JPH05120111A - Access control method for horizontal distributed data base system - Google Patents

Abstract

PURPOSE:To shorten the time required for generation and regeneration of an access module(AM), and to improve the rate of operation by holding the AM in common between application programs(APs). CONSTITUTION:Plural sites 100, 200 and 300 provided with the same data base management systems(DBMSs) 103, 203 and 303 of a compile system are connected with a communication circuit 170. Also, with respect to all APs 101, 201 and 301 operated on each site 100, 200 and 300, the same AP is used in a source code level, and AMs 104, 204 and 304 for controlling in common an access to DBs 105, 205 and 305 of each site of each AP are generated, transferred through the communication circuit 170 and stored in all the sites. In such a state, at the time of executing an arbitrary AP, the site for executing the AP executes an access to the DB through the common AM of the site having the DB of an access destination of the AP. Accordingly, the number of AMs of the whole distributed DB system can be curtailed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distributed database that does not have a shared storage and shares a database between sites connected by a communication line, and in particular, the schema of the database at each site is the same. The present invention relates to an access control method for a horizontal distributed database system, which is suitable for load distribution and realization of high reliability.

[0002]

2. Description of the Related Art In general, a database (hereinafter, referred to as DB) is a query language for SQL (Structured Query Lan).
It is divided into an interpreter method and a compile method depending on the execution method of the guage). In the interpreter system, S
Translate and execute the QL string at runtime. This method is used when it cannot be compiled in advance (for example, when the SQL character string is given from the terminal) or when it is not necessary enough to compile (for example, COM.
This is effective for SQL statements whose syntax such as MIT is simple. In addition, the interpreter method has a simpler processing system structure than the compiling method, and therefore real-time processing is not required for the DBMS (Database Management).
nt System, database management system).

FIG. 4 is a block diagram showing the structure of a DB system using an interpreter type DBMS. The solid arrows in the figure represent the flow of data, and the broken arrows represent the flow of access control. The SQL statement 401 written in the AP (application program, application program) 400 is a syntax analysis 403 by the DBMS 402. Consistency check 404 with schema, access path determination 4
05, optimization 406 is performed, and the data 408 in the DB 407 is accessed.

On the other hand, in the compiling method, as will be described with reference to FIG. 5, the SQL character string described in the AP is compiled in advance and the access module (hereinafter referred to as AM
Stored in the DB for the DBMS system. Then, when the DBMS executes the AP,
To AM ID (IDentifier),
By receiving the runtime parameter, loading the corresponding AM from the DB in the system onto the memory and transferring control to the AM, the DB is accessed and the execution result is returned to the AP.

FIG. 5 is a block diagram showing the structure of a DB system using a compiler type DBMS. Figure 5
FIG. 5A shows a configuration particularly related to the creation of AM, and FIG.
(B) shows an outline at the time of access execution, in which a solid line arrow shows a data flow, and a broken line arrow shows an access control flow. In FIG. 5A, AP
The SQL statement 501 written in 500 is parsed by the DBMS 502 when the AP 500 is compiled.
3. Consistency check 504 with schema, access path determination 505, optimization 506, and AM creation 50
7 and system DB5 as AM508-511
At 12, AM registration 513 is performed. Then, at the time of executing the AP 500, as shown in FIG. 5B, the DBMS 502, based on the AM number 514 embedded in the AP 500,
AM of the corresponding AM 508 from the system DB 512
Read / execute 515 is performed to access data 517 in the DB 516. As described above, in the case of the compile method, the SQL statement is parsed, the consistency with the schema is checked, the access path to the DB is determined, and various other optimizations are performed at the time of AM generation. Compared with the interpreter method that executes the execution of SQL, it has a feature that the execution of SQL is faster.

However, in the compiling method, since the access path is embedded in AM, when the DB configuration is changed, for example, when the index is deleted or added, the access path used up to that point cannot be used. Occasionally, and sometimes not optimally. In order to deal with such a problem, the DBMS has a function of automatically recreating an AM when a schema change affecting the access path is made after the AM is created. This is A
M regeneration or simply regeneration.

The DB is referred to as a distributed DB system, for example, as described on pages 2558 to 2562 of "Electronic Information and Communication Handbook" edited by Institute of Electronics, Information and Communication Engineers (1988, published by Ohmsha). , Without shared memory,
There is a system that shares a DB between sites connected by a communication line. The site is an information processing system that executes a DBMS. Further, this distributed DB system is roughly classified into a horizontal distributed type and a vertical distributed type. Horizontally distributed type has the same DB schema at each site,
It is used for load balancing and high reliability. For example, D
When B is divided for each prefecture, it becomes a horizontal distributed DB system. In the vertically distributed type, the DB schema at each site is different, and there are often hierarchical relationships between DBs.
Used for DB integration, etc. For example, DB in a company
Is a vertically distributed DB system.

When the compilation method is used in the horizontal distributed DB system, the processing for searching the DB is the same as the case of the non-distributed DB system shown in FIG. It will be registered as AM on the site where there is. At this time, a function for switching the inquiry destination to each site that is not necessary in the non-distributed DB system, editing the inquiry result, and processing at the time of failure, that is, an AM control module (hereinafter referred to as CM), AP
It is required on the site side with. In this way, non-distributed DB
Whereas in the system, only AM is used, in the case of the distributed DB system, CM and AM are used at the time of access. The CM and AM are collectively called an access routine.

FIG. 6 is a block diagram showing a basic pattern of arrangement of access routines in the horizontal distributed DB system. In this figure, the horizontal distributed DB system includes a site (x) 600, a site (y) 610, and a site (z).
620 are connected via a communication line 670,
The site (x) 600, the site (y) 610, and the site (z) 620 are respectively DBMS 602, 612,
622, system DBs 606, 616, 626, and DBs 605, 615, 625. Further, the site (x) 600 has an AP 601 to which the terminals 607 to 609 are connected and which accesses the DBs 605, 615 and 625. In the horizontal distributed DB system having such a configuration, the AM is created as follows. First, to register AP601,
Site (x) that received a request to create an access routine
600 is the main site, and sites other than the main site (y)
610 and site (z) 620 are subordinate sites. Then, at the main site (x) 600, CM 603 for its own site
Is created and stored in the system DB 606. In addition, in the main site (x) 600, AM604,61 of all sites
Create 4, 624. AM6 created in this way
AM61 of sub site among 04, 614, 624
4, 624 for each secondary site (y) 610,
(Z) 620, and the main site AM 604 is stored in the system DB 606 of the main site (x) 600. The slave sites (y) 610 and (z) 620 store the sent AMs 614 and 624 in the system DBs 616 and 626, respectively.

Next, the operation at the time of regeneration of the AM is as follows. Site (x) 60 where the schema was changed
0 is the main site, and sites other than the main site, that is, the site (y) 610 and the site (z) 620 are the sub sites. Main site (x) 600 is AM for own site
604 and CM603 are played back, and system DB60
6 and sends the schema change notification to all the subordinate sites, and the subsites (y) 610 and (z) 620 have their own AMs 614 and 624, respectively.
Request playback of. Based on this request, the secondary site (y)
The site 610 and the site (z) 620 store the regenerated AMs 614 and 624, which do not change the schema for their own sites, in the system DBs 616 and 626, respectively.

The flow of execution of the AM thus created is as follows. The search request from AP601 is D
CM of the site with AP601 via BMS602
It is transmitted to 603. Then, the CM 603 determines the site to be used and issues a search request to the AM of the determined site. The AM that has received the request accesses the DB of its own site and returns the result to the CM 603. CM6
03 edits the received result and returns it to AP 601.

The horizontal distributed DB system shown in FIG. 6 represents the case where there is only one AP in the system. In reality, however, such APs simultaneously operate at all sites. The structure is as shown in FIG. FIG. 7 is a block diagram showing an actual pattern of arrangement of access routines in the horizontal distributed DB system. In this figure, all sites (x) 700, (y) in the horizontal distributed DB system
710 and (z) 720, terminals 707-70, respectively.
9, 717 to 719, 727 to 729 are connected, and each site (x) 70 is connected by the communication line 770.
AP (1) 70 at 0, (y) 710, (z) 720
1, (2) 711, and (3) 721 operate at the same time. The flow at the time of creating an AM is that the processing of one AP 601 shown in FIG. ,
This is performed for (2) 711 and (3) 721. That is, the flow of AM creation is as follows. First, for AP (1) 701, the site (x)
700 becomes the main site, and the site (y) 710 and the site (z) 720 other than the main site (x) 700 become the slave sites. Then, the main site (x) 700 creates CMs 703 for its own site and stores them in the system DB 706. On the main site (x) 700, AM for all sites
(1) Create 704, 714, and 724. Of the AM (1) 704, 714, 724 created in this way, the AM (1) 714, 724 for the slave site is transferred to each slave site, and the AM (1) 70 for the main site is transferred.
4 is stored in the system DB 706 of the main site (x) 700. In addition, the secondary sites (y) 710, (z) 720
Stores AM (1) 714 and 724 in the system DBs 716 and 726, respectively. Similarly,
The remaining site (y) 710 and site (z) 720, which require registration of AP (2) 711 and (3) 721, respectively become the main sites and repeat the above-mentioned operation, and system DB 706 of site (x) 700. CM (1) 7
03, AM (1) 704, (2) 732, (3) 74
2 in the system DB 716 of the site (y) 710,
CM (2) 730 and AM (1) 714, (2) 73
1, (3) 743, and CM (3) 740 and AM (1) in the system DB 726 of the site (z) 720.
724, (2) 733, and (3) 741 are stored. When the AM is regenerated, the AP described in FIG.
The processing is the same as in the case of a book.

The flow of execution of the AM thus created is as follows. First, a search request from the AP (1) 701 is sent via the DBMS 702 to the AP (1) 701.
It is transmitted to CM (1) 703 of a certain site (x) 700. This CM (1) 703 determines the site to be used and issues a search request to the AM of the corresponding site. The AM that received the request is the DB (DB 705, 715,
725) and access the result to the original CM.
(1) Return to 703. The CM (1) 703 edits the returned result and returns it to the AP (1) 701. Also,
The execution of AP (2) 711, which runs simultaneously with AP (1) 701, is as follows. The search request from the AP (2) 711 is transmitted to the CM (2) 730 of the site (y) 710 having the AP (2) 711 via the DBMS 712. The CM (2) 730 determines the sites to be used and issues a search request to the AM of each site. The AM that has received the request accesses the DB of its own site and returns the result to the CM (2) 730. CM (2) 73
0 edits it and returns it to AP (2) 711. AP
The same applies to (3) 721.

As described above, in the conventional horizontal distributed DB system, the relationship between AMs of other APs has not been considered. R is for building a distributed DB system.
DA (Remote Database Acces)
In order to use standard technologies such as s, remote DB access), many DBMS have adopted an interpreter method, and a DBMS adopting a compile method.
However, it is difficult to share a general AM because the AM changes depending on the index state and the slight difference in the SQL statement in the AP (for example, only the search columns are different, This is because the AM is different). Therefore, for example, as shown in FIG. 7, when considering a distributed DB system in which all sites have the same compiled DBMS, when the AP is replaced or the DB configuration is changed by modifying or adding functions. In addition, it is necessary to create and regenerate the AM, and in the conventional method of arranging the AM as shown in FIG. 7, it takes a long time to create and regenerate the AM and the service stop time becomes very long. was there.

[0015]

The problem to be solved is that in the prior art, in the distributed DB system,
It takes a lot of time to create and regenerate M, and it is not possible to shorten the service stop time. The object of the present invention is to solve these problems of the prior art and to provide a horizontal distributed DB.
It is an object of the present invention to provide an access control method for a horizontal distributed DB system that enables sharing of AMs in a system, shortens the time required to create and regenerate AMs, and improves the service operation rate.

[0016]

In order to achieve the above object, an access control method for a horizontal distributed DB system according to the present invention comprises: (1) connecting a plurality of sites equipped with the same compiled DBMS to each other via a communication line. Horizontally distributed DB
All APs that are systems and operate on multiple sites
In addition, using the same AP at the source code level, precompiling the AP at any site to generate an AM that commonly controls access to each DB of multiple sites of all APs. Common AM to all APs is transferred via a communication line and stored in all sites, and when any AP is executed, the site that executes this AP is
It is characterized in that the DB is accessed through the common AM of the site having the DB of the access destination of P.

[0017]

In the present invention, source code executed at different sites enables sharing of the substance of AM among the same AP. As a result, the number of AMs in the entire distributed DB system can be reduced. For example, if the number of sites is n,
Assuming that one AP moves at each site, the conventional method needs to generate n × n AMs.
It becomes better to generate M, and it is possible to reduce n × (n−1) pieces. As a result, it is possible to shorten the AM generation time and the regeneration time, further reduce the AM storage area in the system DB, and also reduce the memory consumption when making the AM resident in the memory. Become.

[0018]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a flow chart showing an embodiment of an access control method of a horizontal distributed DB system according to the present invention, and FIG. 2 shows a first embodiment of a structure of the horizontal distributed DB system used for the implementation according to the present invention. It is a block diagram. In FIG. 2, the horizontal distributed DB system according to the present embodiment has a site (x) 100, a site (y) 200, and a site (z), each of which has the same compiled DBMS.
300 are connected via a communication line 170. And site (x) 100, site (y) 20
0, the site (z) 300, and DBs 105, 205, 305 of the same structure, and these DB10, respectively.
5, 205, 305, the same DBMS 102, 202, 302 of the same compiling method for controlling access to the
Further, AMs 104, 204 and 304 that control access to the own DB based on requests from other sites, and CM that requests access to other sites.
(1) 103, (2) 203, and (3) 303 are provided with system DBs 106, 206, and 306, respectively. Also, site (x) 100, site (y)
200 and the site (z) 300 have terminals 107
, 111, 207 to 211, 307 to 311 are connected, and further, the same AP (1) 101 at the source code level for accessing the DBs 105, 205 and 305,
It has (2) 201 and (3) 301. Then, at the sites (x) 100, (y) 200, (z) 300,
AP (1) 101, (2) 201, (3) 3 respectively
01 operates simultaneously.

Thus, the present horizontal distributed DB system is
All sites (x) 100, (y) 200, (z) 30
0 is the same DBMS 102, 20 of the compiling method
2 and 302 are the same type of distributed system, and each DB 105, 205, 305 has the same structure,
The part of the schema that affects AM is the same at all sites, and the same AP operates at the source code level at all sites. With this configuration, the horizontal distributed DB system can realize load distribution and high reliability, realize the sharing of AM according to the present invention, and stop the system by generating and regenerating AM. The time can be shortened. Note that sharing the AM means, for example, AP (1) 101 at a different site,
It means that (2) 201 and (3) 301 use the same AM 104 on the site (x) 100 to access the DB 105. Also, each site (x) 10
0, (y) 200, (z) 300 address information difference, CM (1) 103, (2) 203, respectively,
(3) Can be absorbed by 303.

The flow of creating and regenerating AMs 104, 204, 304 by the site (x) 100 based on AP (1) 101 will be described below. First, the site (x) 100 that has received a request to create an AM or the site (x) 100 that has changed the schema becomes the main site, and is a site other than the main site, that is, the site (y).
200 and (z) 300 are secondary sites. Then, at the main site (x) 100, CM (1) 103 for all sites,
(2) 203 and (3) 303 are created, and the AM 104 for the own site is created. Furthermore, AM for your site
The AMs 204 and 304 that are copies of 104 and the CMs (2) 203 and (3) 303 corresponding to the slave sites are transferred to the slave sites (y) 200 and (z) 300, respectively. If there is a schema change, a schema change request is also transferred at the same time. Secondary sites (y) 200, (z) 30
0 indicates the transferred AMs 204 and 304, respectively.
And CM (2) 203, (3) 303 and system D
It is stored in B206 and 306. If there is a request to change the schema, the schema of the own site is changed.

Next, the AM10 thus generated
The flow of execution of 4, 204 and 304 will be described. For example, A
The search request from the P (1) 101 is sent via the DBMS 102 to the C of the site (x) 100 where the AP (1) 101 is located.
Reached to M (1) 103. This CM (1) 103
Determines the site to be used, and connects the determined site, for example, the AM 204 of the site (y) 200 to the communication line 17
A search request is issued via 0. AM20 that received the request
3 accesses the DB 205 of its own site and sends the result via the communication line 170 to the original CM (1) 10
Return to 3. Then, the CM (1) 103 edits the received result and returns it to the AP (1) 101.

Further, for example, the execution of the AP 201 running at the same time as the AP (1) 101 is as follows. AP2
The search request from 01 is sent to the AP via the DBMS 202.
It is transmitted to the CM (2) 203 of the site (y) 200 where 201 is located. The CM (2) 203 determines the site to be used and, for example, the AM of the site (z) 300.
A search request is issued to 304 via the communication line 170. The AM 304 that received the request is the own site (z) 300
The DB 305 is accessed and the result is returned to the original CM (2) 203 via the communication line 170.
Then, the CM (2) 203 edits the result and
Return to P201. The same applies to AP301.

Such an access control method for a horizontal distributed DB system is effective for a horizontal distributed DB system having one site in each prefecture and 47 sites nationwide. In other words, each site is connected to the terminal in each prefecture,
Assuming that it is a server for providing services,
At each site, a DB is sent according to the message received from the terminal.
A server AP that issues SQL to the MS operates. In such a case, this server AP can be the same at all sites, and a common AM can be used.

Next, the processing operation of the horizontal distributed DB system of FIG. 2 according to the present invention will be described with reference to the flowchart of FIG. First, generation and regeneration of AM will be described (step 1). The DBMS of the main site compiles the same AP in advance at the source code level (step 2), switches the inquiry destination to each site, edits the inquiry result, and CMs for all sites that perform fault handling. Then (step 3), AM for the own site is created (step 4). Copy of AM for own site and C for subsidiary site
M and M are transferred to the respective sites (step 5).
At this time, if there is a schema change (step 6), a schema change request is also transferred at the same time (step 7). The secondary site stores the transferred AM and CM in its own system DB.
It is stored inside (step 8). At this time, if there is a schema change request (step 9), the schema of the own site is changed (step 10). In this way, the AMs stored in each site can be commonly used if the APs to be executed are the same at the source code level.

Next, the flow of execution of the AM thus generated and regenerated will be described (step 1).
1). The main site DBMS that executes the same AP at the source code level as the one used to generate the AM is the AP
Sends the search request from the site to the CM of its own site (step 12). Based on this search request, the CM determines the site to be used (step 13), and determines the AM of the determined site.
Then, a search request is issued (step 14). Upon receipt of the search request, the AM accesses the DB of its own site (step 15) and returns the result to the CM of the main site (step 16). The CM of the main site edits the returned result (step 17) and returns it to the running AP (step 18). In this way, in the horizontal distributed DB system, AM can be made common to each AP by using the same AP at the source code level.

Even if the main site access routine cannot be decomposed into CM and AM due to optimization or the like (the access routine in this case will be referred to as CMAM hereinafter), As shown in 3, an AM sharing method is possible. FIG. 3 is a block diagram showing a second embodiment of the configuration according to the present invention of the horizontal distributed DB system used for implementing the access control method of the horizontal distributed DB system in FIG. The horizontal distributed DB system according to the present embodiment has terminals 107 to 111 and 207 to 21, respectively.
1, site (x) 120 connected to 307-311,
(Y) 220 and (z) 320 are connected by a communication line 170, and each site (x) 120,
(Y) 220 and (z) 320 are D of the same structure, respectively.
B105, 205, 305 and these DB105, 2
The same DBMS 130, 230, 330 of the compile method for controlling access to 05, 305, and AM 150, 250, 350 for controlling access to the own DB based on requests from other sites, and
Based on the request from the AP between the own sites, it controls access to its own DB and requests access to other sites C
MAM (1) 160, (2) 260, (3) 360,
System DBs 140, 240, 340 storing respectively
It is equipped with. Further, each of the site (x) 120, the site (y) 220, and the site (z) 330 is a DB.
APs (1) 101, (2) 201, and (3) that have the same source code level to access 105, 205, and 305
It has 301. And site (x) 120,
For (y) 220 and (z) 330, AP (1) 101,
(2) 201 and (3) 301 operate simultaneously.

In this way, the horizontal dispersion D of the second embodiment is
The B system, like the first embodiment described with reference to FIG. 2, has all the sites (x) 120, (y) 220,
(Z) 330 is the same as the compiling method DBMS 13
It is a homogeneous distributed system with 0, 230, 330,
A system in which the DBs 105, 205, and 305 have the same structure, and the part that affects AM in the schema is the same at all sites, and the same AP operates at the source code level at all sites. Is. As a result, the present horizontal distributed DB system enables load distribution and high reliability, and also realizes AM sharing according to the present invention, thereby reducing system downtime due to AM generation and regeneration. It can be shortened.

In the horizontal distributed DB system of the present embodiment, the access routine of the main site is
Cannot be separated into CM and AM. Therefore, the AM is shared by using an integrated access routine.
In this case, in the configuration of FIG. 2, for example, CMAM (1) 1
In the horizontal distributed DB system according to the present embodiment, the CMAM (1) 160 directly accesses the DB 105 without using the AM 150, while the 60 uses the AM 150 to access the DB 105. Become. In this way, in the horizontal distributed DB system of this embodiment, the AM can be shared between the slave sites.

As described above with reference to FIGS. 1 to 3, in the horizontal distributed DB system of this embodiment, AM is shared between applications by using the same AP at the source code level. As a result, the dispersion D
The number of AMs in the entire B system can be reduced. Then, it is possible to shorten the AM generation time and the regeneration time,
It is possible to shorten the service stop time of the system due to generation and regeneration of AM. Furthermore, A in the system DB
It is possible to reduce the M storage area and the memory consumption when making the AM resident in the memory. In addition, the present invention is shown in FIG.
The present invention is not limited to the embodiment described with reference to FIG. 3. For example, the number of sites may be increased. In short, all sites are distributed DB systems that have the same DBMS of the compilation method, and It is sufficient that the part that affects the access module is the same for all sites and that the same AP operates at the source code level for all sites.

[0030]

According to the present invention, AM is shared between APs, the number of AMs stored in each site is reduced, the time required for AM generation and regeneration is shortened, and a horizontal distributed DB system is realized. It is possible to improve the service utilization rate.

[0031]

[Brief description of drawings]

FIG. 1 is a flowchart showing an embodiment of an access control method for a horizontal distributed DB system according to the present invention.

FIG. 2 is a block diagram showing a first embodiment of a configuration according to the present invention of a horizontal distributed DB system used for implementing the access control method of the horizontal distributed DB system in FIG.

3 is a block diagram showing a second embodiment of the configuration according to the present invention of the horizontal distributed DB system used for implementing the access control method of the horizontal distributed DB system in FIG.

FIG. 4 is a block diagram showing a configuration of a DB system using an interpreter-type DBMS.

FIG. 5 is a block diagram showing a configuration of a DB system using a compiler type DBMS.

FIG. 6 is a block diagram showing a basic pattern of arrangement of access routines in a horizontal distributed DB system.

FIG. 7 is a block diagram showing an actual pattern of arrangement of access routines in a horizontal distributed DB system.

[Explanation of symbols]

 100 site (x) 101 AP (1) 102 DBMS 103 CM (1) 104 AM 105 DB 106 system DB 107-111 terminal 120 site (x) 130 DBMS 140 system DB 150 AM 160 CMAM (1) 170 communication line 200 site (Y) 201 AP (2) 202 DBMS 203 CM (2) 204 AM 205 DB 206 System DB 207-211 Terminal 220 Site (y) 230 DBMS 240 System DB 250 AM 260 CMM (2) 300 Site (z) 301 AP (3) 302 DBMS 303 CM (3) 304 AM 305 DB 306 System DB 307-311 Terminal 320 Site (z) 330 DBMS 340 System DB 350 AM 360 CMAM (3) 400 AP 401 SQL statement 402 DBMS 403 Parsing 404 Consistency check 405 Access path determination 406 Optimization 407 DB 408 Data 500 AP 501 SQL statement 502 DBMS 503 Parsing 504 Consistency check 505 Access path determination 506 Optimization 507 AM creation ~ 511 AM 512 system DB 513 AM registration 514 AM number 515 AM reading / execution 516 DB 517 data 600 sites (x) 601 AP 602 DBMS 603 CM 604 AM 605 DB 606 system DB 607 to 609 terminals 610 sites (y) 612 DB MS 614 AM 615 DB 616 System DB 620 Site (z) 622 DBMS 624 AM 625 DB 626 System DB 67 0 communication line 700 site (x) 701 AP (1) 702 DBMS 703 CM (1) 704 AM (1) 705 DB 706 system DB 707 to 709 terminal 710 site (y) 711 AP (2) 712 DBMS 714 AM (1) ) 715 DB 716 System DB 717 to 719 Terminal 720 Site (z) 721 AP (3) 724 AM (1) 725 DB 726 System DB 727 to 729 Terminal 730 CM (2) 731 to 733 AM (2) 740 CM (3) ) 741-743 AM (3) 770 communication line

Claims (1)

[Claims] 1. A horizontal distributed database system in which a plurality of sites having the same compiled database management system are connected by a communication line, and a source is provided for all application programs operating on the plurality of sites. Use the same application program at the code level, compile the application program in advance at any site, and generate an access module that commonly controls the access of all the application programs to each database at the multiple sites. Then, the access module common to all the application programs is transferred via the communication line and stored in all the sites, and the application program is executed when any of the application programs is executed. An access control method for a horizontal distributed database system, wherein a site executing a program accesses the database via the common access module of a site having a database to which the application program is accessed.