JPH01142966A - Distributed processing system - Google Patents

Abstract

PURPOSE:To selectively activate the application processing means of respective nodes and to attain the sharing of a purpose work by defining the execution instruction contents in accordance with the resources distribution between the nodes to a control table for demarcating a work provided at respective nodes. CONSTITUTION:In nodes A and B, control parts Cp successively refers to and interprets a control table CT and successively, activates application processing parts Ipi having an execution designation. Application processing parts Ip1 communicates the information through a communicating memory part PCM in a self-node. When the designation memory contents on a memory part PCM are adjusted so as to be the same ones by respective memory synchronizing updating mechanism (u) of the nodes A and B, the information can be communicated at the section of application processing parts Ip2 of the node B through a communicating memory part PCM. For this reason, the program to realize the application processing parts Ip2 can be prepared without considering by which of the nodes A and B an action is executed.

Description

[Detailed Description of the Invention] [Objective of the Invention] (Field of Industrial Application) The present invention provides a distributed processing system suitable for performing a series of tasks among nodes connected to an online network. Regarding.

(Prior Art) Conventionally, when distributed computer nodes (distributed nodes) connected via an online network work together to share tasks, an application program is placed on each node, and the opposing application program is A communication agreement was established between the two companies, and a series of processing procedures for conducting online communication in accordance with this agreement were incorporated into each program. For example, the a program of A node is A
It is used to access resources (files, etc.) on the node B, and the b program on the B node is used to receive the access results from the a program and perform output processing to the resources (printer, etc.) on the B node. In order to divide the functions, a series of processes for passing information from program a to program b are realized in program a.
For example, a series of processes for receiving information from program a are implemented in program b.

In the above function sharing (distribution) method, for example, in the above example, some of the resources of the A node (some files of the file group to be accessed) are moved to the B node for the purpose of load adjustment, etc. When it becomes necessary to relocate/ill-check the resources to be accessed within the network,
It became necessary to review and change the program's processing function itself. Also, a pair of nodes with different resource allocations, for example, file X.

A pair of node A with y and node B with printer (first
For the pair (second node pair) of node C with file It was necessary.

(Problems to be Solved by the Invention) As mentioned above, in the past, in order to have multiple nodes work together to share and execute a series of tasks, the application program for each node was created in consideration of the resources each node had. Therefore, if the resource distribution relationship between each node changes, or furthermore, if the resource distribution relationship differs for each network even if the business content is the same, significant program changes are required. There was a problem.

Therefore, in this invention, nodes connected to an online network work together to share and execute a series of tasks, but if the tasks are the same, the resource distribution relationship between each node in the network changes. An object of the present invention is to provide a distributed processing method that can standardize application programs placed on each node even in networks with different resource distribution relationships.

[Structure of the Invention] (Means for Solving the Problems) This invention provides for each distributed node to share and carry out the target work,
several applied processing means common to each node for executing the target work separately for several small tasks; a control table in which execution specification information for specifying an execution node for each applied processing means is set; Execution control means refers to this control table and determines whether or not the application processing means to be activated next is to be operated on its own node, and a designated memory of communication memory means provided for information exchange between each application processing means. The present invention is characterized in that a memory synchronization update means is provided for matching the contents of the areas.

(Function) According to the above configuration, even if all the applied processing means necessary to carry out the target work are commonly provided in each node, the execution instruction contents of the control table are used to separate (divide) the work. As long as it is defined according to the resource distribution relationship between nodes, only the applied processing means defined in the control table will be selectively activated in each node according to the business execution status, and each node will be linked. The purpose of the work can be divided and carried out.

Therefore, if the business content is the same, all the application processing means necessary for the execution of the business will not be affected by the resource distribution relationship between each node, and therefore the functions of these application processing means and the above-mentioned execution control means, etc. It becomes easy to create an application program that realizes this.

(Embodiment) FIG. 1 shows a block configuration of a distributed network system to which the present invention is applied. In the same figure, A and B are nodes (distributed nodes, computer nodes) that are arranged in a distributed manner and perform a series of tasks in conjunction with each other, and L
is a transmission line interconnecting nodes A and B. Node A
, B are as follows. In nodes A and B, Tpl, Ip2...lpn are prepared for each small task that constitutes the target task, and CT is an application processing component for executing the corresponding small task, and CT provides execution instructions for application processing components 1p1 to Ion. Control table to be defined, Cp is control table C
This control component has a control table interpretation section Cpi that interprets the specified contents of T, and an application component execution control section CpC that performs activation control of application processing components Ip1 to II)n according to the interpretation result of the control table interpretation section Cpi.

Control parts CP, control table CT and applied processing parts I
p1 to II)n are functional elements realized by application programs installed in nodes A and B.

PCM is control part Cp, applied processing part Ipl~11)n
SI is a synchronization establishment unit for establishing synchronization between the own node and the other node, and U is a memory synchronization update mechanism. This memory synchronization update mechanism U detects changes in the content of the designated memory area (area indicated by diagonal lines) on the communication memory unit PCM of the own node and other nodes in order to match the memory information of the same area, and the other node (the designated memory area of the communication memory unit PCM). Ll is each synchronization establishment unit 3m of nodes A and B
L2 is a logical bus (logical path) between each memory synchronization update mechanism 0 of nodes A and B.
It is.

FIG. 2 shows an example of the control table CT. As shown in the figure, the content of each entry in the control table CT includes an identifier (applied processing component name) of an applied processing component necessary for performing a series of tasks and execution designation information. Here, as execution specification information, P (Primary), S
(S secondary) and B (Both)
There are three types available. P is applicable applied processing (small work)
S means that the primary node executes the corresponding application process, S means that the secondary node executes the corresponding application process, and B means that both the primary and secondary nodes execute the corresponding application process in parallel. Each is shown below. Note that the control table CT of this embodiment also includes information necessary to manage the activation order of the application processing components 1p1 to Ipn necessary for performing a series of tasks. It is also possible to manage it by another means and set only the pair of the application processing component name and execution designation information in the control table CT.

Next, the operation of one embodiment of the present invention will be explained.

First, in the nodes A and B, the control component Cp sequentially refers to and interprets the control table CT, and sequentially activates application processing components 1pi (1≦i≦n) that are specified to be executed. The activated application processing component pi communicates information via the communication memory unit PCM within its own node. Now, here, the application programs for realizing the respective functional elements of the control component Cp1 control table CT and the application processing components Ipl to Ipn are equally placed in the nodes A and B connected online by the transmission path. Therefore, if the memory synchronization update mechanism U of nodes A and B adjusts the contents of the designated memory areas on each communication memory unit PCM of nodes A and B to have the same contents,
For example, the application processing component 1pl of node A communicates with, for example, application processing component Ip2 of node A, the communication memory unit PC within its own node.
In exactly the same way as communication can be performed via PCM, information can be exchanged with the application processing component 1p2 of node B via the communication memory unit PCM. For this reason, applied processing parts I
In order to execute the corresponding applied processing (small task) using p2, it is only necessary to arbitrarily decide whether it should be executed on node A or node B, taking into account the resource situation, etc.
The program portion that implements the application processing component 1p2 can be created without considering whether it will be operated on node A or node B. This means that the application processing component Ip1. Ip2...Ipn are common to all. Furthermore, control parts Cp and control table C
The same is true for T, and the same control component Cp and control table CT can be placed in nodes A and B, respectively, and no changes are necessary.

Now, as shown in the control table CT of FIG. 2, it is defined whether each applied processing component Ip1 to Ion should operate on a primary node or a secondary node. Therefore, at the start of processing, node A,
It is necessary to determine which of B will be the primary node and which will be the secondary node. This determination method can be defined in advance as part of the system-generated information for nodes A and B, or it can be determined based on instructions from the operator during execution and a sequence is executed to obtain consensus between nodes A and B. method etc. can be applied. In any case, the above primary -
/Secondary partition is preset at nodes A and B.

If this setting is made, the following operations will be possible.

First, control components Cp prepared for nodes A and B, respectively.
The control table interpreter CI)I in the control table CT
When it is necessary to interpret and start the applied processing component Ipl,
Calls the application component execution control unit Cpc. The application component execution control unit Cpc uses (preset) primary/secondary classification information of its own node and the control table interpretation unit Cp.
The execution designation information of the application processing component Ipi requested (to be activated) by the call of i is compared, and if they match, it is determined that the local node should be executed, and if they do not match, it is determined that the other node should be executed. Also,
The execution specification information of the applied processing component It)+ is B (Both
), the application component execution control unit Cpc determines parallel execution of the local node and the partner node. When the application component execution control unit CpC determines execution on its own node or parallel execution, it starts the requested application processing component Ipi. On the other hand, if the execution of the other node is determined,
The application component execution control unit Cpc is the control table interpretation unit Cpi.
, it appears as if the application processing component 1pi has been activated, and the control table interpreter C is executed without actually controlling the activation.
Return control to pi.

This is called "empty control." This blank control allows the control table interpreter cl to move on to interpreting the application processing component to be activated next.

The application component execution control unit Cpc makes the above determination every time it is called by the control table interpretation unit Cp1.

At this time, when the application component execution control unit CpC detects a change from continuous execution of the own node to execution of the other node,
Synchronization establishment unit SI within own node! Notify l of the “remote activation request”. The synchronization establishment unit SL1 sends this request to the synchronization establishment unit S11 of the partner node via the logical bus 1. Contrary to the above case, when the application component execution control unit Cpc detects a change from continuous execution of the other node to execution of the own node, the application component execution control unit Cpc sends a WAIT (request for start of the other party) to the synchronization establishment unit S11 in the own node. weight). In this case, the synchronization establishment unit Sl in the own node has a “partner startup request” from the synchronization establishment unit Sm in the other node, and upon receiving this request, it sends a request to the application component execution control unit C in the own node.
Notify PC. The application component execution control unit Cpc remains in a holding state until the synchronization establishment unit S11 within its own node receives a “partner activation request” from the other node, and is forced to wait for activation of the application processing component requested by the control table interpretation unit Cp1. . Then, when the synchronization establishment unit SIl in its own node receives the “partner activation request” from the other node and this fact is notified to the application component execution control unit CpC, the control unit Cpc receives the request from the control table interpretation unit Cpi. Start the applied processing component. Note that at this time, in the other node, the synchronization establishment unit S
11, the control is returned to the application component execution control unit Cpc, and the "empty control" of the execution of the partner node is repeated.

Now, when changing from execution on the other node to execution on the own node and activating the application processing component that is the target of execution on the own node, what is important is that the specified memory on the communication memory part PCM that has changed as a result of execution on the other node. The contents of the area are changed to the communication memory part PCM of the own node by the operation of the memory synchronous update mechanism U.
is reflected in the corresponding memory area of the node, and the contents of the specified memory area of both nodes match. Therefore, when the memory synchronization update mechanism U detects that writing to a specified memory area on the communication memory unit PCM in the node in which the mechanism U is located occurs and memory information changes, the memory synchronization update mechanism U updates the communication memory unit PCM of the other node. In order to reflect the change in the corresponding specified memory area above, the write information is transmitted together with the position information to the memory synchronization update mechanism U of the other node via the logical bus L2. The memory synchronization update mechanism U that has received this transmission information uses this transmission information to write the write information to the corresponding position in the designated memory area on the communication memory unit PCM in its own node. Through this operation, the contents of the designated memory areas of each communication memory unit PCM in both nodes are made to match.

Now, when the application component execution control unit Cpc notifies the synchronization establishment unit SIl in its own node of the “other party activation request”, it sets a synchronization flag at a predetermined position in the designated memory area of the communication memory unit PCM. There is. It is now assumed that the synchronization flag setting described above has been performed by the synchronization establishment unit SI of node A. If the memory synchronization update mechanism U of nodes A and B processes the contents of the specified memory areas on each communication memory unit PCM of nodes A and B to match, then the specified memory area of the communication memory unit PCM of node A The synchronization flag set at the location should be set at the corresponding location in the specified memory area on the communication memory unit PCM of the node 8. Therefore, when the synchronization establishment unit S11 in the node B, which is the counterpart node of the node A, receives the “activation request” from the synchronization establishment unit Sl of the node A, the synchronization establishment unit S11 in the node B, which is the counterpart node of the node Check the sync flag,
Based on the result of this check, it is determined whether or not the content matching of the designated memory area on the communication memory unit PCM has been completed (by the memory synchronization update mechanism U). After confirming that the contents of the designated memory area match, the synchronization establishment unit Sm of the node B returns control to the application component execution control unit CpC in the node B.

With the above processing functions, for example, node A becomes the primary
・Defined and defined as a node, node B is secondary ・
Control table CT is defined as a node.
In the applied processing parts ip1゜Ip2. When Ip5 is designated as P and applied processing component Ip3゜Ip3 is designated as S, as shown in Fig. 3, applied processing components Ipi and Ip2゜lp5 operate in node A, and in node B, The application processing component ID3°Ip4 operates and a series of tasks are performed. If a resource (for example, a file) to be processed by the applied processing component 1p4 is moved from node B to node A, the execution specification information corresponding to the component name indicating the applied processing component Ip4 in the control table CT is By rewriting S to P, as shown in FIG. 4, at node A, application processing components Ipt, Ip2. Ip4.
It can be easily adjusted so that the ID5 operates and the applied processing component 1p3 operates on the node i. Also, for example, node A
.

On both nodes of B, there is a file X and an application processing component IDX that performs the process of outputting the contents of the communication memory unit PCM to this file In this case, the applied processing component 1px may be designated as B (BOth) in the control table CT. By doing this, as shown in FIG.
Output processing to file X is performed in parallel on both nodes of B. In other words, processing that does not change the contents of the communication memory section PGM, such as reflecting the contents of the communication memory section PCM in an external file, can be performed in parallel at nodes A and B, so that performance is significantly improved.

Now, when the application component execution control unit CDC detects a change from execution on the own node to parallel execution on both nodes due to the above 8 specifications, or a change from parallel execution on both nodes to execution on the other node, In the synchronization establishment section S-
The application component execution control unit C
When the PC detects a change from execution on the other node to parallel execution on both nodes, or a change from parallel execution on both nodes to execution on its own node, the PC sends a "request to start other party" to the synchronization establishment unit 3m in its own node. "WAIT" to wait for the execution of the other node to finish, thereby preventing the contents of the communication memory unit PCM from being changed during execution of the other node. In addition,
When it is detected that the parallel execution on both nodes has changed to the parallel execution on both nodes, the execution can be continued as is, and there is no need to notify the other node of a "partner activation request" or to issue a WAIT instruction.

In the above embodiment, P (primary), S (secondary), etc. are used as the execution instruction information, but it is also possible to use a node name or the like. Further, in the above embodiment, a case has been described in which two nodes perform one task, but the present invention can also be applied to a case in which three or more nodes perform one task.

[Effects of the Invention] As detailed above, according to the present invention, the following effects can be achieved.

■ When attempting to carry out a task in conjunction with nodes connected to an online network, an application program is created assuming that each node has all the resources necessary to carry out the task; In order to separate (allocate) tasks, the desired task can be accomplished by simply defining the execution instruction contents of the control table according to the actual resource distribution state.
Program creation becomes significantly easier.

■ From (■) above, even if the resource distribution relationship between nodes changes, it can be easily handled by simply rewriting the control table.

In addition, when performing the same task in any of several distributed network systems with different resource distribution relationships, the same application can be executed simply by rewriting the control table.
The program can be applied.

■ Since it is easy to make adjustments according to the resource distribution relationship, the optimum division of inter-node interlocking performance can be found by verifying the point while changing the resource allocation.

[Brief explanation of the drawing]

FIG. 1 is a block configuration diagram of a distributed network system to which this invention is applied, FIG. 2 is a diagram showing an example of the table structure of the control table shown in FIG. 1, and FIGS. 3 to 5 are similar to FIG. 1. 2 is a diagram illustrating an example of division of work between nodes A and B shown in FIG. A, 8... Node (distributed node), Cp... Control component, CT... Control table, Ip1 to Ipn... Application processing component, PCM... Communication memory unit, Sll...
Synchronization establishment unit, U: memory synchronization update mechanism. Applicant's agent Patent attorney Takehiko Suzue

Claims (1)

[Claims] In a distributed network system comprising a plurality of nodes interconnected through an online network, each of the nodes is prepared for each of several small tasks constituting a target task, and executes corresponding small tasks. An applied processing means common to each of the nodes mentioned above to perform the above operations, a control table in which execution specification information for specifying the execution node for each applied processing means is set, and a control table that refers to this control table to perform the next activation. Execution control means for determining whether or not to operate the application processing means on its own node and activating the application processing means in accordance with the result of this determination; and a communication memory provided for information exchange between the respective application processing means. means, and memory synchronization update means for detecting changes in memory information in a specified memory area of the communication memory means, transmitting the same information to other nodes, and matching the contents of the specified memory areas of the communication memory means of the own node and the other node. and synchronization establishing means for synchronizing the operation of the applied processing means between the own node and the other nodes, so that each of the nodes interlocks and performs the objective work in a shared manner. Distributed processing method.