14 of the first to third embodiments is that a “cache attribute” field is provided.

  As in the first embodiment, the administrator also specifies a cache attribute when defining / creating an LU from the management console 4. In the figure, the “resident” attribute is set for the shared LU 170 (internal LUN = 0) that stores the shared application and the shared OS. On the other hand, the “normal” attribute is set for the other dedicated LUs.

  Usually, the LRU (Least Recent Use) method is used for cache control. This is a method in which newest data is stored by discarding data in the cache 18 in order from the least recently used data. Thereby, the cache hit rate can be effectively improved with a small cache capacity.

  On the other hand, the LU cache resident in this embodiment is a method of keeping the data of the designated LU regardless of the LRU once it is stored in the cache 18 without being discarded. By performing such control, for example, when a shared application with diskless PC # 0 (2a) is used, if this application is stored in the cache 18, the other diskless PC # 1 (2b) will then be the same. When using a shared application, high-speed access due to cache hits can always be realized. If control is performed using the LRU method, data used by diskless PC # 0 (2a) and other applications are stored in the cache 18, and the above shared application may be discarded from the cache 18. is there. In this case, when the diskless PC # 1 (2b) uses the shared application, the cache 18 again stores data, and the cache does not work effectively.

  In the above explanation, it was explained that only the data of the shared LU is resident in the cache 18. However, when sufficient cache capacity can be secured, or when you want to speed up a specific diskless PC 2 intentionally, etc. Of course, it is also possible to designate a dedicated LU as a cache.

(4) Effects According to the present embodiment, part or all of the contents stored in the shared LU can be resident in the cache, and as a result, the shared application and the shared OS can be resident in the cache with high access performance. Therefore, once one diskless PC2 uses a shared application or shared OS, it can obtain very fast access performance when another diskless PC uses the same application or the same OS. effective. As a result, in addition to reducing the operation and management costs of applications and OSs, there is an effect that higher performance can be realized. In addition, since a cache that is much faster than a disk device is utilized, there is an effect that each PC can increase the access speed when storing the OS and application in the built-in disk device.

The lineblock diagram of the client computer system of a first embodiment. The block diagram of the diskless PC of 1st embodiment. The block diagram of the storage of 1st embodiment. The block diagram of the management console of 1st embodiment. The block diagram of the LUN management table of 1st embodiment. The figure which shows the relationship between diskless PC and LU of 1st embodiment. The figure which shows the relationship between diskless PC and LU of 2nd embodiment. The figure which shows the relationship between diskless PC and LU of 3rd embodiment. The block diagram of the storage of 4th embodiment. The block diagram of the LUN management table of 4th embodiment. The block diagram of the NC system of a prior art.

Explanation of symbols

1 ... Storage
11 ... Central control unit
12 ... LU definition part
13… Communication control circuit
14… LUN management table
15 ... Fiber Channel I / F control circuit
16 ... Access control unit
17 ... LU
18 ... Cash
19 ... LU cache resident control section
2 Discless PC
21 ... Central control unit
22 ... Fiber Channel I / F control circuit
23 ... I / O devices
221 ... Boot-up control circuit
3 ... Fiber Channel connection circuit
4 ... Management console
41 ... Central control unit
42 ... I / O devices
43 ... Communication control circuit
44… LU management department
45 ... Fiber Channel I / F control circuit
46… Local disk
5 ... Fiber Channel
6 Communication path

Claims (13)

  A computer system for interconnecting a plurality of client computers, a storage shared by the client computers, a management console for managing the storage, and the plurality of client computers, the storage and the management console A computer system comprising a connecting device for the purpose. With multiple calculators,
A storage system connected to the plurality of computers and having at least one disk device and a cache memory;
The storage system has a plurality of logical units on the disk device,
A logical unit that is shared among the plurality of logical units is set for a plurality of computers,
Each logical unit is assigned an internal logical unit number and a virtual logical unit number,
The internal logical unit number is a number used for managing logical units in the storage system, and different internal logical unit numbers are assigned to the plurality of logical units,
The virtual logical unit number is a number used by the computer to confirm the logical unit accessed from the computer.
One or more logical units accessed from each computer are assigned a virtual logical unit number starting from 0 for each computer, and virtual logical unit number 0 is shared by a plurality of computers for each computer. Assigned to the logical unit
A system in which data of the shared logical unit is resident in the cache memory. The system of claim 2, comprising:
The storage system stores a correspondence relationship between an internal logical unit number, a virtual logical unit number, and a computer using the virtual logical unit number. The system according to claim 3, wherein
Each of the plurality of computers, when searching for a logical unit accessible from the computer, starts a search from a logical unit having a virtual logical unit number of 0 and searches for logical units having consecutive virtual logical unit numbers. A system characterized by The system of claim 2, comprising:
Each of the plurality of computers searches for a logical unit accessed from the computer and executes a boot-up process of an operating system stored in the detected logical unit. The system of claim 2, comprising:
Each computer performs boot-up processing of an operating system stored in a shared logical unit among a plurality of computers having a virtual logical unit number of 0. A storage system connected to a plurality of computers and having at least one disk device, a cache memory, and a control unit that controls the disk device, the device, and the cache memory,
A plurality of logical units on the disk device;
Among the plurality of logical units, a shared logical unit is set for a plurality of computers.
Each logical unit is assigned an internal logical unit number and a virtual logical unit number,
The internal logical unit number is a number used for managing logical units in the storage system,
The virtual logical unit number is a number used when each computer confirms the logical unit accessed from the computer.
One or more logical units accessed from each computer are assigned a virtual logical unit number starting from 0 for each computer, and virtual logical unit number 0 is shared by a plurality of computers for each computer. Assigned to the logical unit
The storage system according to claim 1, wherein the shared logical unit data resides in the cache memory. The storage system according to claim 7, wherein
The control system stores a correspondence relationship between an internal logical unit number, a virtual logical unit number, and a computer using the virtual logical unit number. The storage system according to claim 7, wherein
A storage system, wherein an operating system used by a plurality of computers is stored in a logical unit shared by a plurality of computers. A method of using a storage system having a plurality of logical units by a plurality of computers,
Providing the plurality of logical units in at least one disk device of the storage system;
An internal logical unit number assigning step for assigning an internal logical unit number used for managing logical units in the storage system to each of the plurality of logical units;
For each of the plurality of logical units, setting a computer that accesses the logical unit;
Virtual logical unit number assignment that assigns a virtual logical unit number used when each of the plurality of computers detects a logical unit accessed from the computer to a logical unit accessed from the computer among the plurality of logical units. Steps,
For each of the plurality of computers, a logical unit detection step of detecting a logical unit accessed from the computer using the virtual logical unit number,
The virtual logical unit number assignment step includes a step of assigning a virtual logical unit number starting from 0 for each computer to one or more logical units for each computer accessed from each computer, and the virtual logical unit number 0 is Including a step assigned to a logical unit shared by a plurality of computers,
Furthermore, the method of using the storage system, further comprising the step of making the data of the shared logical unit resident in a cache memory included in the storage system. A method of using the storage system according to claim 10,
Furthermore, the storage system has a step of storing a correspondence relationship between an internal logical unit number, a virtual logical unit number, and a computer using the virtual logical unit number. A method of using a storage system according to claim 11,
In the logical unit detection step, each computer starts a search from a logical unit having a virtual logical unit number of 0, searches for logical units having a continuous virtual logical unit number, and the logical unit having a certain virtual logical unit number A method of using a storage system, comprising a step of stopping a search for a logical unit when a response indicating that a unit does not exist is received. A method of using the storage system according to claim 10,
The use of the storage system further comprising a step of executing boot-up processing of an operating system stored in a logical unit whose virtual logical unit number detected by each of the plurality of computers is 0 Method.

Images