JPH08272632A - Exclusive control method for shared resources - Google Patents

Abstract

(57) [Summary] [Purpose] To reduce the processing time for multiple jobs that use shared resources without compromising the reliability of operations. [Constitution] In a disk subsystem, channel synchronization control is performed for a synchronous command job 900, a staging job 901, an asynchronous destage monitoring job 902, and an asynchronous destage job 903, which perform one job using a cache slot 905 in a cache memory. Job group 906, asynchronous destage job group 90
Different job group identifiers are given to each of the seven job groups, and the job group identifier match check is performed for the reservation requests 908, 909, 910, and 911 for the cache slot 905 from each job, and sharing is performed only when they match. By permitting, the cache slot 905 is used simultaneously between a plurality of jobs in the same job group, and exclusive between different job groups.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a shared resource exclusive control technique, and more particularly, to a disk subsystem including a disk controller which employs a multiprocessor system for performing parallel processing by a plurality of processors for the purpose of function / load distribution. It is related to effective technology when applied to etc.

[0002]

2. Description of the Related Art In the prior art, in a computer system that uses the same resource for a plurality of jobs,
For example, in order to prevent a logical inconsistency from occurring due to simultaneous use of shared resources such as a CPU, a channel, a main storage device, and a cache memory, exclusive control for exclusively using shared resources is performed. According to this technique, access to the resource by another job is suppressed until the execution of the processing such as the data update request for the shared resource by the specific job is completed.

As a similar conventional technique, for example,
In the technique described in Japanese Patent Laid-Open No. 1-303537, exclusive control is performed by associating the presence / absence of a shared resource with the presence / absence of a specific file, and forcibly executing the specific file every certain period of time. By accessing the shared resource by a specific job,
An exclusive control method for a shared resource is disclosed that exclusively prevents access to the resource by another job and avoids a deadlock.

[0004]

In the former conventional technique described above, for example, a reference request for a resource by another job during execution of a data update request for the resource for a specific job is exclusively suppressed, so that a plurality of requests for reference to the resource are not accepted. There is a problem that the processing time of the entire job becomes long on average. That is, the execution time of the job processed using the shared resource is
Since the time required for execution of each job becomes a simple sum and becomes long, there is a disadvantage that, for example, the CPU and channels, the main storage device, and the cache memory, which are the resources of the host computer, are occupied for a long time to lower the availability.

In the latter prior art, since exclusive control of resources is performed depending on the presence or absence of files, exclusive control can be easily realized by a high-level language or the like, but at some point the resources are assigned to one of a plurality of jobs. There is no change to simple exclusive control.

As described above, according to the conventional shared resource exclusive control method, when a certain job uses the shared resource, other jobs cannot use the shared resource until the job is completely completed. It was one of the factors that significantly reduced the use efficiency of shared resources.

An object of the present invention is to provide a shared resource exclusive control technique capable of shortening the processing time of a plurality of jobs using the shared resource without impairing the reliability of the operation.

Another object of the present invention is to provide a shared resource exclusive control technique capable of realizing improvement of availability of shared resources without impairing the reliability of operation.

[0009]

In the present invention, a plurality of jobs that share the same resource and process them are considered as one group, and they are defined as a job group, and among a plurality of jobs within the same job group, Operation is performed such that simultaneous use is permitted regardless of a reference request and an update request for the same resource, and shared resources are mutually exclusive between different job groups.

Further, a sharing permission condition indicating whether or not each combination of processing modes indicating a processing type for a shared resource of an individual job is set, and belonging to the same job group,
In addition, it is possible to perform an operation of permitting simultaneous use of the shared resource only to a plurality of jobs that satisfy the sharing permission condition.

For example, the control logic can be constructed by the following means as an example.

(1) First means for judging whether or not a shared resource is in use.

(2) When the shared resources are in use, it is checked whether or not the job group numbers match. If they match, the resource securing (sharing) is successful, and if they do not match, the resource securing (sharing). A second means of determining failure.

(3) One to one for each resource when resource securing is successful
A third means for allocating resources by registering allocated job numbers and job group numbers in the resource management table corresponding to, and registering reserved resource numbers in the job management table corresponding to each job on a one-to-one basis.

(4) When releasing resources, the release request job is
At the time of the last allocation job of the resource, the resource is deleted by deleting the job group number from the resource management table, usually deleting the allocation job number from the resource management table, and deleting the reserved resource number from the job management table. Fourth means to release.

[0016]

FIG. 3 is a conceptual diagram for explaining an example of the operation of the present invention, and the operation of the present invention will be described below with reference to FIG.

For example, job A300 and job B30
1, job C302, job D306, job E307
However, it is assumed that the shared resource 305 can be independently used through the procedures of the shared resource securing request 309, the shared resource securing request 310, the shared resource securing request 311, the shared resource securing request 312, and the shared resource securing request 313. And job A3
00, job B301, and job C302 perform one job as a system, and similarly job D306,
Two jobs E307 perform one job as a system. Job A300, Job B301, Job C302
The job group number A303 is assigned as the three job groups, and the job group number B308 is assigned as the two job groups D306 and E307.

The job A 300 is the resource management processing unit 304.
A shared resource securing request 309 is made to Resource management processing unit 30
In step 4, the first means determines that the shared resource 305 is not in use, and the third means allocates the shared resource 305 to the job A300 by the third means. Then, the job B301 makes a shared resource securing request 310 to the resource management processing unit 304. In the resource management processing unit 304,
It is judged that the shared resources are being secured by the first means, and the second
The job group number is checked by this means. It is determined that they are in the same job group, and the shared resource 305 is assigned to the job B301 by the third means. next,
The job C302 makes a shared resource securing request 311 to the resource management processing unit 304. Job C302 similar to the above procedure
The shared resource 305 is allocated to the third means by the third means. At this time, the shared resources 305 are the job A300 and the job B30.
1, job C302 can be used simultaneously.

On the other hand, job D3 having a different job group
When 06 issues a shared resource securing request 312 to the resource management processing unit 304, the sharing becomes impossible by the second means. Similarly, the job E307 cannot be shared. That is, the job D306 and the job E307 of the job group B308 are exclusive to the jobs of the job group A303. When the job C302 makes a request 314 for releasing the shared resource 305 to the resource management processing unit 304, the shared resource 30 is transmitted from the job C302 by the fourth means.
5 is released.

[0020]

Embodiments of the present invention will now be described in detail with reference to the drawings.

1 and 2 are flow charts showing an example of the operating principle of the shared resource exclusive control method according to one embodiment of the present invention, and FIG. 4 is applied to the shared resource exclusive control method of the present invention. 5, 6 and 8 are conceptual diagrams showing an example of the control information used in the shared resource exclusive control method of the present embodiment, and FIG. FIG. 9 is a conceptual diagram for explaining the types of processing (shared resource use mode) in the disk subsystem, and FIG. 9 is a conceptual diagram showing an example of the operation of the shared resource exclusive control method of this embodiment in the disk subsystem of FIG. is there.

First, the configuration of a cached disk subsystem in which the shared resource exclusive control method of this embodiment is implemented will be briefly described with reference to FIG.

In FIG. 4, the disk controller 402
Is connected to the host computer 400 through the signal lines 416 and 417 and the channel control device 401 on the upper side, and the drive 41 which is a magnetic storage medium on the lower side.
Connected with 5. The disk controller 402 reads and writes data on the drive 415 in response to a request from the host computer 400. Host computer 40
0 and the drive 415 are controlled by the control processor 4 built in the disk controller 402.
03, control processor 404, control processor 405,
The control processor 406. These control processors are connected to the host computer 400 and the drive 415 via the channel control device 401, the drive interface 413, and the drive interface 414.

The control memory 407 is a common memory that can be accessed by all the control processors 403, 404, 405, 406, and includes a resource management table 408 and a job management table 409 for managing resources by the disk controller 402. Common control information is stored. A specific description of this common control information will be given later. Cache memory 4
11 and a directory for storing the management information thereof are memories accessible from all the control processors 403, 404, 405, 406, and the cache memory 41.
1 is from the drive 415 to the control processors 403, 40
It is used to temporarily store the data read by 4,405,406. In addition, the cache memory 411 is
The control processors 403, 404, 405, 406 are shared resources that can be used independently of each other. The cache slot 412 is a data management unit of the cache memory 411. Control processor 403, 404, 405, 4
06 is a cache memory 41 via a signal line 410.
1. Data and control signals are exchanged with the control memory 407.

Such a plurality of control processors 403
9 to 406, a disk subsystem including a disk controller 402 and a drive 415 having a multiprocessor configuration includes a cache slot 905 (412) and control processors 403 to 40 as illustrated in FIG.
It can be regarded as an information processing system shared by a plurality of jobs (control processors 403 to 406) via a resource management unit 904 composed of control logic for controlling 6. Resource management department 9
The control logic 04 performs control operation as shown in FIGS. 1 and 2, which will be described later.

Then, via the cache memory 411 (cache slot 412), the host computer 40
In the disk control device 402 that controls the exchange of data between 0 and the drive 415, there is a process of synchronizing with a command of the host computer 400 and a process that occurs internally. As an example of the former, the access command processing from the host computer 400 includes a synchronization command job 900 and a staging job 901. For example, when performing the data transfer process, the staging job 901 drives the drive 415 to the cache slot 905 which is a data management unit of the cache memory 411.
Transfer data from. The sync command job 900 is
From the cache slot 905 to the channel controller 401
Data transfer to. As an example of the latter, an asynchronous destage job 903 that performs data transfer (hereinafter referred to as destage) of unreflected data to the drive 415 from the cache slot 905 to the drive 415, and an asynchronous destage monitoring job 902 (secure request) 910 is used).

From this point of view, the job groups are classified, and the synchronization command job 900 and the staging job 9 are classified.
01 is one channel synchronization control job group 906, and the asynchronous destage job 903 and the asynchronous destage monitoring job 902 are one asynchronous destage job group 907. Further, a different job group number is given to each. When the synchronous command job 900 processed by the control processor 403 executes the reservation request 908 for the cache slot 905 by the resource reservation processing 100 described later, the cache slot 905 is allocated. In addition, the staging job 90 having the same job group number processed by the control processor 404.
1 is a reservation request 90 for the same cache slot 905.
9 is executed, it is possible to allocate the same cache slot 905 by the resource securing process 100 and use the same cache slot 905 at the same time. On the other hand, when the asynchronous destage job 903 having a different job group number processed by the control processor 406 executes the reservation request 911 for the same cache slot 905, the resource reservation processing 100 rejects the reservation of the cache slot.

As described above, it is possible to share a cache slot 412 (905) with a job having the same job group number and to exclude a job having a different job group number.

Next, various common control information used in this embodiment to realize the above-mentioned exclusive control will be described. The common control information in this embodiment includes a job management table 409 and a resource management table 408. Hereinafter, they will be described in order.

FIG. 5 shows an example of a job management table 500 (Job Control Block; hereinafter abbreviated as JCB) used in this embodiment. JCB500 has 1 for each job
There are as many as one to one, and each of them has three setting areas in which three data are individually stored. That is, the job number setting area 5 in which the job number 501 for the resource management processing unit 904 to identify each job is stored.
01a, a reserved resource number setting area 502a in which a reserved resource number 502 reserved by the job is stored, and a job group number setting area 503a in which a job group number 503 to which the job belongs is stored. The job number 501 is stored before the job is generated. In the reserved resource number setting area 502a, when the resource reservation process 100 allocates a resource to the job, the allocated resource number 601 described later is stored. The job group number 503 is stored when the job is generated.

FIG. 6 shows an example of the resource management table 600 used in this embodiment. The resource management table 600 of this embodiment has a number corresponding to each resource in a one-to-one correspondence, and each has a setting area in which four types (six) of data are individually stored. That is, the resource management processing unit 904
, A resource number setting area 601a in which a resource number 601 for identifying each resource is stored, an allocation job number 602 to which the resource is allocated, an allocation job number 603
Assigned job number setting area 602a in which
The allocated job number setting area 603a, the job group number setting area 604a in which the job group number 604 that secures the resource is stored, the reserved resource use mode 605, and the use mode 606 (hereinafter, "fun").
6), that is, a use mode setting area 605a and a use mode setting area 606a.

The resource number 601 is the disk controller 40.
2 is stored when started. The assigned job numbers 602 and 603 and the job group number 604 are stored when the resource management processing unit 904 permits sharing of the resource. functions 605 and 606 represent resource use modes, and when the resource management processing unit 904 permits sharing of the resource, a read mode 801 to be described later,
Any of the write mode 802, staging mode 803, and destaging mode 804 is stored.

FIG. 7 shows an example of the definitions of the above-mentioned read mode 801 (read 701), write mode 802 (write 702), staging mode 803 (staging 703), and destaging mode 804 (destaging 704) in this embodiment. Indicates. Read mode 801
A (read 701) indicates data transfer from the cache slot 412 to the host computer 400, and a write mode 802 (write 702) indicates data transfer from the host computer 400 to the cache slot 412. Staging mode 803 (Staging 70
3) shows data transfer from the drive 415 to the cache slot 412, and the destaging mode 804 (destaging 704) shows data transfer from the cache slot 412 to the drive 415.

FIG. 8 shows an example of conditions (shareability condition 800) set during a plurality of usage modes of the cache slot 412 in the disk subsystem. In the case of the present embodiment, there are four types of functions, read mode 801, write mode 802, staging mode 803, and destaging mode 804, as described above, and whether or not resources (cache slot 905) can be shared among them. But entry 805, entry 8
06, entry 807, and entry 808 are set. In the case of the present embodiment, from the viewpoint of data protection, the read mode 801 is permitted to be shared with the staging mode 803 (entry 805), but is not permitted to be shared with the destaging mode 804 (entry 80).
6) is set as follows. Similarly, write mode 8
No. 02 is permitted to be shared with the staging mode 803 (entry 807), but is not permitted to be shared with the destaging mode 804 (entry 808). The sharing permission / inhibition condition 800 is a condition valid in the disk subsystem of this embodiment.

For example, due to a bug in the program,
Jobs in the same job group are read mode 801
Then, when an attempt is made to reserve the cache slot 412 in the destaging mode 804 with respect to the secured cache slot 412, the entry 806 indicating the sharing relationship between the two is set to be unshareable from the viewpoint of data protection. Sharing is not allowed even for jobs in the job group.

Next, the operation of the control processors 403, 404, 405, 406 of the disk controller 402 of this embodiment which executes the above-mentioned shared resource exclusive control method will be described.

First, according to the flowchart of FIG.
The resource securing process 100 will be described. First, it is determined whether or not the reserved resource is free, by whether or not the job group number 604 of the resource management table 600 is set (step 101). If the job group number 604 is not set, the following processing is executed. JCB5
The reserved resource number 601 of 00 is set to the reserved resource number setting area 502a (step 109). Then, the job group number 503 is stored in the job group number setting area 604a of the resource management table 600 (step 110). Next, the resource management table 60
The job number 501 of the reservation request source is stored in the assigned job number setting area 602 of 0 (step 111).
The reservation request source job has designated the func specified in the function 605 setting area of the resource management table 600.
stored in step 112. Then, the return code is set to secure success (step 113) and the process returns to the secure request source (step 114).

If the job group number 604 is already set, the following processing is performed. A match check is performed between the job group number 604 set in the resource management table 600 and the job group number 503 on the JCB 500 that is the reservation request source (step 102). If they do not match, the sharing failure is set in the return code (step 108),
The process returns to the reservation request source (step 114). If the job group numbers 503 match, the following processing is executed. Fu set in the resource management table 600
The condition match of the naction 605 is checked (step 103). If the conditions do not match, resource reservation failure is set in the return code (step 108), and the process returns to the reservation request source (step 114). If the conditions match, the reserved resource number setting area 502a of the JCB 500
The resource number 601 to be secured is stored (step 104). Then, the resource request source job number 501 is stored in the assigned job number registration area 603a of the resource management table 600 (step 105). The function requested by the reservation requesting job is stored in the usage mode setting area 606a of the resource management table 600 (step 106). Then, the securing success is set in the return code (step 107), and the process returns to the securing request source (step 114).

Next, an example of the resource release processing 200 will be described with reference to the flowchart of FIG. First, the assigned job number 60 in the resource management table 600 of resources to be released
2, 603, the release request source job determines whether the last secure job of the resource is present (step 2).
01). If it is the last job, the resource management table 600
The job group number 604 of (step 20
4). Subsequently, the job number 501 of the release request source is deleted from the assigned job number setting area 602a of the resource management table 600 (step 202). Subsequently, the reserved resource number setting area 5 of the JCB 500 of the release request source job
The resource number 601 of the resource requested for release from 02a is deleted (step 203), release success is set in the return code (step 205), and the process returns to the release request source (step 206).

If it is not the last release job, the assigned job number setting area 602a of the resource management table 600
The release request source job number 501 is deleted from (step 202). Then, the release request source job JCB500
The resource number 601 of the resource requested to be released is deleted from the reserved resource number setting area 502a (step 203), and release success is set to the return code (step 205).
The process returns to the release request source (step 206).

As described above, in this embodiment, when exclusive control of the cache memory 411 (cache slot 412) which is a shared resource of the plurality of control processors 403 to 406 is performed, the plurality of control processors 403 to 406 are controlled by the channel synchronization control job. Grouping into a group 906 and an asynchronous destage job group 907, and according to the job group number 604 assigned to each job group, sharing of the cache slot 412 is allowed even when different control processors 403 to 406 have the same job group number 604. Therefore, the control processors 403-
Compared with the case where access to the cache slot 412 of 06 is managed by simple exclusive control, for example, the synchronous command job 900 and the staging job 901 can share the cache slot 412 and process in parallel. That is, the cache slot 412
Data transfer processing from the cache slot 412 to the channel control device 401 can be executed in parallel even if the staging processing for the disk storage is not completely completed, the data transfer processing time is shortened, and the file processing throughput of the disk subsystem is improved. To do. Further, availability of the cache memory 411 and the cache slot 412 is improved.

Further, the plurality of control processors 403 ...
If 406 is divided into job groups and the plurality of control processors 403 to 406 belonging to each job group are allowed to share unconditionally, one of the jobs sharing the cache slot 412, which is a shared resource, is reset. In that case, there is a concern that a job different from the combination of the jobs before the reset will share the shared resource after the reset. In addition, there is a concern that a program bug may cause sharing with a job that should not be shared. As a result, for example, dataization occurs.

In order to prevent this, in the present embodiment, as described above, the job group number 503 and the funcitons 605 and 606 capable of sharing the shared resource are stored in the resource management table 600. Every time, along with checking the job group number 604,
Shared resource usage mode (funciton) 605, 6
06 is checked, and even if the job group numbers 604 match, jobs that do not satisfy the sharing permission conditions of funcitons 605 and 606 are not permitted to share the cache slot 412, which is a shared resource. As a result, when the cache slot 412 is shared by a plurality of jobs executed by the plurality of control processors 403 to 406 and the like, the cache is shared between jobs that should not be shared due to a program bug or processor failure. The problem that the slot 412 is shared can be reliably prevented, and the reliability of operation is improved.

As described above, according to this embodiment,
When a plurality of jobs executed by the plurality of control processors 403 to 406 and the like access the cache slot 412, a plurality of jobs that do not cause a logical inconsistency in the access to the cache slot 412 belong to the same job group. The cache slots 412 are grouped into two groups, and the jobs belonging to the same job group are allowed to use the cache slots 412 at the same time. Data transfer processing to the control device 401 can be executed in parallel, and the file processing throughput of the disk subsystem can be improved. Further, availability of the cache memory 411 and the cache slot 412 is improved.

Further, the cache slot 4 of each job
Since it is determined whether or not a logical inconsistency occurs even at the processing level (functon) for 12, a plurality of logical inconsistencies caused by a processor failure, a program bug, a reset, or the like are generated. It becomes possible to prevent the cache slot 412 from being shared by the above job, and the reliability of the disk subsystem is improved.

Although the invention made by the present inventor has been specifically described based on the embodiments, the present invention is not limited to the above embodiments, and various modifications can be made without departing from the scope of the invention. Needless to say.

For example, the system to which the shared resource exclusive control technique of the present invention is applied is not limited to the disk subsystem and the like, but can be widely applied to general information processing systems having shared resources.

[0048]

According to the shared resource exclusive control method of the present invention, it is possible to reduce the processing time of a plurality of jobs using the shared resource without deteriorating the reliability of the operation. .

Further, it is possible to obtain the effect that the availability of the shared resource can be improved without impairing the reliability of the operation.

[Brief description of drawings]

FIG. 1 is a flowchart showing an example of the operating principle of a shared resource exclusive control method according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an example of the operation principle of the shared resource exclusive control method according to the embodiment of the present invention.

FIG. 3 is a conceptual diagram illustrating an example of an operation of a shared resource exclusive control method that is an embodiment of the present invention.

FIG. 4 is a block diagram showing an example of a configuration of a disk subsystem to which the shared resource exclusive control method of the present invention is applied.

FIG. 5 is a conceptual diagram showing an example of control information used in the shared resource exclusive control method of the present invention.

FIG. 6 is a conceptual diagram showing an example of control information used in the shared resource exclusive control method of the present invention.

FIG. 7 is a conceptual diagram showing an example of definition of a usage mode in a disk subsystem to which the shared resource exclusive control method of the present invention is applied.

FIG. 8 is a conceptual diagram showing an example of a sharing permission condition of a use mode in a disk subsystem to which the shared resource exclusive control method of the present invention is applied.

FIG. 9 is a conceptual diagram showing an example of an operation of a disk subsystem to which the shared resource exclusive control method of the present invention is applied.

[Explanation of symbols]

402 ... Disk control device (storage control device), 403,
404, 405, 406 ... Control processor, 407 ... Control memory, 411 ... Cache memory (shared resource), 4
13, 414 ... Drive interface, 400 ... Host computer, 410 ... Signal line between control processor and cache memory and control memory, 415 ... Drive (rotary storage device), 416, 417 ... Channel control device and control processor signal Line, 500 ... Job management table, 600 ... Resource management table.

Claims (3)

[Claims] 1. A group of a plurality of jobs to be processed by using the same shared resource is defined as a job group, and an identifier corresponding to each job group is assigned to each job group. Managing the jobs belonging to each of the job groups, permitting simultaneous use of the shared resource between the plurality of jobs having the same identifier, and allowing the shared resource to be shared between the plurality of jobs having the same identifier. An exclusive control method for a shared resource, which is characterized by performing exclusive control such as exclusive access. 2. A sharing permission condition indicating whether or not each combination of processing modes indicating a processing type for the shared resource of each of the plurality of jobs belonging to the same job group is set, and the same job group is set. The shared resource exclusive control method according to claim 1, wherein the simultaneous use of the shared resource is permitted only to a plurality of jobs that belong to the job and satisfy the sharing permission condition. 3. The shared resource is a rotary storage device that permanently stores data, a cache memory that holds a copy of the data in the rotary storage device, and the rotary storage device via the cache memory. A cache slot set in the cache memory in a disk subsystem including a storage controller including a plurality of processors for controlling the transfer of the data between a device and a host device, and the job is the cache Read to read data from the slot to the host device, write to write data from the host device to the cache slot, staging to read data from the rotary storage device to the cache slot, data from the cache slot to the rotary storage device Consists of destaging The job group is set such that the read and write requests from the host device for the same cache slot and the staging and destaging requests from the rotary storage device can be executed in parallel as much as possible. The shared resource exclusive control method according to claim 1 or 2.