JPH08335211A - Decentralized object environment execution system - Google Patents

Abstract

PURPOSE: To facilitate the communication from a client to each decentralized object. CONSTITUTION: A name server 20 administers the names 22 of computers where respective decentralized objects 11, 12, 13, and 14 are present and port numbers 23 for communications. Further, respective clients 32, etc., are provided with interface parts 30 for communications with the respective decentralized objects. To request one of the decentralized objects to operate data, each client 32, etc., sends the decentralized object name to the name server 20. Consequently, the name server returns information required for a communication with the decentralized object. Therefore, the respective clients 32, etc., need only to be aware of the presence position of the name server 20. Consequently, each client need not pay attention to the presence position of a decentralized object operating on a computer connected to the network 10 and the communication protocol for the decentralized object.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention realizes a distributed object environment in which a distributed object operating on a plurality of computers connected to a network communicates with a distributed object operating on the computer or on another computer. It is about the system.

[0002]

2. Description of the Related Art Generally, when a process exchanges messages on a plurality of computers connected by a network, a message communication mechanism provided by an operating system is used. TCP / IP and UDP / IP protocols are used for the message communication mechanism. R to execute the procedure of the process running on the remote computer using this protocol
There is a mechanism called PC (Remote Procedure Call). This is a procedure for a client process to send parameters to a server process on a computer and obtain a result from the server process. At this time, the client process performs a procedure for executing the above-described RPC (for example, converting the transmission data into a format that can be transferred on the network, setting a port number, etc.), and communicating with a server process on which computer. Specify whether to do.

The above has been put into practical use as a basic function of network communication of a conventional computer operating system (for example, "UNIX Network Programming: W. Richard Stevens / Translated by Yoichi Shinoda,"
See Toppan).

[0004]

However, the above-mentioned conventional techniques have the following problems. That is, when the client process executes the above RPC,
I had to always specify on which computer the server process was running. Therefore, when the client process executes the PRCs of a plurality of server processes, it is necessary to manage which computer has the server process of each RPC, which complicates the structure of the client process. It was Also,
When the client process is on a different computer from the server process, RPC is used, and when the server process is on the same computer, it is possible to communicate using a different cost-effective method than RPC. Since the protocols are different, we could not handle them in a unified manner. For example, when a server process on a remote computer was moved to a local computer, the client program had to be rewritten significantly.

[0005]

In order to solve the above-mentioned problems, the distributed object environment realization system of the present invention provides:
The following features are provided in a client-server system that controls communication between a distributed object that operates on a plurality of computers connected to a network and a client that uses the distributed object as a server.

(1) 1. Each distributed object is provided on each computer connected to the network. 2. A name server that manages the protocol for communication with each distributed object and the distributed object name in association with each other is provided. 3. Provided in the client provided on each computer,
An interface unit for transmitting the object name of the communication destination to the name server is provided.

(2) In (1), a history object is provided in each distributed object for accumulating a client name of an operation request source and a history of operation contents. (3) In (1) or (2), a recovery object is provided in each distributed object, which stores the data content immediately before the operation and restores the data content to the state immediately before the operation when a failure occurs.

[0008]

[Action]

(1) According to the distributed object environment realization system of the present invention, objects are managed by the name server, and each client is provided with an interface unit for communication with the distributed objects that the distributed object environment has. When each client requests one of the distributed objects to operate the data, the distributed object name is transmitted to the name server. As a result, the name server returns information for communication with the distributed object to the client, and the client performs communication processing based on the information. Therefore, each client need only be aware of the location of the name server. As a result, when each client communicates with a distributed object running on a computer connected to the network,
There is no need to be aware of which computer the distributed object exists on. Also, it is not necessary to be aware of the communication protocol provided by the operating system.

(2) In (1), as a result of the history object in the distributed object taking the history of communication, the debugging work of the application program in the distributed object environment becomes easy. (3) In (1) or (2), as a result of managing the data content immediately before the operation by the recovery object in the distributed object, a failure such as an operation error can be easily dealt with.

[0010]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a block diagram of an embodiment of the distributed object environment realization system of the present invention. The illustrated system includes distributed objects 11, 12, 13, and 14, a name server 20, clients 32 and 33, and the like.
The distributed objects 11, 12, 13, and 14 are arranged in the computers 1, 2, 3, and 4, respectively. Here, a distributed object is a process created using the object-oriented concept. The structure of each distributed object is
It is as shown in FIG.

As shown in FIG. 2, each distributed object includes a server 101, a distributed object body 102, a history object 103, and a recovery object 104.
Etc. Of these, the server 101 processes requests from clients. The server 101 is provided in a distributed object environment, and there are two types: a server for remote access and a server for local access. When a distributed object receives a message from a remote computer, it uses a server for remote access, and when it receives a message only from a local computer, it uses a server for local access. Both servers have the same interface, so there is no need to change the client program when switching between servers.

FIG. 3 shows the structure of the distributed object body. The distributed object body 102 is the operation 2
00 and a set of data 300. Then, according to the message received by the server 101,
One of the operations held by the distributed object body is called. When an operation is called, it performs operations such as operations on data that has the state of distributed objects. The data 300 is not operated except by the operation of the distributed object body. The processing when an operation is called is described by the application programmer. You can also access other distributed objects from within the operation process.

FIG. 4 shows a client program which advances processing while communicating with the distributed object.
It is created by the application programmer. As shown in FIG. 4, the structure of the client program includes an interface unit 41 with the name server, an interface unit 42 with the distributed object, and a program body unit 43. The interface unit 41 with the name server is provided in the distributed object environment.
This is to communicate with the name server to get information about the distributed object. This allows the information of the distributed object to be obtained from the name server.

The interface 42 with the distributed object is provided in the distributed object environment. this is,
It is intended to be used with information obtained from name servers to communicate with distributed objects. This interface is provided by the operating system, RP
The purpose is to make it easy to use a communication protocol interface such as C. When a client program communicates with multiple distributed objects, each interface is required. The program body 43 is a part for processing the result obtained by communicating with the distributed object. This is written by the application programmer.

FIG. 5 shows the structure of the history object. The history object 103 keeps a history of messages sent from the client to the distributed object. That is, the server 101 activates the operation of the history object to store the message received by the server 101 in the history object 103.

Here, the description returns to FIG. 1 again. In FIG. 1, the recovery object 104 is for returning the distributed object to its original state when a failure occurs in the execution of the method of the distributed object. By calling the operation of the recovery object from the distributed object body, the contents of the data immediately before the operation of the distributed object can be saved. by this,
Even if a failure occurs in the distributed object, the state before the operation can be restored by returning the save information stored in the recovery object to the distributed object. Therefore, it is possible to reduce the influence of the failure on the distributed object environment.

The name server 20 stores information about each distributed object 11, 12, 13, 14. As shown, there is only one name server 20 in the distributed environment. Such a name server 20
The information stored in is the distributed object name 21, computer name 22, and port number 23. The distributed object name 21 is a name unique to the distributed object.
The computer name 22 is the unique name, device number, etc. of the computer in which the distributed object exists. Port number 23
Is the port number specified when communicating with the distributed object.

Next, the operation of the above system will be described. In the above-mentioned distributed environment, the client and the distributed object communicate with each other to proceed with data processing. In order for the client 32 to communicate with the distributed objects 11, 12 and the like operating on the computers connected to the network 10, the client 32 first makes an inquiry to the name server 20. When a request comes from the client 32 to the name server 20, the name server 20 returns the information of the distributed object corresponding to the request. That is, the name server has the information of the distributed object and returns it to the client 32. The client 32 can communicate with the distributed objects 11, 12, etc. based on the information.

The distributed objects 11, 12 and the like perform corresponding processing in response to a request from the client program. That is, the distributed objects 11 and 12 are
As described above, it is composed of a set of data and a set of operations, and operations on data are performed only by operations. As a result, when the distributed object receives a message, it starts the corresponding operation. In the distributed objects 11, 12, etc., the message from the client 32 is processed by the server 101 as follows. First, the message from the client 32 is stored in the history object 103 in order to obtain the history of the message from the client 32. When storing a message, the operation of the history object 103 is called. When debugging, by using this history object 103,
You can simplify debugging client programs that use this distributed object.

Next, the operation of the distributed object body 102 corresponding to the message from the client 32 is called. This activates one of the operations of the distributed object. When the operation is activated, the distributed object body performs the following processing. First, before processing the operation of the distributed object body 102, the state information of the distributed object body 102 for restoring the state of the distributed object when the distributed object fails is stored in the recovery object 104. At this time, when the state of the distributed object body 102 is stored, the operation of the recovery object 104 is called. Then, the operation of the distributed object body 102 is executed. And after performing the operation,
Returns the answer of the execution result to the client.

In this way, when each client 32 or 33 communicates with the distributed objects 11, 12 etc. operating on the computers 1, 2 etc. connected to the network 10, the distributed objects 11, 12 etc. It is not necessary to be aware of which computer etc. exist in, and it is possible not to be aware of the communication protocol provided by the operating system. Further, since the history of communication is obtained by the history object 103 in the distributed object, it is easy to debug the application program in the distributed object environment. Furthermore, since the recovery object 104 in the distributed object manages the data content immediately before the operation, it is possible to easily deal with a failure such as an operation error.

The present invention is not limited to the above-mentioned embodiments, and it goes without saying that various modifications can be made. For example, in the above-described embodiment, the case where both the history object 103 and the recovery object 104 are provided has been described, but the present invention is not limited to this.
The case where either or both of these are not applied is also applied. Further, in the above-described embodiment, the case where the name server 20 simply provides the communication information has been described.
You may make it substitute the communication processing to 1, 12, etc.

[0023]

As described above, according to the distributed object environment realization system of the present invention, an object is managed by a name server, and an interface for communicating with a distributed object in the distributed object environment is provided. Since this is done, there are the following effects. That is, when a client communicates with a distributed object running on a computer connected to a network, it is not necessary to be aware of which computer the object exists in. Therefore, the structure of the client process can be simplified. In addition, when the client performs the above-mentioned communication, it is possible not to be aware of the communication protocol provided by the operating system. Therefore, the communication can be handled uniformly by each client. This gives, for example,
Even if the server process on the remote computer is moved to the local computer, the client program need not be rewritten.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of a distributed object environment realization system of the present invention.

FIG. 2 is an explanatory diagram of a configuration of a distributed object.

FIG. 3 is an explanatory diagram of a distributed object body.

FIG. 4 is an explanatory diagram of a client.

FIG. 5 is an explanatory diagram of a history object.

[Explanation of symbols]

 1, 2, 3, 4 Computers 11, 12, 13, 14 Distributed objects 20 Name server 30 Interface section

Claims (3)

[Claims] 1. In a client-server system for controlling communication between a distributed object operating on a plurality of computers connected to a network and a client having the distributed object as a server, each client connected to the network is connected to each computer. A name server that manages each distributed object that is placed, a protocol for communicating with each distributed object, and each distributed object name in association with each other, and a name server that is provided in the client provided on each computer, A distributed object environment realization system comprising an interface unit for transmitting a communication destination object name to a name server. 2. Provided in each of the distributed objects,
The distributed object environment realization system according to claim 1, further comprising a history object for accumulating a history of a client name of an operation requesting source and an operation content. 3. Provided in each of the distributed objects,
The distributed object environment realization system according to claim 1 or 2, further comprising a recovery object for storing data contents immediately before an operation and returning the data contents to a state immediately before the operation when a failure occurs.