JPH04112242A - Data base processing device and definition change processing method - Google Patents

Abstract

PURPOSE:To shorten the stop period of operation for the definition change of a data base and to easily perform the previous test of the state after the change and the stepwise definition change by obtaining formal and general frame and mechanisms which arrange relations between data base definition and an application program in the data base. CONSTITUTION:A definition change managing means 11 is provided to manage definition information of the old version, that of the new version, and relations between them, and definition information of the old version or the new version is selected by an access selecting means 17 at the time of requesting the access. Definition change 18 of a data base 12 is divided to an application phase 14, a verification phase, and a settling phase, and mechanisms operated in respective phases are formalized and generalized to settle the general frame for change definition. Thus, definition change processing procedures 10 and restriction management 11 can be taken into a data base management system 13 in the general level.

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] The present invention relates to a database processing device and definition change processing method for accessing a database according to definition information of the database.

The purpose is to shorten the period of operational outage when changing database definitions, and to facilitate preliminary testing of post-change status and step-by-step definition changes.

Definition change management means for managing definition information for each definition element, including old verge versions, new versions that have not yet been finalized, and their relationships, and access requests.

Equipped with an access selection means for selecting whether to use the old or new version of definition information for access.

A mechanism is provided within the database management system to replace the old version information with the new version when the change process is finalized. Further, the configuration is such that each definition element is changed in accordance with the definition change model, with a definition change model consisting of an application phase, a verification phase, and a finalization phase as a unit.

[Industrial Application Field] The present invention relates to a database processing device and a definition change processing method that access a database according to definition information of the database.

A database definition expresses an interpretation of the database. The data in the database is stored in a manner consistent with interpretation according to this definition. An application program that accesses a database uses database definition information to access data. in this way,
The definition of a database, the data in the database, and the application program that accesses the database are interrelated.

Therefore, when the definition of a database is changed, it also affects the data in the database and the application programs that access the database. Therefore, while the data in the database is frequently accessed and changed, the definition of the database is not changed as often. but.

It is usually unavoidable that database definitions need to be changed due to database operation requirements.

In such cases, if the database management system has a built-in mechanism that supports definition changes based on a general framework, instead of the traditional at-hoc method of leaving things up to the user, the burden on users in changing definitions can be reduced. can be done.

[Conventional technology]

The usage pattern of the database changes over the course of two hours. Changes in the storage structure may become necessary as the amount of data stored in the database increases or decreases.

Changes in the frequency or importance of database access requests may require changes to the storage structure.

Furthermore, there are cases where the real world itself, which is reflected in the logical structure of the database, has changed and this must be reflected in the database. Simple things like changing column attributes (for example, what used to be 8 digits has now become 10 digits due to an increase in the amount of data), or adding columns (for example, adding more than just phone numbers to an address book) , a fax number is required). Furthermore, it may be necessary to add or delete table integrity constraints, or to partition or merge tables.

At the time when such various definition changes become necessary, the database already has a large amount of accumulated data and users via application programs, unlike at the initial design stage. therefore.

It is necessary to confirm that changes in definitions will have as little negative impact on existing systems and their operations as possible before making any actual changes. Also, if it is necessary to reorganize the stored data due to definition changes, this process will be used.

It is required to keep the period during which the database is unavailable as short as possible.

With conventional data management systems, if you want to conduct a preliminary test of the state after a definition change, you have to create a test system of the same size as the existing system to perform sufficient testing. Furthermore, the database management system itself does not provide a mechanism that allows changes to be made in accordance with a common definition change model for each definition element.

It is generally difficult to make gradual changes.1 In normal cases,
− It was necessary to create a system after each change.

[Invention tries to solve it! II! ! As above,
In the conventional technology, the relationship between the database definition, the data in the database, and the application program that accesses the database was not organized, so the procedure for changing the data base definition was left to users such as system operators.

It was not possible to standardize the change procedure.

For example, the following are the main methods that users use to change definitions.

The first method is to completely stop the database.

This method involves rebuilding a database that reflects definition changes and then restarting operations using that database.

This may involve recreating data and changing application programs. Although this method is a safe method as it is possible to avoid an unstable defined state because the actual operation is stopped during the transient state, the long-term suspension of operation will hinder business operations.

The second method allows definition changes to be made unconditionally in some cases even during operation, immediately after the changes are made.

This is a way to make that definition valid. The person making the definition change will deal with the impact of the change on an ad hoc basis. In other words, system operators, etc. should fully understand the nature of data related to the definition being changed, and then reflect the change without completely halting operations by localizing the scope of the change. do.

In the second method, since the measures are taken on a case-by-case basis, there is a risk that errors may occur in determining the scope of the change's impact. Furthermore, when an error occurs, it may not be possible to quickly return to the state before the change.

The present invention aims to solve the above-mentioned problems, and has data heir definition,
We organized the relationship between data in a database and application programs that access the database, and created a formal and generalized framework and a mechanism to realize it, compared to the procedure for changing definitions that was previously left up to the user. By providing this service, the aim is to shorten the period of operational outage when changing database definitions, and to facilitate preliminary testing of the post-change status and step-by-step definition changes.

[Means to solve the problem]

FIG. 1 is a diagram explaining the principle of the present invention.

In FIG. 1, 10 is a database processing device equipped with a CPU and memory, 11 is a definition change management means, 12 is a database, 13 is an old version database definition, and 14 is a new version database definition 715a to 1.
5c, 16a, 16b are definition elements related to the logical structure and storage structure of the database; 17 is access selection means;
18 represents a definition change model.

The definition change management means 11 stores the database definition 13 before the change (old version) regarding the database definition information.
and the database definition 14 after the change (new barge overturn) are managed for each definition element. It also has a mechanism to check the consistency of new version definition information.

The access selection means 17 determines that in response to an access request to the database 12, the access using the database definition 13 before the change is the one after the change 17) Cheetah Base Definition 1
4. Select which access you want to use.

Access using the selected definition information is realized.

The database processing device 1o of the present invention is equipped with such a definition change management means 11, manages definition information of an old version, definition information of a new version, and the relationship between them, and furthermore has an access selection means 17 to select between the old version and the new version. A database management system that allows users to select either version of definition information at the time of an access request, and replaces the old version of information with the new version when the change processing for all necessary definition elements is finalized. It is located inside.

Furthermore, in the definition change processing method according to the present invention, as shown in FIG.
3 and the new version of the database definition 14 coexist, and an application phase in which an operation is performed to enable application of the new version of the database definition 14;
Regarding the new version of the database definition 14, which can be used five times while continuing to operate with the old version of the database definition 13, check whether it matches the intention of the change.
a verification phase in which the verified new version of the database definition 14 is verified by running a test program;
A definition change model 18 consisting of a confirmation phase to replace the database definition 13 of the old barge town is realized as a function within the database management system, and the definition is changed one step at a time for each definition element using this framework as a unit. .

[Effect] Database definition elements include 1, for example, logical structure definition elements such as schemas, tables, columns, and integrity constraints;
There are various definition elements, such as storage structure definition elements related to data organization, physical structure definition elements indicating the arrangement of data on physical media, etc., and the database definition is based on these definition elements.

It consists of information on their mutual relationship.

The requirements for changing definitions are: ∎ A framework is required to ensure that the resulting database interpretation is consistent by repeating the operation of changing definitions, and ∎ Using actual test data to confirm whether the changes are as intended It is necessary to be able to verify the information by, etc.

To this end, the definition must be changed in accordance with the mutual relationships of the definition elements, and any conflicts that arise must be able to be corrected through repeated operations in a manner consistent with the intent of the change.

For example, if a table is deleted, the storage structure definition elements that reference the table will be affected. By performing operations such as deleting or changing definitions for each definition element that is affected, consistency as a whole is maintained. Then, it must be possible to bind the access module to consistent definition elements and verify with actual definition data whether the definition changes match the intent.

According to the present invention, a framework for the operations necessary for changing these definitions is prepared within the database management system, so instead of changing the definitions on an ad hoc basis (rather than changing the definitions in a uniform plan according to the definition change model). - Under the operation, it becomes possible to perform verification etc. and create a definition change model while minimizing the impact on the actual operation of the database. ゛ [Example] Figure 2 shows the definition change management shown in Figure 1. An example of the configuration of the means 11 is shown.

The definition change management means 11 includes, for example, as shown in FIG.
Database definition information management mechanism 20, change information management mechanism 21, consistency monitoring mechanism 22, and consistency monitoring mechanism 30
It consists of

The database definition information management mechanism 20 integrates information on each of the definition elements of the database, and
It is managed along with its identifier at each level such as storage and physical level.

The change information management mechanism 21 manages information before change (old version) and information after change (new version) in each definition element of the database, including mutual relationship information. In the example shown in FIG. 2, definition element 15a is changed to definition element 16a.

Definition element 15b has been changed to definition element 16b. Manage relevant information about these changes.

Information on the new version of database definition 14 is 9
In implementing this invention, it is not necessary to have all the information, and only the parts that have changed can be used to store the old version of the database definition 13.
It is only necessary to have the part that differs from the original, that is, the difference information associated with the change. There may be no difference information, and a copy method may be used.

In two application operations, the consistency monitoring mechanism 22 identifies a scope of application and checks consistency between definition elements within the scope of application, that is, applicability.

FIG. 3 shows the flow of processing of the consistency monitoring mechanism 22 accompanying the common operation.

■ At the start of the apply operation, accept the 5 apply operations,
Identifies the definition element to which it is applied.

■ Identify the scope of application. In the example shown in FIG. 2, the scope of application is the new version of the database definition 14 that includes the definition element 16a and definition element 16b that are subject to change.

■ Check whether the definition elements to be applied are consistent within the scope of application. If there is no contradiction, proceed to 9 process ■; if there is a contradiction, proceed to process ■.

■ If there is no conflict, incorporate the applicable definition element into the applicable scope. After that, the process moves to process (2).

■ If there is a conflict, invalidate the previous apply operation.

■ Determine whether all necessary apply operations have been completed; if not, repeat the process. If it is completed, it is applicable and the application operation is completed.

Consistency requires that there be no contradiction between definition elements within a specific scope of application (generally a partial definition of a database definition). On the other hand, consistency is checked in a range that includes the entire database and application programs related to it.

FIG. 4(a) shows the positioning of the consistency monitoring mechanism 30 shown in FIG.

The consistency monitoring mechanism 30 is a database definition element that is affected by finalizing a definition change.

Integrity with other objects such as application program 32 is monitored. Inspection range 3
1 includes an application program 32. New version database definition 14. This includes data 33 to be actually accessed.

FIG. 4 (b) shows the consistency monitoring mechanism 30 associated with the ra setting operation.
This shows the processing flow.

■ To start the confirmation operation, press l+! After performing certain operations, the definition element to be confirmed is identified.

■ Check whether the definition element to be confirmed is consistent with other definition elements and the application program 32. If they are consistent, the confirm operation is completed.

■ If not consistent, perform a reconciliation operation. The consistency monitoring mechanism 30 changes the consistency, but the user may intervene when actually performing the consistency operation.

■ If the matching operation is successful, the confirming operation is considered complete; if the matching operation fails, the confirming operation is invalidated.

FIG. 5 is an explanatory diagram of the access selection means 17 shown in FIG. 1.

The access selection means 17, when accessing the definition information,
It is possible to select either the definition information before the change (old version) or the definition information after the change (new version). Note that this access selection means 17 does not enable actual operation by an application program that uses a mixture of definition information of new and old versions at the same time. This makes it possible to verify the following.

FIG. 5(b) shows the processing flow of the access selection means 17.

■ When there is an access request, it is determined whether the request is for access after the change (new version).

■ In the case of an access request after the change, select the new version of the database definition 14.

■ If you are requesting access to the previous version, select the old version of database definition 13.

(2) The selected database definition 13 or database definition 14 is then accessed to obtain necessary definition information.

Next, a specific example of definition change will be explained with reference to FIGS. 6 to 8.

As shown in Figure 6 (a), the database before change (identifier DB) is expressed by a logical structure definition table (identifier T), a storage structure definition (identifier S), and a physical structure definition (identifier P). shall be.

Taking as an example the case where column C is added to this table T, each I! Explain lateral movement.

7 new database definitions are generated by performing an apply operation using the definition change statement as input. The new version database identifier can be expressed implicitly by adding a subscript to the old version database identifier, or can be specified explicitly.

In this example, the following table change statement (column addition) "ALTERTABLE ADD COLUMN C column description" written in SQL language will be applied to a new database (identifier DBI).

The consistency monitoring mechanism 22 shown in FIG. 2 checks whether the new version of the table definition is consistent, as shown in FIG. 6(b).

In this example, 1 table definition statement "CREATE TABLE T, 9."

By applying it to the target database DBI at the same time, together with the above table change statement (column addition),
A consistent and applicable table definition is generated. Old Hargegon's definition of Dekuhes is:
It exists independently of the new version. The relationship between the new and old versions is shown in Figure 2.
0 and the change information management mechanism 21.

The validation process for the new version of the table uses actual data and programs to verify that the changes meet the intended purpose. At this time, there are the following two methods.

One is the method shown in FIG. 7(a), in which lower layers (storage definitions, physical definitions) of the old version database are borrowed.

The other method is the method shown in Figure 7 (b), which also applies to lower layers of the old version database.

This method involves applying definition changes, closing the database to the new version, and making the database consistent as a whole. In the latter method, by adding a column to table T, record length change is applied to storage structure S, and space expansion is applied to physical structure P.

At this stage, as shown in FIG. 4(a), the integrity monitoring mechanism 30 checks the integrity of the new version of the database as a whole and the integrity of the access module (application program). is inspected.

Furthermore, the access selection means 17 shown in FIG.
Allows you to select and access any of the older version databases. However, it does not enable mixed use of the new and old versions.

In this example, the following two judgments can be made as a result of the verification.

Case E 1] If it is determined that the value of the additional column C is to be treated as invalid data (null) due to the addition of column 5C, there is no need to change the storage structure definition and physical structure definition.

[Case 2: If it is determined that it is necessary to change the record length of the storage structure definition and further increase the amount of space in the physical structure definition due to the addition of column C, then change the record length of the storage structure definition and the physical structure definition. Changes will be necessary.

As a result of the verification, if it is confirmed that the purpose of the definition change is met, the finalization phase of the table definition change can be started. In the finalization phase, the completion of the finalization of the upper layer is generally based on the premise that the finalization of the lower layer is completed.

Therefore, in case 1 above, only the interpretation of the storage structure definition and physical structure definition changes.

If the definition change is not performed (copy level), [Case 2], the finalization of the table definition change is completed at the storage structure definition (
The storage structure definition (record length change) is determined on the premise that the physical structure definition (space amount increase) is determined.

Data movement or reorganization may, but need not, be coupled with the physical definition layer.

FIG. 8(a) shows the order of confirmation completion when borrowing the lower layer definition of the old version database. Further, FIG. 8(b) shows the order of confirmation completion when changing the lower layer definition of the old version database.

Next, we will explain the overall basic model of definition change operations and states.

Conventional definition change models have generally been transition models in which a state changes from a pre-change state through an intermediate state to a final post-change state. The intermediate state is a transitional state in which definitions are being changed, and during this time, data operations on the database cannot be performed and database operations must be stopped.

On the other hand, the definition change model according to the present invention, as shown in FIG. The impact on database operations is minimized.

This new definition change model 18 is divided into an application phase, a verification phase, and a finalization phase. The outline of each phase is as follows.

(a) The application phase is a phase in which the description of the new definition is accepted, definition information in the application state (hereinafter referred to as 1 metadata) is generated, and a consistent state, that is, an applicable state is reached.

(b) The verification phase is a phase in which whether the applied metadata matches the intention of the change is verified using actual data and programs.

(C) [The fixed phase is a phase in which the old version of the database is replaced with the new version of the database after it is determined in the verification phase that the applied state metadata matches the intent of the change.

This definition change model 18 includes logical level, storage level,
It is commonly used in various abstractions at the physical level. 9th
The figure shows the flow of definition changes across multiple abstractions. In addition.

The hierarchy of definition elements is not limited to the logical, storage, and physical levels, and the hierarchy of this definition change model 18 is an example.

When the definition changes have been reflected in all abstractions, the definition changes as a whole are completed.

When changing the logical structure definition, first, the new definition is applied at the logical level, and the new definition is verified while coexisting with the old definition. The start of finalization of the logical structure after verification triggers definition changes at the storage level, if necessary. Similarly, the start of finalization of storage structures induces definition changes at the physical level.

Each phase will be explained in detail below.

[Application Phase] In the application phase, one or more definition changes are specified and an application operation is performed. FIG. 10 shows an example of the application operation and generation of metadata.

After performing a basic syntax check on the definition change specified in the apply operation, a new version of the definition is generated that reflects the specification. At this time, the old version remains as is, as shown in FIG.

Data related to the old version is also not affected by the apply operation.

Once a consistent version of the definition has been constructed, the end of the second application phase can be declared. This state is called a plicable state.

If a single apply operation does not result in consistency, a combination of multiple apply operations can bring the application to a consistent state, that is, an applicable state. Although there may be an apply operation that replaces an existing apply operation, generally multiple apply operations are treated at the same level, and there is no priority based on the order of application.

The results of the apply operation can be canceled.

At this time, the new hajon generated by the corresponding application operation can be deleted.

[Verification phase] In the verification phase, a test program is actually executed against the applicable version (this includes creation of access module AM in the new version, creation of test data, etc.). to confirm that the changes were made as intended. FIG. 11 shows an example of verifying the new version.

As a result of the verification, ■It is acceptable as is.■It is necessary to continue changing the definition to match the intention. ■
Reject changes. One of the following judgments will be made. In the case of (2), the definition of the 5 new version is deleted by canceling the application operation of the corresponding definition change.

[Confirmation phase If the co-verification determines that the offer is accepted, the *decision phase will begin. FIG. 12 shows an example of confirmation processing that is completed in a single layer.

During the finalization phase, replace the new version with the old version. How to replace it is shown in the verification result above.
, ■ Varies depending on.

② (Accept as is), as shown in FIG. 12, the confirmation phase is completed in a single layer. ■If so,
This will be completed by accumulating changes to the definitions of multiple layers.

If the process is completed by accumulating changes in the definition of multiple layers, it may cause changes in definitions at lower levels.

Interpretation between definition elements may change.

(1) When a lower-level definition change is caused in the commit phase If an apply operation (for example, adding a column) requires a lower-level secondary definition change (for example, changing the record length of a storage structure definition) In this case, the procedure for changing the definition is performed according to the criteria for changing the definition across multiple abstractions as described above with reference to FIG. That is, as a prerequisite for the completion of the upper-level confirmation, the confirmation phase of the lower-level definition change must be completed.

(2) When changing the interpretation of the relationship between definition elements In response to a definition change (for example, adding a column), the relationship with the related definition element is reinterpreted and the definition of the related definition element is not changed. There is also. For example, all accesses to additional columns may be ignored.

In such a case, definitive completion is possible without the need for a secondary definition change at a lower level and its definitive completion. Of course, since the interpretation of the data has been changed, there may be cases where all previous inquiries may not be valid.

〔Effect of the invention〕

As explained above, according to the present invention, a change in the definition of a data field is divided into a generalization phase, a verification phase, and a finalization phase, and the mechanism that operates in each phase is formalized and categorized. A framework can be established. As a result, definition change processing procedures and constraint management, which were traditionally left to the user, can now be incorporated into the database management system at a simple level. Incidentally, it is also possible to implement stepwise database reorganization processing due to definition changes within this framework.

[Brief explanation of drawings]

FIG. 1 is a diagram explaining the principle of the present invention. FIG. 2 is a configuration example of the definition change management means according to the embodiment of the present invention. FIG. 3 i Process flow of the consistency monitoring mechanism according to the embodiment of the present invention. FIG. Explanatory diagram of the monitoring mechanism FIG. 5 is an explanatory diagram of the access selection means according to the embodiment of the present invention. FIG. 6 is an example of a definition change for explaining the present invention in detail. FIG. 7 is an example of a verification method for changing definitions according to an embodiment of the present invention. FIG. 8 is an explanatory diagram of the order of confirmation completion according to the embodiment of the present invention. FIG. 9 is an explanatory diagram of definition changes across multiple abstractions according to an embodiment of the present invention. FIG. 1O is an example of common operations and metadata generation according to an embodiment of the present invention. FIG. 11 is an example of verifying a new version according to an embodiment of the present invention. FIG. 12 shows an example of a confirmation process completed in a single layer according to an embodiment of the present invention. In the figure, 10 is a data cache processing device, 11 is a definition change management means 512 is a data cache, 13 and 14 are data cache definitions, 15a, 15b, 15c, 16a, 16b
17 represents a definition element, 17 represents an access selection means, and 18 represents a definition change model.

Claims (1)

[Claims] 1) In the database processing device (10) that accesses the database according to the definition information of the database, during the process of changing the definition information of the database, the definition information of the old version, the definition information of the new version before finalization, and A definition change management means (11) that manages these relationships for each definition element; and a definition change management means (11) that manages access requests to the database by using definition information of an old version or by using definition information of a new version that has not yet been finalized. an access selection means (17) for selecting access to be performed, manages the old version definition information, the new version definition information, and the relationship thereof during the change processing, and when the change processing for necessary definition elements is finalized; A database processing device characterized in that a database management system is provided with a mechanism for replacing information of an old version with a new version. 2) A definition change processing method in a database processing device (10) that accesses a database according to database definition information includes an application phase in which an operation is performed to make new definition information of the database applicable in a state where it coexists with old definition information; , a verification phase in which the new definition information that is applicable continues to operate with the old definition information, and verifies whether it matches the intent of the change by creating test data or running a test program; The framework of the definition change model (18) provided by the database management system, which consists of a finalization phase in which a database based on new definition information is replaced with a database based on old definition information, is used as a unit, and the above definition change model is updated for each definition element. A definition change processing method characterized by changing definitions in accordance with a framework.