JPH02252050A - Data concurrency control method - Google Patents

Abstract

PURPOSE:To reduce the volume of programs and to reduce labor also when a control function is added by providing a master node with a data changing means. CONSTITUTION:The master node 1 controls the data of all control functions and distributedly arranges the data of the partial control functions in slave nodes 2 to 4 connected through a bus and respective slave nodes 2 to 4 execute control processing corresponding to the arranged control function data. The master node 1 is provided with a control table 10 dividing the data of the control functions in each record and indicating which node is to receive the divided record, a processing means 11 for executing IPL in each record and a data change processing means 12 for changing data in each record. At the time of IPL, each record is transmitted to the node shown in the table 10 by the processing means 11 and processed, and at the time of changing the data, the record to be changed by the means 12 is transmitted to the node specified by the table 10 and processed.

Description

[Detailed Description of the Invention] [Summary] The present invention relates to a data concurrency control method for a distributed processing system, and provides a data concurrency control method that requires a small amount of programs and requires less effort even when a control function is added. For the purpose of this, the parent node manages the data of all control functions, and the data of some control functions is distributed to multiple child nodes connected by a bus, and the child nodes manage the data of all control functions. In a distributed processing system that performs control processing corresponding to data, data of a control function is divided into record units in the parent node, and the divided records are distributed to which node.The present invention relates to an improvement of a data concurrency control method of the system.

It has a processing means at the time of IPL that performs (referred to as . send and process,
When changing data, the data change processing means is configured to send the record to be changed to the node indicated in the delivery management table for processing.

[Industrial application field]

The present invention uses a parent node used in building management systems, etc., to manage data for all control functions, and to transfer data for some control functions (schedule start/stop, power outage processing, etc.) to nodes connected via a bus. Distributed processing that is distributed to a plurality of child nodes, and the child nodes perform control processing corresponding to the arranged control function data [Prior Art] FIG. 6 is a block diagram of a conventional distributed processing system.

In the case of FIG. 6, it is a diagram showing a distributed processing system in which the A control function is distributed to child nodes 2 and 4, and the B control function is distributed to child nodes 2 and 3. 2 has control function data 20 and B control function data 30.

Since the data 20 of this A control function is different from the data 30 of the 8111111 function, the data change program 21 of the A control function, the data change program 31 of the B@control function, and the A control function data IPL time processing program 2, respectively.
2 and B control function data IPL processing program 32
have.

Therefore, when starting up the system, the data 2 of the A control function is stored in the disk memory of the parent node 1'' and child nodes 2 and 4.
0 is loaded by the A control function data IPL time processing program 22, and the B control function data 30@B control function data IPL time processing program 32 is loaded in the disk memory of the parent node 1 and child nodes 2 and 3. and load it.

When it becomes necessary to change the control function data, for example, the A control function data 20, the operator displays the A control function data 20 of the parent node 1'' on the man-machine interface 50, specifies the correction location, and makes the change. When instructed, the data change program 21 of the A control function changes the data of the parent node 1° and also sends the changed data to the child nodes 2 and 4 to make them change.

Of course, when it becomes necessary to change the B control function data 30, the B control function data change program 31 is used to change the data.
Make the changes in the same way as above.

In this way, always parent node 1" and child nodes 2 to 4
Data simultaneity control is performed so that the data of the control functions possessed by each are the same.

[Problem to be solved by the invention]

However, with conventional data concurrency control methods,
Data change program and IP for each control function data
There is a problem that there is a large amount of programs because it is necessary to have an L time processing program, and when adding control functions, a data change program and IPL are required for the added control function data.
There is a problem that a time processing program must be provided, which takes a lot of time and effort.

An object of the present invention is to provide a data concurrency control method that requires a small amount of programs (and can also reduce effort when adding control functions).

[Means to solve the problem]

FIG. 1 is an ivy diagram representing the principle of the present invention.

As shown in Fig. 1, a parent node 1 manages data for all control functions, and transfers data for some control functions to multiple child nodes 2, 3, 4, etc. connected via a bus. In a distributed processing system that is distributed and where child nodes perform control processing corresponding to the allocated control function data, the parent node 1 has a system that divides the control function data into records and assigns the divided records to which node. Delivery management table 10 indicating whether to deliver and IPL for each record
It has a processing means 11 for performing IPL and a data change processing means 12 for changing data in units of records.

At the time of IPL, the processing means 11 at the time of IPL transmits record units to the nodes shown in the delivery management table 10 for processing, and when data is changed, the data change processing means 12 transmits the records to be changed. is sent to the node shown in the delivery management table 10 for processing.

[For production]

According to the present invention, the parent node 1 is provided with a placement management table 10 that divides control function data into record units and indicates in which node the divided records are placed.

Further, an IPL time processing means 11 performs IPL on the nodes shown in the delivery management table 10 in record units, and a data change processing means 12 performs data changes on the nodes shown in the delivery management table 10 in record units. This is provided.

At the time of IPL, the IPL time processing means 11 transmits each record to the node shown in the delivery management table 10 to perform IPL, and when data is changed, the data change processing means 12 selects the record to be changed. Delivery management table 1
The data is transmitted to the node indicated by 0, data is changed, and data concurrency is controlled.

That is, since IPL and data changes are made to the nodes shown in the delivery management table 10 on a record-by-record basis, IPL
Only one type of time processing means 11 and one type of data change processing means 12 are required each, and the number of programs can be reduced.

Also, even if a control function is added, if the added control function data is divided into records and a delivery management table is added that indicates which node each record should be delivered to, IPL and data changes can be made using the existing IPL. Since the processing can be performed by the time processing means 11 and the data change processing means 12, there is no need to provide special IPL and data change processing means, so that the effort required when a control function is added can be reduced.

〔Example〕

FIG. 2 is a block diagram of a distributed processing system according to an embodiment of the present invention, FIG. 3 is a diagram showing a delivery management table and scheduled destination information according to an embodiment of the present invention, and FIG. 4 is a diagram showing data according to an embodiment of the present invention. FIG. 5 is a flowchart of a processing program during IPL according to an embodiment of the present invention.

In FIG. 2, as in the conventional example shown in FIG.
FIG. 3 is a diagram illustrating a distributed processing system in which distributed processing is performed in No. 3;

In the case of FIG. 2, in the parent node 1, the data 20 of the A control function and the data 30 of the B control function are recorded in record units as shown in FIG. 2(A).

b, c, ..., record 1. m, n, ...
The divided records a, b, c, ...
For each record Q, m, n, ..., Fig. 3 (A) [Delivery to the node indicated by 1 in Fig. 3 (A)]
A delivery management table 10 is provided as shown in FIG.

Further, the parent node l is provided with one type of data change processing program 12 as shown in FIG. 4, and one type of IPL processing program 11 as shown in FIG.

When starting up the system, the IPL processing program 11
IPL will be performed in this step, and this will be explained with reference to FIG.

In step 1, the arrangement information as shown in FIG. 3(A) is searched from the delivery management table 10 for the record to be IPLed].

In step 2, for each record, scheduled delivery destination information is created by tracing the delivery information in (A), as shown in FIG. 3(B), and is delivered to the node to which the delivery is to be made.

In the case of FIG. 3(B), parent node 1. Deliver to disk memory of child node 2.4.

In step 3, it is checked whether the transmission has been made, and for the node that has sent, the corresponding node part of the scheduled delivery destination information is cleared in step 4. In step 3, if the node cannot transmit due to a failure etc. In step 5, the corresponding node portion of the scheduled delivery destination information is left as is without being cleared.

Then, it is transmitted when the node failure is recovered.

Next, data changes will be explained according to FIG. 4.

In the case of changing data, the operator changes the data of the parent node 1 using the man-machine interface 50 in FIG. Instruct.

Then, in step 2, the delivery management table 10 is searched for delivery information as shown in FIG. 3(A) for the specified record.

In step 3, the identified record is
As shown in b), scheduled delivery destination information is created by tracing the delivery information in (A) and is delivered to the node to which the delivery is to be made.

In the case of FIG. 3(B), parent node 1. It is delivered to the child node 2.4 to change the specified record.

In step 4, it is checked whether the transmission has been made, and for the node that has sent, the corresponding node part of the scheduled delivery destination information is cleared in step 5. In step 4, if the node cannot transmit due to a failure etc. , in step 6, the corresponding node location in the scheduled delivery destination information is left without being cleared.

Then, when the node failure is recovered, the specified record is changed by sending it.

In this way, data concurrency control is performed.

That is, in this case, since IPL and data changes are performed on a record-by-record basis, only one type of processing program for IPL and one type of processing program for data change are required, reducing the number of types.

Also, even if a control function is added, if the added control function data is divided into records and a delivery management tail is added that indicates which node to send it to, IPL and data changes can be made using the existing control function data. Since the processing can be performed using the processing program 11 during IPL and the program 12 when changing data, there is no need to provide a special program, and the effort required when adding control functions can be reduced.

〔Effect of the invention〕

As explained in detail above, according to the present invention, IPL and data changes can be processed with one type of program each, so the number of types of programs can be reduced, and when control functions are added, the added If you divide the control function data into records and add a delivery management table that indicates whether to send them to the nodes, IPL and data changes can be made easily.
This has the effect of eliminating the need for existing IPL processing programs and data modification programs.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the principle of the present invention; FIG. 2 is a block diagram of a distributed processing system according to an embodiment of the present invention; FIG. 3 is a diagram showing a delivery management table and scheduled destination information according to an embodiment of the present invention; FIG. 4 is a flowchart of a processing program when changing data according to an embodiment of the present invention, FIG. 5 is a flowchart of a processing program during IPL according to an embodiment of this invention, and FIG. 6 is a block diagram of a conventional distributed processing system. It is. In the figure, 1 and 1° are parent nodes, 2 to 4 are child nodes, 10 is a delivery management table, and 11 is a processing means at the time of initial program loading. Processing program at IPL, 12 is processing means at data change, processing program at data change, 20 is A control function data, 21 is A control function data change program, 22 is A control function data IPL processing program , 30 is data of the B control function, 31 is a data change program for the B control function, 32 is a B control function data IPL processing program, and 50 is a man-machine interface.

Claims (1)

[Claims] The parent node (1) manages data of all control functions,
Part of the control function data is distributed to multiple child nodes (2, 3, 4, etc.) connected by a bus, and the child nodes perform control processing corresponding to the placed control function data. In a distributed processing system, the parent node (1) has a delivery management table (10) that divides control function data into record units and indicates to which node the divided records are to be delivered, and an initial program for each record. A processing means (11) for loading an initial program and a data change processing means (12) for changing data in units of records are provided. The data is transmitted to the node shown in the delivery management table (10) in record units for processing, and when data is changed, the data change processing means (10)
2), change the record to the corresponding delivery management table (1).
A method for controlling data concurrency, characterized in that data is sent to and processed by the node shown in item 0).