JP2000311112A - Information system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a cache fast write(CFW) function between plural main frame hosts having an access interface for a fixed length. SOLUTION: When there is a limitation that an application program to utilize the CFW function is to be operated on one host from processing start to end, a CFW ID is not shared but independently managed by respective hosts an only when CFW data managed by a certain host are erased, the CFW ID of the present host is updated but when CFW data on the other host are erased, the CFW ID is not updated. Besides, one CFW ID is shared for each subsystem between the plural main frame hosts, the initial values of all the sub-systems are set to respective hosts at the time of system start, and at timing when the CFW data are erased on a certain host, updating of the CFW ID is reported to the other host so that the same type of a CFW ID value can be maintained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information system, and more particularly, to an information system including a plurality of mainframe hosts and a storage device shared by the mainframe hosts, and realizing a cache high-speed writing function.

[0002]

2. Description of the Related Art In a so-called mainframe computer on which an operating system such as IBM OS / 390 or Hitachi VOS is mounted, a magnetic disk device having a variable-length record data format is applied as an external storage device. The access to the magnetic disk device includes CK that enables data access of variable-length records.
A D (Count Key Data) interface is used.
Conventionally, data in the variable-length record format uses this C
It has been physically implemented on a disk device having a KD interface.

However, in recent years, as a storage device for variable-length record data, a RAID comprising a relatively inexpensive disk device used in a PC or a workstation has been proposed.
(Redundant Array of Inexpensive Disks) has been widely used. RAID has been announced by D. Patterson, G. gibson and RH Kart at the ACM SIGMOD conference at the University of Illinois in Chicago.
z; A Case for Redundant Arrays of Inexpensive Disk
s (RAID), ACM SIGMOD Conference, Chicago, IL,
(June 1988), pp. 109-116. In RAID, by connecting a plurality of relatively inexpensive and low-reliability disk devices and distributing and allocating data to the plurality of disk devices, the performance is improved by parallelizing data transfer, and the redundant data is retained. High availability is achieved.

The disk drive used for RAID is S
It has an interface such as CSI, and has an FBA (Fixe Blo
ck Architecture) has a fixed-length data format. Therefore, the storage subsystem that receives variable-length record format data from the CKD interface and writes the data to the FBA-format disk device,
A conversion mechanism for conversion to the BA format and conversion from the FBA format to the variable-length record format is provided internally. As an example of such a format conversion mechanism, for example, an apparatus disclosed in Japanese Patent Application Laid-Open No. Hei 6-150557 is known.

On the other hand, recently, an external storage device interface of a mainframe computer having a SCSI interface has appeared. For example, IBM Multiprise2000 described in “Mainframe '98” Nikkei BP, pp.53-54, has a SCSI interface, incorporates an FBA format disk device in the housing, and generates a variable A processor specializing in storage control interprets a command group for long data access (CCW: Channel Common Work), converts it into a command group for fixed length access (SCSI command), and inputs / outputs to / from an FBA format disk device. Execute the process. With this conversion mechanism, the conventional C
The FBA disk device can be used without changing application programs, OS, and the like, which are based on the KD disk device. Multiprise 2000
Is a registered trademark of IBM Corporation. In order to extract variable-length record format data to an FBA format disk device, the record to be accessed must be an FBA format.
You need to find out where it is on the format. In this prior art, when a write request occurs,
There is also disclosed a technique for once reading track data including a record to be accessed from an FBA disk device, calculating a record position on a cache, and writing the record. However, if the position of the record to be written can be obtained in advance by calculation, the data is not read.

[0006] In an external storage device for a mainframe host, a cache high-speed write function (CFW: Cache
e Fast Write). This CFW function is to store, in an external storage device in which a cache is mounted in the device, specific write data specified by the host only in the cache, to avoid the overhead required for accessing the disk medium. It is mainly applied to temporary data retention during application processing. In the CFW function, the CFW stored in the cache
Data is allowed to be lost due to power failure or device failure, but the external storage device must report the loss of CFW data to the host without fail. The following method is used for reporting to the host. When starting processing using the CFW function, the host receives a value called a CFW ID from an external storage device and holds the value in a main storage or the like. The CFW ID is set for each storage subsystem composed of an external storage and its control mechanism, and is added when CFW data managed by the external storage is lost. At the time of access from the host by the CFW processing, the CFW ID held by the host is added. If the CFW ID differs from the current CFW ID held by the external storage device, the CFW data used for the processing is added. Is reported to the host. With this report, the application is interrupted and executed again from the beginning of the process.

[0007]

SUMMARY OF THE INVENTION The aforementioned IBM Multip
The rise 2000 relates to a built-in disk, but if this conventional technique is applied, an external storage device having an FBA access interface can be connected to a mainframe computer. FBA disk drive is CK
Compared with the D disk device, since it is supplied from many vendors, it is possible to select a model from various variations in price, performance, and reliability. Also, PC
Given the current situation in which business coordination between a PC and a mainframe is becoming indispensable due to the increase in the amount of data handled and the higher functionality and higher performance of applications, the use of FBA disk devices that can be connected to a PC has been increasing. The benefits of being able to connect to the mainframe are huge.

It is premised that such an external storage device is shared by a plurality of hosts in order to share data and reduce device costs. For example, a storage subsystem of the CKD interface conventionally connected to a mainframe computer usually has a plurality of channel paths, is simultaneously connected to a plurality of mainframe hosts, and shares a volume arranged therein. Therefore, F
The external storage device of the BA format is also shared and operated between a plurality of mainframe computers.

On the other hand, in a mainframe computer having an access interface to an external storage device in the FBA format, a variable-length record format and an FBA
Due to the nature of the format conversion process, it is necessary to have a cache memory that can be accessed from the processor in charge of the external storage device control. In the above prior art, a part of the main memory is used as a cache memory. Since the data in the cache memory can be accessed without passing through the channel, the read performance can be significantly improved by holding a large amount of data in the cache memory. At the stage where the write data is held in the cache memory, the completion of the write process is reported to the CPU, and thereafter, the held update data is asynchronously written to the external storage device. Time can be greatly reduced. The performance improvement by the data access without passing through the channel is one of the advantages of the prior art.

Therefore, when an FBA disk device is shared by a plurality of mainframe computers as an external storage device, each mainframe computer has a variable-length format and FBA format conversion mechanism, each of which locally holds a cache memory. Will be. That is, a plurality of storage control mechanisms and cache memories independent of hardware are connected to one external storage device.

By the way, in the mainframe computer, the past program assets are emphasized, and the interface of the storage device is required to maintain the compatibility with the past model. In terms of functions, supporting functions used in frequently used applications greatly contributes to improving the competitiveness of products. Therefore, the support of the CFW function is desired also in the information system in which the FBA disk device is shared by a plurality of mainframe computers as the object of the present invention.

First, in the same manner as in the prior art, when there is a restriction on the execution of an application program using the CFW function so that it runs on one mainframe computer at all times, cache management is performed for each mainframe computer. Even if the CFW data in the cache memory of one mainframe computer is lost, the CF operating on another mainframe computer
There is no need to affect applications using the W function. That is, applications using the CFW function that operate on each mainframe computer can operate independently.

However, according to the conventional definition of the storage device having the CKD interface, the CFWID has one value for each storage device subsystem, and therefore, one CFWID is provided between each mainframe computer for each storage device subsystem. The CFW ID will be shared. In this case, when the CFW data is lost in the cache memory of one mainframe computer, the CFWID of the storage subsystem is updated, and the CFW data of the same storage subsystem is updated on another mainframe computer. If the application program using the logic is operating, the C of the storage subsystem
The process is interrupted due to the change of the FW ID. When an application program using the CFW function operates across a plurality of mainframe computers in order to distribute the load in the information system, the CFW is independently provided for each mainframe computer.
When the ID is managed, every time the mainframe computer on which the program operates changes, the C
The FW ID does not match the CFW ID held by the mainframe computer, and the processing is interrupted. Further, the CFW data held in the cache of a certain mainframe computer cannot be accessed by transmitting to another mainframe computer.

Therefore, the first problem to be solved by the present invention is as follows:
In an information system in which a fixed-length format external storage device is shared between mainframe computers having a fixed-length access interface, a restriction is provided such that an application using the CFW function is continuously operated on one mainframe computer. Another object of the present invention is to prevent interruption of CFW processing in operation on another mainframe computer due to loss of CFW data in one mainframe computer. Further, a second object of the present invention is to share a CFW ID among a plurality of mainframe computers and further share CFW data in the same information system, so that a CFW function over a plurality of mainframe computers can be achieved. Is to enable the execution of application programs using

[0015]

According to the present invention, the CFW ID for the same storage subsystem is managed independently by each mainframe computer and is not shared. The CFW ID of each mainframe computer is updated only when CFW data managed by itself is lost due to a cache memory failure of the own device, and is not updated even when CFW data of another mainframe computer is lost. Control. As described above, it is possible to avoid interruption of CFW processing in operation on another mainframe computer due to loss of CFW data in one mainframe computer.

In order to solve the second problem, according to the present invention, CFW data written in a cache of a certain mainframe computer is transferred to all the other mainframe computers and held in the cache. The CFW ID for the storage subsystem is shared by multiple mainframe computers. here,
As means for sharing, two means are conceivable. One means stores the initial value of the CFW ID in a data area on a main memory managed by each external storage device control mechanism when the information system is started. When the CFW data managed by each mainframe computer is lost, the update of the CFW ID of the storage subsystem is notified to all other mainframe computers, and the CFW ID held locally by each mainframe computer is updated. , Maintain value identity. Another means is a shared device shared by each mainframe computer (such as an external storage device for storing data of a storage subsystem).
Keep the CFW ID above and thereby share the CFW ID. As described above, application programs using the CFW function can be executed across a plurality of mainframe computers.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a first embodiment of the present invention will be described.
A third embodiment will be described. Note that the present invention is not limited by this.

In the first embodiment, in an information system composed of a plurality of mainframe hosts sharing an external storage device, an application program utilizing the CFW function runs on one host from one end to the other, and the other host processes during processing. Not to In the CFW processing executed by each host, the CFW data is stored only in the cache memory of the host. Each host manages the CFW ID independently of each other, updates the CFW ID only when the CFW data in the host is lost, and updates the CFW ID of the other host.
Control is performed so as not to be updated even if data loss occurs (therefore, the CFW ID is not shared). In the second embodiment, in an information system including a plurality of mainframe hosts sharing an external storage device, in the CFW processing executed by each host, the CFW data is stored not only in the host itself but also in the caches of all other hosts. It is also stored in memory. Then, at the time of system startup, each host
After initializing the ID, when one host updates the CFW ID, it instructs another host to update the CFW ID,
As a result, one CFW ID is shared between all the hosts. In the third embodiment, in an information system including a plurality of mainframe hosts sharing an external storage device, in a CFW process executed by each host, CFW data is stored not only in its own host but also in caches of all other hosts. It is also stored in memory. Then, the CFW ID is held in the shared device, and when a certain host updates the CFW ID, the other host is instructed to take in the CFW ID from the shared device.
Share D.

First Embodiment A first embodiment will be described with reference to FIGS. 1 to 9. FIG. 1 is a block diagram of an information system 1 according to the first embodiment of the present invention. This information system 1
Two or more mainframe hosts 100, 100, and one or more external storage devices 150 of fixed-length format shared by the mainframe hosts 100, 100;
And a communication line 110 for connecting the mainframe hosts 100 and 100.

The mainframe host 100 has one or more CPUs 101, one or more storage control processors 102, a main memory 103, and a channel 104.

The external storage device 150 has a fixed-length access interface, and converts CKD format volume data recognized by the OS into FBA format and stores it. The external storage device 150 may be a single disk device or a RAID subsystem having a fixed-length access interface. A volume of the CKD interface and a conventional storage controller connected to the volume are called a subsystem. The external storage device 150 can store data of volumes for a plurality of subsystems. In this embodiment, it is assumed that the data of one subsystem is stored in the entire external storage device 150 connected to the information system 1.

The communication line 110 enables information exchange between the OSs of the mainframe hosts 100, input / output control programs, and between the OS and the input / output control programs. The communication between different mainframe hosts 100 includes, for example, exclusion of a data area between OSs. Note that the communication line 110 is not used for realizing the CFW function in the first embodiment.

The CPU of the mainframe host 100
In 101, an application program such as a database operates. Control of memory allocation and the like when the application program operates and management of the file system and the like are performed by the OS. Access to the external storage device 150 from the application program is performed via the OS.

In the storage control processor 102, the OS
An input / output control program that executes the input / output processing to / from the external storage device 150 via the CPU is running. The input / output control program has a function of converting between a variable-length record format and an FBA format, receives a read / write request command (CCW) sent from the OS to the external storage device 150 via the CKD interface, and accesses the FBA format. It is converted into an interface command (SCSI command) and transmitted to the external storage device 150. By this conversion, the data of the volume in the CKD format recognized by the OS is stored in the external storage device 150 in the FBA format, and the input / output control program also manages the mapping. For this management, a conventionally known method can be used, and a description thereof will be omitted. Also,
The input / output control program also controls data transfer of input / output data between the external storage device 150 and the main storage 103.

The main memory 103 is allocated a cache memory 105 used for input / output with the external storage device 150, in addition to an area for storing data and programs used by each CPU 100. Also, in the main memory 103,
Cache management information 106 for managing data on the cache memory 105, and CFW ID used for the CFW function
107 is also stored.

The channel 104 connects the mainframe host 100 and the external storage device 150, and the F
Performs a BA format access interface command and data transfer.

The cache memory 10 of the main memory 103
5 includes data read from the external storage device 150 and O
S temporarily holds data to be written to the external storage device 150. FIG. 2 shows an exemplary configuration of the cache memory 105. The cache memory 105 stores the segment 20
It is equally divided into small areas called zero. One segment 200 can store one track of data, such as data read from the external storage device 105 or the external storage device 10.
No. 5 and the like are not stored. When storing data in the cache memory 105, the data is stored in the same FBA format as when storing the data in the external storage device 105. Usually, in the access interface to the variable-length format, the minimum access unit is not a track but a record. For this reason, in the case of accessing a record in the middle of a track instead of accessing from the beginning of the track, it is necessary to find the position of the record in the FBA format from the beginning of the track. When the records in the track satisfy conditions such as equal length, the record position can be specified using a conventionally known method. However, when these methods cannot be used, it is necessary to read the data of the track from the external storage device 150. In the present embodiment, for the sake of simplicity, the occurrence of this read is avoided by using the minimum access unit as a track instead of a record.

The cache management information 106 is information used by the input / output control program to manage the cache memory 105, and stores management information such as the state of segment allocation to each track.

FIG. 3 is a configuration diagram showing information related to the present invention in the cache management information 106. The cache allocation information 300 indicates whether the corresponding segment 200 has been allocated to each track of the external storage device 105. The segment management information 301 includes the segment 20
The status of the data stored in the segment 200 is managed in a one-to-one correspondence with 0. The valid segment management information queue pointer 302 is stored in the external storage device 105.
Segment management information 301 assigned to the track
Is the root pointer of the effective segment management information queue to which is connected. Free segment management information queue pointer 3
03 is the segment management information 3 not allocated to the track.
01 is the root pointer of the free segment management information queue to which the "01" is connected.

FIG. 4 shows the cache allocation information 300.
FIG. Cache allocation information 300
Has pointers for all tracks in the external storage device 105. The pointer value to the segment management information 301 of the segment 200 is stored in the pointer of the track to which the segment 200 has been allocated, and the NULL value is set to the pointer of the track to which the segment 200 has not been allocated.

FIG. 5 is a configuration diagram showing information related to the present invention in the segment management information 301. The segment pointer 500 stores the address of the corresponding segment 200. The valid flag 501 indicates whether the data stored in the segment 200 is valid. The valid flag 5 is stored when data read from the external storage device 150 is stored in the segment.
02 is set to ON, and is set to OFF when releasing the cache allocation of the segment. Unreflected flag 5
02 indicates that the data stored in the segment 200 has not been reflected in the external storage device 150. CFW
The flag 503 indicates that the data stored in the segment 200 is data written by applying the CFW function. The segment queue pointer 504 is a pointer used when connecting the segment management information 301 to a valid segment management information queue or an empty segment management information queue.

Both the non-reflection flag 502 and the CFW flag 503 in the segment management information 301 indicate that the relevant segment has not been reflected in the external storage device 150, but the correspondence for data guarantee is different. Unreflected data flag 5
It is necessary to guarantee that data stored in the segment 200 in which 02 is set is written to the external storage device 150. Therefore, although the data of the segment 200 is temporarily stored in the segment 200, it is necessary to write the data to the external storage device 150 in synchronization with the write instruction from the OS, and to report the completion of the write processing after the write is completed. .
On the other hand, the segment 20 in which the CFW flag 503 is set
0 is basically stored in the cache memory 10.
5, writing to the external storage device 150 need not be guaranteed. Therefore, CFW flag 5
The segment 200 in which 03 has been set is not discarded and continues to be held until the writing to the external storage device 150 is completed even after the completion of writing to the OS is reported.

FIG. 6 shows an example of the configuration of a valid segment management information queue. The segment 200 for storing disk-reflected data such as read data controls the cache retention period by a conventionally known method such as LRU management. In the present embodiment, all valid segment management information 301 is managed in one queue regardless of the attribute, but may be managed in another queue according to the attribute of the segment management information 301.

The CFW ID 107 is conventionally CKD
In a subsystem composed of a volume and a storage control device that controls the volume, it is used to determine the normality of CFW data held in a cache. The value does not change as long as the CFW data remains normal, and is updated when the CFW data is lost due to a cache failure or the like. The application program using the CFW function takes in the CFW ID at the start of the processing, and thereafter adds the CFW ID to the processing request command and transmits it. C
When the FW ID changes and a CFW processing request command to which the value before the update is added is sent, the storage controller notifies the application program that the CFW data used in the processing has disappeared from the cache, Interrupt the process. As described above, since the information of one subsystem is stored in the information system 1 of the present embodiment, there is only one CFW ID.

Next, the operation of the storage control processor 102 will be described. FIG. 7 is an explanatory diagram showing the operation when the CFW data is lost in the right host 100 while the left host 100 is executing the CFW processing. Host 1 on the left
00, in response to a write processing (hereinafter, CFW write) request using the CFW function from the CPU 101,
CFW ID added to W write request and main memory 103
Is compared with the CFW ID 107 held in the other host 100, and if they match, the segment 200 storing the request data is discarded from the cache memory 105 of the other host 100, and the request data is allocated to the segment 20
Store to 0. The host 100 on the right side detects the loss of the CFW data and detects the CFW data stored in the main memory 103.
The ID 107 is updated. The hosts 100 operate independently of each other, and the CFW ID update on the right does not interrupt the CFW write processing on the left.

FIG. 8 is a processing flowchart of the CFW processing in the present embodiment. In step 801, the CPU 1
01 added to the CFW write request transmitted from 01
The W ID is compared with the CFW ID 107 stored in the main memory 103. If the ID values match, the process proceeds to step 802. If the ID values do not match, the process proceeds to step 805.

In steps 802 to 804,
The CFW write process is executed. In step 802, the data of the requested track on the cache memory 105 of the other host is discarded. Specifically, the cache memory 105 of another host is checked, and if there is unreflected data in the external storage device 150, the data is written to the external storage device 150, and the data reflected in the external storage device 150 is allocated. Segment 20
Discard 0. Discarding a segment 200 means that the valid flag 501 of the corresponding segment management information 301 is turned off, the segment management information 301 is moved from the valid segment management information queue to the free segment management information queue, and the corresponding track of the cache allocation information 300 is deleted. Clearing the pointer to be performed. Step 8
At 03, if the segment 200 is not assigned to the track, the segment 200 is newly assigned, and the write data is stored in the segment 200. To newly allocate the segment 200, the segment management information 301 connected to the free segment management information queue is removed from the queue, reconnected to the valid segment management information queue, and the entry of the track in the cache allocation information 300 is added to the segment management information 301. This is done by setting an address. At this time, the CFW flag 50 of the segment management information 301 of the segment 200 in which the CFW data is stored.
Turn 3 on. In step 804, the requested CF
The completion of the W write process is reported to the CPU 101. Then, the process ends.

In step 805, the above-mentioned step 801 is executed.
By comparing the CFW ID of
Recognizing that the CFW data has been lost during the execution of the FW processing, the CPU 101 notifies the CPU 101 that the CFW write request cannot be processed due to a change in the CFW state, and interrupts the CFW processing being executed by the application. And
The process ends.

FIG. 9 shows the CFW I in this embodiment.
It is a processing flow figure of D update processing. In step 901, it is recognized that the CFW data held in the cache memory 105 of the host 100 has been lost, and the main memory 1
03 is updated. As a result of this update, if an application using the CFW function is being processed on the host 100, the application suspends processing.

According to the first embodiment, the loss of CFW data in one host 100 causes another host 100
Can be prevented from interrupting the operating CFW process.

Second Embodiment A second embodiment will be described with reference to FIGS. 10 to 14. The information system according to the second embodiment is basically the same as the information system 1 shown in FIG. In the second embodiment, the communication path 110 is used for transferring CFW data to all other hosts and transmitting a CFW ID update instruction. At this time, the communication line 110 may have a broadcast function capable of transferring data from one host 100 to all other hosts 100. In the second embodiment, an operation of the storage control processor 102 when a write process using the CFW function is requested will be described first. Next, the CFW ID between the multiple hosts 100
Will be described.

FIG. 10 shows that the host 100 on the left side
Storage control processor 10 when write processing is executed
2 is a diagram illustrating an operation of FIG. On the host 100 on the left,
In response to a CFW write request from CPU 101, CFW
The IDs are compared, and if they match, the request data is stored in the allocated segment 200. Thereafter, the CFW data is transferred to all the other hosts 100, and the CFW data is transferred to the cache memory 105 of all the other hosts 100.
Holds W data. After this process, even if the application using the CFW data moves from the left host 100 on which the application using the CFW data is running to the right host 100,
The CFW data in the cache memory 105 can be accessed.

FIG. 11 is a processing flowchart of the CFW processing in the present embodiment. In step 1101, the CP
CFW ID added to the CFW write request transmitted from U101 and CFW ID10 stored in main memory 103
Compare 7 If the ID values match, step 110
Then, the process proceeds to step 1105 if they do not match.

In steps 1102 to 1104, the CFW write process is executed. Step 110
In step 2, if the segment 200 is not assigned to the track, the segment 200 is newly assigned and the write data is stored in the segment 200. At this time, the segment management information 301 of the segment 200 storing the CFW data
Is turned on. Step 1103
Then, the CFW data is transferred to all the other hosts 100 via the communication path 110. At this time, all other hosts 1
00 may be transferred one-to-one with the communication channel 1
By adopting a network having a broadcast function as 10, the CFW data can be transferred to all other hosts 1.
00 may be broadcast. Step 1104
Now, let the CPU know that the CFW write process for the request has been completed.
Report to 101. Then, the process ends.

In step 1105, step 1101
By comparing the CFW ID of
Recognizing that the CFW data has been lost during the execution of the FW processing, the CPU 101 notifies the CPU 101 that the CFW write request cannot be processed due to a change in the CFW state, and interrupts the CFW processing being executed by the application. And
The process ends.

FIG. 12 is a processing flowchart of the CFW data receiving processing in the present embodiment. Step 1201
Then, the CFW data transferred from the other host 100 is received. The received data is temporarily stored in an area on the main memory 103. In step 1202, if the segment 200 is assigned to the track, the CFW data is stored in the segment 200. If not, the segment 200 is newly assigned.
The CFW data is stored in the allocated segment 200. At this time, the CFW flag 503 of the segment management information 301 of the segment 200 is turned on.

FIG. 13 shows that the host 100 on the right side
FIG. 9 is a diagram illustrating an operation of the storage control processor 102 when data is lost. In the second embodiment, the CFW held by each host 100 when the information system is started up
The ID 107 is initialized with the same initial value. Host 1 on the right
00, the CFW stored in the cache memory 105
Detects data loss and stores CFW in main memory
The ID 107 is updated. After that, an instruction to update the CFW ID is transmitted to all other hosts.

FIG. 14 shows the CFW in this embodiment.
It is a processing flow figure of ID update processing. Step 1401
If the event that caused the activation of this processing is that the CFW data of the host 100 has been lost, then step 1 is executed.
The process transitions to 402, and if a CFW ID update instruction has been received from the other host 100, the process proceeds to step 1404.

In step 1402, it is recognized that the CFW data stored in the cache memory 105 of the host 100 has been lost, and the CFW stored in the main memory 103 is recognized.
The FWID 107 is added (updated). Step 1403
Then, an instruction to update the CFW ID is transmitted to all the other hosts 100. This CFW ID update instruction is C
It consists of a processing request identifier indicating the update of the FW ID and an updated value of the CFW ID. In addition, as an update instruction, CFW ID
May be transmitted instead of the updated value of the CFW ID. Also, as an update instruction, CFW ID
May be transmitted alone. Then, the process ends.

In step 1404, the CFW ID update value attached to the CFW ID update instruction from the other host 100 is compared with the CFW ID 107 stored in the main memory 103 of the host 100, and the CFW ID is updated. If the value is larger than the CFW ID 107 of the host 100, the flow goes to step 1405;
The process ends without updating the ID. In addition, CFW ID
Since is a finite size counter, it is necessary to make a comparison in consideration of the possibility that the apparent magnitude relationship may be reversed due to overflow. A conventionally known method can be used for this magnitude comparison. In step 1405, the updated value is registered as the CFW ID 107 in the main storage 103, and the process ends.

According to the second embodiment, it is possible to execute an application program using the CFW function over a plurality of hosts 100.

Third Embodiment A third embodiment will be described with reference to FIG. 15 to FIG. The information system according to the third embodiment is basically the same as the information system 1 shown in FIG. In the third embodiment, the communication path 110 is used for transferring the CFW data to all other hosts and transmitting an instruction to capture the CFW ID from the external storage device. At this time, the communication line 110 may have a broadcast function capable of transferring data from one host 100 to all other hosts 100. In the third embodiment, the operation of the storage control processor 102 when a write process using the CFW function is requested is the same as that of the second embodiment, and thus the description is omitted, and the CFW between a plurality of hosts 100 is omitted. Only the ID sharing process will be described.

FIG. 15 shows that the host 100 on the right side
FIG. 9 is a diagram illustrating an operation of the storage control processor 102 when data is lost. In the third embodiment, the CF stored in the external storage device 150 when the information system is started up.
W ID 1507 and CFW I held by each host 100
D107 is initialized with the same initial value. Host 10 on the right
0, the loss of the CFW data stored in the cache memory 105 is detected, and the CFW stored in the main memory is detected.
The ID 107 is updated. After that, the CFW ID is written in the external storage device 150, and the CFW ID is sent to all other hosts.
An instruction to capture the ID 1501 is transmitted. In the present embodiment, the CFW ID is stored in the external storage device 150, but may be stored in a communication control device shared by all the mainframe hosts 100. In addition, CFW
By holding the ID in a non-volatile medium such as the external storage device 150, the CFW I
It becomes possible to take over the D value. In particular, when data managed by this information system is connected to a mainframe host 100 having only a CKD interface,
The present information system plays the role of a storage controller in the related art.
In order to maintain the W ID invariance, the CFW ID needs to be taken over. Further, in the present embodiment, each host 1 is updated because the trigger for updating the CFW ID is extremely rare, and it is necessary to reduce the reference overhead.
00 has the CFW ID 107 in its main memory,
When the data is stored in a communication control device or the like that requires a small reference overhead, the data may be stored in the main memory 103 of each host 100 and may be taken from the shared device every time the CFW ID is referred.

FIG. 16 shows the CFW in this embodiment.
It is a processing flow figure of ID update processing. Step 160
In step 1, subsequent processes are separated according to the event that has caused the process to be activated. CF of the host 100
If the W data has been lost, the process proceeds to step 1602, and if a CFW ID update instruction from another host 100 has been received, the process proceeds to step 1605. In step 1602, the main memory 1 recognizes that the CFW data stored in the cache memory 105 of the host 100 has been lost, and
03 is updated. In step 1603, the updated CFW ID 107 is written to the external storage device 150. In step 1604, an instruction to update the CFW ID is transmitted to all the other hosts 100. In step 1605, the external storage device 150 is controlled according to the CFW ID update instruction from the other host 100.
From the main memory 103.

According to the third embodiment, it is possible to execute an application program using the CFW function over a plurality of hosts 100.

-Other Embodiments- In the second and third embodiments, data is transferred using the communication path 110. However, separately from the communication path 110, all the hosts 100 and the external storage device 150 are connected between the hosts. The network may be connected to a data transferable network, and the CFW data may be transferred via the network. Further, this network may be a network capable of broadcasting. At this time, the CFW is stored in the external storage device 150.
By not transferring data, the external storage device 1
It is possible to avoid accessing 50.

[0057]

According to the information system of the present invention, when an external storage device having a fixed-length interface is shared among a plurality of mainframe hosts, an application program using the CFW function is used from the start to the end of the processing. In the case of operating on one mainframe host, it is possible to avoid interruption of CFW processing in operation on another host due to loss of CFW data on one host. Further, according to the information system of the present invention, CFW is performed between a plurality of hosts.
The ID and CFW data can be shared, and application programs using the CFW function can be executed across a plurality of hosts.

[Brief description of the drawings]

FIG. 1 is a block diagram of an information system according to a first embodiment of the present invention.

FIG. 2 is a diagram illustrating a configuration example of a cache memory;

FIG. 3 is a diagram illustrating a configuration example of cache management information;

FIG. 4 is a diagram illustrating a configuration example of cache allocation information;

FIG. 5 is a diagram illustrating a configuration example of segment management information;

FIG. 6 is a diagram illustrating a configuration example of a valid segment management information queue.

FIG. 7 is an explanatory diagram showing a process for a write request using a CFW function and an operation of a storage control processor for a CFW ID update process due to loss of CFW data in the first embodiment.

FIG. 8 is a processing flowchart of CFW processing in the first embodiment.

FIG. 9 is a processing flowchart of a CFW ID updating process in the first embodiment.

FIG. 10 is an explanatory diagram showing an operation of a storage control processor at the time of processing for a write request using a CFW function in the second embodiment.

FIG. 11 is a processing flowchart of CFW processing according to the second embodiment.

FIG. 12 is a processing flowchart of CFW data reception processing in the second embodiment.

FIG. 13 is an explanatory diagram showing an operation of a storage control processor for a CFW ID update process due to loss of CFW data in the second embodiment.

FIG. 14 is a processing flowchart of a CFW ID update processing in the second embodiment.

FIG. 15 is an explanatory diagram showing an operation of the storage control processor for a CFW ID update process due to a loss of CFW data in the third embodiment.

FIG. 16 is a process flowchart of a CFW ID update process according to the third embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Information system 100 ... Mainframe host 101 ... CPU 102 ... Storage control processor 103 ... Main memory 104 ... Channel 105 ... Cache memory 106 ... Cache management information 107 ... CFW ID 110 ... Communication line 150 ... External storage device 200 ... Segment

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Takashi Oeda 1099 Ozenji Temple, Aso-ku, Kawasaki City, Kanagawa Prefecture Inside System Development Laboratory, Hitachi, Ltd. (72) Inventor Takahiko Shoyama 1st Horiyamashita, Hadano City, Kanagawa 5B005 JJ01 JJ11 KK13 MM11 NN13 NN14 5B045 DD12 DD15 GG07 5B065 BA01 CE11 CH01 CS06

Claims (9)

[Claims] 1. An information system comprising a plurality of mainframe hosts and at least one storage device shared by the plurality of hosts and storing data of a fixed-length format, wherein the mainframe host comprises: An operating system and an input / output control program that performs format conversion for converting variable-length data to fixed-length data or for converting fixed-length data to variable-length data, and that processes input and output of the operating system to and from the storage device. A first mainframe host having at least one operating processor and a cache memory used for input / output of the storage device; The cache in a first mainframe host Memory, and does not store it in the cache memory in another second mainframe host.In each of the mainframe hosts, independently holds and manages a cache high-speed write identifier used in a cache high-speed write function, When the cache high-speed write data held in the cache memory of the first mainframe host is lost, only the cache high-speed write identifier managed by the first mainframe host is updated, and the other high-speed write identifier is updated. An information system wherein the cache high-speed write identifier managed by the second mainframe host is not updated. 2. An information system comprising a plurality of mainframe hosts and at least one storage device that is shared by the plurality of hosts and stores fixed-length format data, wherein the mainframe host is An operating system and an input / output control program that performs format conversion for converting variable-length data to fixed-length data or for converting fixed-length data to variable-length data, and that processes input and output of the operating system to the storage device And a cache memory used for input and output of the storage device, wherein when the first mainframe host executes writing using the cache high-speed write function, the cache high-speed write data The cache in the first mainframe host. And transferring the cache high-speed write data to another second mainframe host, and transferring the cache high-speed write data transferred from the first mainframe host to the other second mainframe host. An information system, comprising: receiving and storing in the cache memory in the second mainframe host; and sharing the cache high-speed write identifier with all the mainframe hosts. 3. The information system according to claim 2, wherein, in order to share the cache high-speed write identifier, the same cache high-speed write identifier managed by all the mainframe hosts is used when the information system is started up. Initializes with an initial value, and updates the cache high-speed write identifier managed by the first mainframe host when the cache high-speed write data held in the cache memory of the first mainframe host is lost And instructs all other second mainframe hosts to update the cache high-speed write identifier. The second mainframe host receives an update instruction of the cache high-speed write identifier from the first mainframe host. And manage it on a second mainframe host Information system and updates the serial cache fast write identifier. 4. The information system according to claim 3, wherein the first mainframe host manages the update instruction of the cache high-speed write identifier by the first mainframe host having lost the cache high-speed write data. Attach the updated value of the cache high-speed write identifier to all other second mainframe hosts, and send the update instruction from the first mainframe host to the second mainframe host. Received and compares the updated first value of the cache fast write identifier attached to the update instruction with a second value of the cache fast write identifier managed by a second mainframe host. The cache fast write identifier is updated with the first value only if the first value is an update of the second value. Information system which is characterized in that. 5. The information system according to claim 2, wherein at least one shared device having internal storage means is connected to all the mainframe hosts to share the cache high-speed write identifier. Holding the cache high-speed write identifier in the shared device in addition to the mainframe host, and setting the cache high-speed write identifiers of all the mainframe hosts and the shared device to the same initial value when the information system is started up. When the cache high-speed write data held in the cache memory of the first mainframe host is lost, the cache high-speed write identifier managed by the first mainframe host is updated. Write the cache high-speed write identifier to the shared device. And instructs all other second mainframe hosts to update the cache high-speed write identifier. The second mainframe host receives the update instruction of the cache high-speed write identifier from the first mainframe host. And reading the cache high-speed write identifier from the shared device, and updating the cache high-speed write identifier managed by the second mainframe host with a value read from the shared device. 6. The information system according to claim 2, wherein the cache high-speed write data is transferred from a first mainframe host to all other second mainframe hosts. An information system using a communication path implemented for use in processor communication between the mainframe hosts. 7. The information system according to claim 2, wherein all of said mainframe hosts and said storage device are used as a network for transferring data between said mainframe hosts. An information system comprising a network capable of transferring data, wherein the network is used to transfer the cache high-speed write data from a first mainframe host to all other second mainframe hosts. 8. The information system according to claim 6, wherein the communication path or the network has a broadcast function. 9. The information system according to claim 2, wherein at least one shared device having internal storage means is connected to all the mainframe hosts to share the cache high-speed write identifier. Holding the cache high-speed write identifier in the shared device; reading the cache high-speed write identifier from the shared device when referring to the cache high-speed write identifier from the mainframe host; An information system, wherein, when the cache high-speed write data held in the cache memory is lost, the first mainframe host updates the cache high-speed write identifier of the shared device.