JPH01250151A - Library file access system - Google Patents

Abstract

PURPOSE:To shorten the input/output time together with reduction of a reading area and to localize the influence of an input/output fault by forming a hierarchical structure of members of a library file and controlling these members for each submember. CONSTITUTION:A register means 3 for submembers is used together with an access means 4 and a lock means 5. The means 3 registers each information (submembers) formed into a hierarchical structure with the members sorted for each function into an optional member of a library file 7. The means 4 reads and writes the information on submembers. The means 5 inhibits the writing actions of other tasks while the information on the submember is being read and furthermore inhibits the reading/writing actions of other tasks during the updating of information on the submembers. In such a way, the input/output time is shortened together with reduction of a work area and the restoration processing is facilitated.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method of accessing a library file on an external storage medium such as a magnetic disk by a computer system.

[Conventional technology]

Conventionally, this type of library file stored multiple subfiles (hereinafter referred to as members) in one file, and was managed on a member-by-member basis (many times. Therefore, information could only be accessed on a member-by-member basis. Ta.

[Problem to be solved by the invention]

In this manner, the conventional library file access method allows only member-by-member access.

Therefore, even if the access target is not the entire member but a specific logical record in the member, all the logical records of the target member are read, the specific logical record is updated, and then all the logical records are read again. had to write.

Therefore, if the number of logical records for one member is large at this time, input/output takes time, and furthermore, a work area is required to store all the logical records of the member. Additionally, if an input/output failure occurs in the target library file while accessing a member, all logical records in the member will be affected, so recovery processing will be required for the entire member, reducing information reliability. There are drawbacks such as.

[Means to solve the problem]

The library file access method according to the present invention includes submember registration means, access means, and locking means.

In a computer system that writes data to a library file created on an external storage medium, the registration means stores each information layered in arbitrary members of the library file by classifying the members into functional units (hereinafter referred to as submembers). This is for registering.

The access means is for reading and writing information on submembers registered in the library file by the registration means.

The lock means is to prevent other tasks from writing information while sub-member information is being read, and to prevent other tasks from reading or writing information while updating sub-member information.

〔Example〕

Next, the present invention will be explained in detail with reference to the drawings.

Referring to FIG. 1, there is a library file 7 having a plurality of members 8, an external storage medium 6 that stores this library file 7 and a lock file 9, and a submember ( (not shown) registration means 3; access means 4 for accessing sub-member units registered by this registration means 3;
Locking means 5 for inhibiting writing from other tasks during access by this access means 4 or restricting read and write access operations from other tasks while updating submember information, these registration means 3, and access means 4. and software 2 equipped with a locking means 5, and a plurality of workstations 1.2...n, 1.2, which operate the plurality of software 2, respectively. including.

This embodiment shows that one library file 7 and one lock file 9 on the external storage medium 6 are being accessed by a plurality of workstations 1-n.

FIG. 2 shows member 8 of library file 7 in FIG.
An example of the logical structure of is shown. A means that it is the highest submember among the members; submember B1 and submember C exist under A, and submembers D to E exist under B. A library file access method according to the present invention is used. and member 8
The information can be classified into submember A to submember I and stored in the library file 7.

FIG. 3 shows the directory section 7 of the library file 7.
1 and the data section 72 are shown. Registration of a submember by the registration means 3 in FIG. This is achieved by securing a data record, setting a pointer to the data record in the directory, and writing variable-length logical records to the data record. By the registration means 3 as described above, data records are secured for each submember as shown in FIG.

FIG. 4 shows an example of how sub-members are stored in the directory at this time. Here, directory type ゛1
' means the top-level submember directory entry, "2" means the other submember directory entries. Each submember is
A pointer to a submember directory one level higher in each directory entry (MOTHERPO
INTER), and each directory has a pointer (DAUGHTERPOINTER) to a submember directory entry one level lower in the hierarchy. In this way, the parent-child relationship between sub-members is shown. Furthermore, if there are many submembers that have the same immediate parent, forward pointers (SISTERPOINTER) and backward pointers (ELDERPOINTER) are provided between submember directory entries.

Next, the operation of securing a directory entry and setting a pointer will be explained in detail with reference to FIG.

First, when registering the highest submember A, one entry is secured in the directory, a directory type '1' is set, and A is stored. When registering B and C under A,
One entry is secured in the directory, directory type "2" is set, and B1 and C are stored.At this time, pointers to the directory entries of B1 and C are set in DAUGH-TERPOINTER of A, and B1 and C are stored in DAUGH-TERPOINTER of A.
and MOTHERPOI of the directory entry in C
A pointer to A is set in NTER. Next, when registering work from D under B, one directory entry is secured and directory type "2" is set, and E.

F, G, and H are stored in order. If the data cannot be stored in one entry, one more entry is secured and the data is stored. At this time, B's DAUGHTERPOINT
A pointer to the directory entry from D to H is set in ER, and the 5IST of the directory entry from D to H is set.
ERPOINTER contains a pointer to the directory entry of
A pointer to the directory entry from D to H is set in ERPOINTER. In addition, in the MOTHERPOINTER of the directory entry from D to H and engineering,
A pointer of B is set for each.

In this way, one member is constructed by chaining the hierarchical structure between submembers using pointers.

Access to a submember by the access means 4 of FIG. 1 can be performed without being aware of the storage location of the logical record by specifying the submember name.

The submember names to be accessed are filled with periods (“,”) starting from the highest level submember name and ending with the target submember name.
DAU-GHTERPOINTE is specified fully qualified with
While referring to R, submember names separated by periods are followed one after another to search for the directory of the target member name.

When submember I shown in FIG. 4 is accessed, submember names A, B, and name are specified. First, the directory entry of directory type "1" is searched for submember name A. Next, the directory entry pointed to by A's DAUGHTERPOINTER is searched for submember name B.

Furthermore, the first direct entry pointed to by B's DAUGHTERPOINTER is searched for the submember name .

If no work is found in the target directory entry,
By referring to the pointer to the data record set in the directory tree pointed to by 5ISTERPOINTER, the logical record can be accessed.

At this time, since the information of each sub-member among the members is independent, it is possible to access a plurality of sub-members at the same time. In addition, since the hierarchical structure between submembers is chained using pointers, member-by-member access such as copying, moving, deleting, and locking of members is specified by the top-level submember name. This is possible when accesses are made consecutively.

The locking means 5 in FIG. 1 is for setting a lock on the target member when there is a possibility of accessing one library file 7 from a plurality of workstations 1 as shown in FIG. .

Table 1 Member access restrictions based on lock type R mark: Reference possible W mark: Update possible × mark: Lock error Table 1 shows member access restrictions based on lock type. There are two types of access locks.

Write lock is a method that sets a write prohibition attribute for each member to prevent other tasks from adding/updating other submembers in the same member when reading hierarchical information in submember units. It can only be read by the issuing task and by other tasks.

When hierarchical information is read on a sub-member basis and then modified and written, access locks are applied on a member-by-member basis to prevent similar processing from multiple tasks and discrepancies in the information. It sets an access-prohibited attribute to the task, making it impossible for other tasks to read or write to it.

Locks are normally set and released in memory, but the locking means 5 according to the present invention stores a lock table created in a lock file 9 on an external storage medium 6 so that it can be referenced from a plurality of workstations 1-n. Do it using

Access restriction by the locking means 5 involves setting a specified type of lock in the lock table for the target member 8, and specifying a lock type set within the same task for the target member 8. This is done by releasing the lock from the lock table and referring to the lock table when attempting to access a submember to check whether access restrictions have been set for the target member.

[Effects of the Invention] As explained above, the present invention hierarchically organizes the members of a library file and manages them on a sub-member basis, thereby making it possible to access each sub-member, thereby reducing input/output time and reducing the read area. This has the effect of being able to localize the effects of reduction and input/output failures.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing one embodiment of the present invention, and FIG. 2 is a diagram showing an embodiment of the present invention.
3 is a diagram showing the concept of the library file 7 of FIG. 1, and FIG. 4 is a diagram showing an example of storing submembers in a directory. l...Work stage 3, 2...Software, 3...Registration means, 4...
Access means, 5...Lock means, 6...
...external storage medium, 7...library file, 8...member of library file, 9...
...Lock file. Agent: Patent Attorney Otoo Uchihara / Zuken 3

Claims (1)

[Claims] In a library file access method that accesses multiple subfiles in one file on an external storage medium, one subfile is classified into functional units, and each piece of hierarchical information is registered as a unit. an access means for accessing the information registered by the registration means to each of the hierarchical information units; and an access restriction for preventing changes from being made by other tasks during access by the access means. A library file access method comprising means.