JPH05216850A - Hierarchical data expression converting system - Google Patents

Abstract

PURPOSE:To provide the hierarchical data expression converting system of a distributed processing system, which can execute a communication at a high speed between computers and can execute an inter-computer communication by a compact converting mechanism. CONSTITUTION:In the case a computer 1 executes a data communication to the computer 1 of other computer group 2, an internal data expression format of the computer of a transmitting origin is converted to an in-group standard data expression format 3 of a transmitting destination and transmitted to other computer group 2, and the computer 1 of the transmitting destination converts the transmitted data to an internal data expression format of the own computer. In the case of executing a data communication to the computer 1 in the same group, the computer 1 executes the data communication by converting the internal data expression format of the own computer to the in-group standard data expression format 3 or the internal data expression format of other computer 1. Also, between each computer, an in-group standard data expression format 4 for executing a communication by the computer 1 is provided, and it is also possible to execute mutually a data communication between each computer group 2.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a distributed processing system in which a plurality of computers are connected via a network,
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a hierarchical data representation conversion method for converting internal data representation formats between computers, and in particular, the present invention is a hierarchical type that can perform efficient data conversion when connecting computers having different internal representation formats. The present invention relates to a data representation conversion method.

[0002]

2. Description of the Related Art With the recent development of distributed processing systems,
A distributed processing technique is widely adopted in which various computers are connected to one large-scale network, and program-to-program communication is performed with each other in a procedure call format to distribute the load. However, in the case of such a procedure calling method,
Since the internal representation format of data differs from computer to computer, it is necessary to convert the internal representation format of data. It is desirable for the conversion of the internal data expression format that the load on the communication between the computers is small and the speed is high. However, it is necessary to convert all the internal data expression formats of a wide variety of computers. The conversion mechanism becomes large-scale.

Therefore, in the above distributed processing system, it is important to efficiently convert the internal data representation format of the computer. FIG. 7 is a diagram showing two representative examples of the data conversion method in the conventional distributed processing system. Reference numerals 101 to 105 denote computers.
Reference numeral 06 is an inter-computer standard data representation format, in which FIG. A shows the standard method and FIG.

Further, in the figure, the computers connected by a straight line have a conversion mechanism for converting the internal data representation format of the own computer into the internal data representation format of the other computer, and the computers directly communicate with each other. A computer that is a computer that can be connected and is connected by a straight line through the inter-computer standard data representation format 106 converts the internal data representation format of the computer into the inter-computer standard data representation format at the sending computer. Is transmitted to the destination computer, and the destination computer converts the inter-computer standard data representation format to the internal data representation format of the computer for data transmission.

In the standard system of FIG. 1A, each computer 10
When 1 or 105 performs data communication with other computers 101 to 105, the transmission source uniformly converts the internal data representation format of each computer into the inter-computer standard data representation format 106, and transmits the data. The receiver converts the inter-computer standard data representation format 106 into the internal data representation format of each computer.

Further, in the complete connection system of FIG.
Each of the computers 101 to 105 has a conversion mechanism for converting the internal data representation format of its own computer into the internal data representation format of all computers on the other side with which data communication is performed, and each computer 101 or 105 has another computer 101. When performing data communication with Nos. 105 to 105, each computer converts the data into the internal data representation format of the transmission destination and transmits the data.

By the way, in the standard system of FIG. 7a,
Even if the transmission source and the reception source have the same internal data representation format, since the conversion of the data representation format is always required twice, the efficiency is low and the data communication speed is generally slow. Further, in the complete connection method of FIG. 7b, since each computer 101 to 105 has a conversion mechanism for converting into the internal data representation format of all computers on the other side, the number of data conversions is small and the data communication speed is high. It's fast. However, this method has a drawback in that the conversion mechanism becomes large in scale and it is necessary to provide a conversion mechanism to all the computers each time a computer having a new internal data representation format is connected to the network, which is inflexible. there were.

[0008]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned drawbacks of the prior art.
By enabling the efficiency of communication speed and the scale of the conversion mechanism to be determined according to the target network, between computers that differ only partially in the data representation format, the minimum required data conversion is used to achieve high-speed computer-to-computer conversion. In addition to enabling communication between computers, a compact conversion mechanism enables inter-computer communication, and even if a computer with a new internal data representation format is added, it is not necessary to change the conversion mechanism. Hierarchical data representation conversion The purpose is to provide a scheme.

[0009]

FIG. 1 is a block diagram showing the principle of the present invention. In order to solve the above-mentioned problems, the invention of claim 1 of the present invention is, as shown in FIG. 1, a means for connecting a plurality of computers 1 via a network and each computer 1 calling a program of another computer 1. And a means for providing a service to another computer 1 and a data representation conversion method of a distributed processing system in which the data representation format is converted between computers 1 having different internal data representation formats to communicate data with each other. , A computer connected to the network 1
Is hierarchically divided into a plurality of computer groups 2 and each computer group 2 is provided with at least an in-group standard data representation format 3 representing the computer group 2.

When each computer 1 performs data communication with a computer 1 of another computer group 2 having a different internal data representation format, each computer group 2 communicates with each other via the intra-group standard data representation format 3. Data communication to. The invention of claim 2 of the present invention is the same as that of the computer of claim 1,
In addition, a conversion mechanism is provided between the in-group standard data expression format 3 of the own computer group 2 and all the in-group standard data expression formats 3 of the other computer groups 2.

According to a third aspect of the present invention, a plurality of computers 1 are connected via a network, each computer 1 has means for calling a program of another computer 1, and means for providing a service to another computer 1. In a data representation conversion method of a distributed processing system in which data representation forms are converted between computers 1 having different internal data representation forms and data is mutually communicated, a plurality of computers 1 connected to a network are hierarchically arranged. It is divided into computer group 2 and each computer group 2
A standard data representation format 3 within a group representing the computer group 2 is provided, and the computer 1 is provided between the computer groups and between the computer groups 2.
Is provided with an inter-group standard data representation format 4 for communication.

When each computer 1 communicates with another computer 1 of another computer group 2 having a different internal data representation format, data is mutually exchanged between the computer groups 2 via the inter-group standard data representation format 4. Communicate. According to the invention of claim 4, in the invention of claim 3, each computer 1 is provided with a conversion mechanism between the intra-group standard data representation format 3 and the inter-group standard data representation format 4 of its own computer group.

According to a fifth aspect of the present invention, in the first, second, third or fourth aspect of the present invention, each computer 1 has an internal data representation format of the computer 1 and standard data within a group of the computer group 2 to which the computer 1 belongs. A conversion mechanism between the expression format 3 and the expression format 3 is provided. The invention of claim 6 is the invention of claim 1, 2, 3 or 4, in which each computer 1 has an internal data representation format of the computer 1 and all types of data representation formats of the computer group 2 to which the computer 1 belongs. And a conversion mechanism between and.

The invention of claim 7 is the invention of claims 1, 2, 3,
In the invention of 4, 5 or 6, one computer group 2 is constituted by computers 1 having similar internal data. According to an eighth aspect of the present invention, in the first, second, third, fourth, fifth or sixth aspect of the present invention, one computer group 2 is configured by the computers 1 that need frequent data communication.

[0015]

When the computer 1 performs data communication with the computer 1 having another internal data representation format of another computer group 2,
The internal data representation format of the computer 1 of the transmission source is converted into the standard data representation format 3 of the transmission destination group and transmitted to the other computer group 2. The destination computer 1 in the other computer group 2 converts the transmitted data into the internal data representation format of its own computer and receives it.

When the computer 1 performs data communication with the computers 1 having different internal data representation formats in the same group, the computer 1 sets the internal data representation format of its own computer to the standard data representation format 3 within the group or other. It is converted into the internal data representation format of the computer 1 and transmitted. When the data is transmitted in the in-group standard data expression format 3, the recipient computer 1 converts the in-group standard data expression format 3 into the internal data expression format of its own computer 1 and receives the data.

Further, an inter-group standard data representation format 4 for the computer 1 to communicate between the computer groups and between the computer groups 2
It is also possible to perform data communication between the computer groups 2 via the inter-group standard data expression format 4. Divide the computers that make up the network hierarchically into multiple groups,
By providing the intra-group standard data representation format 3 in each computer group and performing data communication, the data conversion speed can be improved without increasing the scale of the conversion mechanism.

When the computers are classified into groups, if the similarity of the internal data representation format is emphasized and the groups are organized, the scale of the conversion mechanism becomes small and the frequency of format conversion is emphasized. Organizing groups makes the overall conversion more efficient in speed.

[0019]

FIG. 2 is a diagram showing an embodiment of the present invention. 11
1 to 13 are computer groups, 11a to 11c, 12a to 12c, 13a to 13C are computers included in each computer group, and 14 is a standard data representation format between groups, and FIG. Figure b shows the intra-group standard and inter-group standard methods, Figure c shows the intra-group complete connection and inter-group complete connection methods, and Figure c shows the intra-group complete connection and inter-group standard methods. ..

In the figure, the computer connected by a straight line has a conversion mechanism for converting the internal data representation form of the own computer into the internal data representation form of the other computer.
Computers that can directly communicate with each other and that are connected by a straight line via the standard data representation format 14 between computers are It is a computer that performs data transmission by converting to a standard data representation format and transmitting it to a destination computer, and then at the destination computer converting the inter-computer standard data representation format to the internal data representation format of the computer.

Further, the computer groups 11, 12, 13 are further provided.
In the table, computers having similar internal data representation formats of computers or computers having a high frequency of conversion of mutual data (high) are classified and collected, and computers 11c, 12c in each group, Reference numeral 13c is a computer having a group standard data representation format representing each computer group. FIG. 3 is a diagram showing a data conversion mechanism of each computer, 21 is an application of each computer, 21a is internal data of each computer, 21b is a destination address, 22 is a conversion control unit, 2
2a is an intra-group processing unit, 22b is an inter-group processing unit, 22c is a conversion control unit, 22d is a computer group management table, 23 is a communication control unit,
Reference numeral 23a is transmission data, and 23b is a destination address.

The intra-group processing unit 22a in the figure converts the internal data of the computer into intra-group standard data or destination internal data, and the inter-group processing unit 22b converts the internal data of the computer or intra-group standard data. The conversion control unit 22c refers to the computer group management table 22d to determine whether the communication destination is within the group or between the groups, and performs the intra-group processing by referring to the computer group management table 22d. It is means for controlling the conversion processing of the unit 22a and the inter-group processing unit 22b, and determining the destination address.

FIG. 4 is a diagram showing various forms of the intra-group processing unit 22a and the inter-group processing unit 22b used in each system of FIG. 2, and FIG. 4A shows the intra-group processing unit 31 used in the standard system. FIG. 7B shows the intra-group processing unit 32 used in the complete combination method, FIG. 6C shows the inter-group processing unit 33 used in the standard method, and FIG.

In FIG. 4, the intra-group processing unit 31 in FIG. 4A is means for converting internal data into intra-group standard data, and the intra-group processing unit 32 in FIG. 4B is a plurality of conversion processing units 32-1 to 3-3.
2-n, which is means for converting internal data into internal data according to the destination by the conversion processing units 32-1 to 32-n. Further, the intra-group processing unit 33 in FIG. 13C is a means for converting intra-group standard data into inter-group standard data, and the intra-group processing unit 34 in FIG. 9D is a plurality of conversion processing units 34-1 to 34-.
n, and by the conversion processing units 34-1 to 34-n,
It is a means for converting the intra-group standard data into intra-group standard data of the destination corresponding to the destination.

Next, the data conversion process in each of the methods shown in FIGS. 2A to 2D will be described with reference to FIGS. 3 and 4. (A) Intra-group standard and inter-group perfect coupling method (FIG. 2a) In the case of the intra-group standard and inter-group perfect coupling method of FIG.
4a is used as the intra-group processing unit 22a shown in FIG. 4 and the inter-group processing unit 34 shown in FIG. 4d is used as the inter-group processing unit 22b shown in FIG.

In FIG. 2a, when the computer 11a in the computer group 11 transmits data to the computer 11b in the same group, the application of the computer 11a processes the transmission data within the group 22a (FIG. 4a is used). Send to. The in-group processing unit 22a uniformly converts the data in the internal data expression format of the computer 11a into the in-group standard data expression format of the group 11 (internal data expression format of the computer 11c), and the inter-group processing unit 22b (FIG. 4d). Is used).

The conversion control unit 22c does not perform the conversion process by the inter-group processing unit 22b because the transmission destination of this data is within the own group 11 and the inter-group processing unit 22b does not perform the conversion process by the intra-group processing unit 22a. The data converted into the data representation format is sent to the communication control unit 23 as it is, and the destination address 23b
At the same time, the transmission data 23a is sent to the network.

In the recipient computer 11b, the communication controller 23 receives the data and sends it to the inter-group processor 22b.
Since the inter-group processing unit 22b does not require conversion as in the case of the computer 11a, the received data is sent to the intra-group processing unit 22a as it is. The in-group processing unit 22a converts the in-group standard format data into the internal data representation format of the computer 11b and sends it to the application 21.

Next, the computer 11a is replaced by another computer group 12
When sending data to the computer 12a in the
The intra-group processing unit 22a of the computer 11a uniformly converts the internal data representation format data of the computer 11a into the intra-group standard data representation format of the group 11, and the inter-group processing unit 22b (FIG. 4d is used). send. Since the transmission destination is the computer 12a of the group 12, the conversion control unit 22c has an inter-group processing unit 22b.
Of the conversion processing units (34-1 to 34-n in FIG. 4d), the conversion processing unit that converts to the standard data representation format within the group 12 (internal data representation format of the computer 12c) is selected,
The in-group standard data expression format of the group 11 is converted into the in-group standard data expression format of the group 12 and sent to the communication control unit 23, and the transmission data 23a is sent out to the network together with the destination address 23b.

In the recipient computer 12a, the communication control unit 23
Receives the data and sends it to the inter-group processing unit 22b. In the inter-group processing unit 22b, since the data has already been in the standard data representation format within the group of its own group 12, the data is sent to the intra-group processing unit 22a as it is. The in-group processing unit 22a converts the data in the in-group standard format of the own group into the internal data representation format of the computer 12a and sends it to the application 21. (B) In-group standard, inter-group standard system (FIG. 2b) In the case of the intra-group standard and inter-group standard system of FIG. 2b, the intra-group processing unit 31 shown in FIG. 4a as the intra-group processing unit 22a shown in FIG.
Is used, and the inter-group processing unit 33 shown in FIG. 4c is used as the inter-group processing unit 22b shown in FIG.

In FIG. 2b, when the computers 11a in the computer group 11 send data to the computers 11b in the same group, the same method as the intra-group standard and inter-group complete coupling method (FIG. 2a) described in (1) is used. In the intra-group processing unit 22a,
The data in the internal data representation format of the computer 11a is converted into the standard data representation format within the group 11 (internal data representation format of the computer 11c), and is transmitted from the communication control unit 23 without performing the conversion process by the inter-group processing unit 22b. The transmission data 23a is sent to the network together with the address 23b.

In the recipient computer 11b, as shown in FIG.
Similarly to the case of (3), the conversion processing is not performed by the inter-group processing unit 22b, and the received data is sent to the intra-group processing unit 22a as it is. The in-group processing unit 22a converts the in-group standard format data into the internal data representation format of the computer 11b and sends it to the application 21. Next, when the computer 11a transmits data to the computer 12a in another computer group 12, first, the intra-group processing unit 22a of the computer 11a transfers the data in the internal data representation format of the computer 11a to the group of group 11 It is converted into a standard data representation format and sent to the inter-group processing unit 22b (FIG. 4c is used).

Since the transmission destination of the conversion control unit 22c is the computer 12a of the group 12, the inter-group processing unit 22b (FIG. 4c).
To convert the data to the standard data representation format between groups,
The data is sent to the communication control unit 23, and the send data 23a is sent to the network together with the destination address 23b. In the recipient computer 12a, the communication controller 23 receives the data and sends it to the inter-group processor 22b. Group processing unit 22b (see FIG. 4)
In c), the inter-group standard data expression format is converted to the intra-group standard data expression format of the group 12 (internal data expression format of the computer 12c) and sent to the intra-group processing unit 22a. The in-group processing unit 22a converts the in-group standard data expression format into the internal data expression format of the computer 12a and sends it to the application 21. (C) Intra-group perfect coupling, inter-group perfect coupling method (FIG. 2c) In the case of intra-group perfect coupling and inter-group perfect coupling method of FIG. 2c,
The intra-group processing unit 32 shown in FIG. 4b is used as the intra-group processing unit 22a shown in FIG. 3, and the inter-group processing unit 22 shown in FIG.
The inter-group processing unit 34 shown in FIG. 4d is used as b.

In FIG. 2c, when the computer 11a in the computer group 11 transmits data to the computer 11b in the same group, the application of the computer 11a processes the transmission data in the group processing unit 22a (FIG. 4b is used). Send to. Since the transmission destination is the computer 11b of the own group 11, the conversion control section 22c has the conversion processing section (FIG. 4b) of the intra-group processing section 22b.
32-1 to 32-n), a conversion processing unit for converting to the internal data expression format of the computer 11b is selected, and the internal data expression format of the computer 11a is converted to the internal data expression format of the computer 11b, It is sent to the inter-processing unit 22b.

The inter-group processing unit 22b does not perform the inter-group processing because the destination of the data is within the self-group 11 and the inter-group processing is not required, and the data converted into the internal data representation format of the computer 11a is communication-controlled as it is. And sends the transmission data 23a together with the destination address 23b to the network. In the recipient computer 11b, the communication control unit 23 receives the data and sends it to the inter-group processing unit 22b. Since the inter-group processing unit 22b does not need to convert, the received data is sent to the intra-group processing unit 22a as it is. In-group processing unit 2
Like the inter-group processing unit 22b, the data 2a does not perform the conversion process and sends the data converted into the internal data representation format of the computer 11b to the application 21.

Next, the computer 11a is replaced by another computer group 12
When sending data to the computer 12a in the
The application of the computer 11a sends the transmission data to the intra-group processing unit 22a (FIG. 4b is used). Since the transmission destination of the conversion control unit 22c is not the own group 11, the conversion control unit 22c performs conversion processing units (32-1 to 32 in FIG. 4b) of the intra-group processing unit 22b.
In (n), a conversion processing unit for converting to the in-group standard data expression format (internal data expression format of the computer 11c) is selected, and the internal data expression format of the computer 11a is converted to the in-group standard data expression format of the group 11. And sends it to the inter-group processing unit 22b.

In the inter-group processing unit 22b, the conversion control unit 2
In 2c, since the transmission destination is the computer 12a of the group 12, the conversion processing unit (34-i in FIG. 4d) of the inter-group processing unit 22b.
34-n), a conversion processing unit for converting to the standard data expression format of the group 12 is selected, the standard data expression format within the group 11 is converted to the standard data expression format of the group 12, and the communication control unit 23 is selected. And send the send data 23a to the network together with the destination address 23b.

In the recipient computer 12a, the communication control unit 23
Receives the data and sends it to the inter-group processing unit 22b. In the inter-group processing unit 22b, since the data has already been in the standard data representation format within the group of its own group 12, the data is sent to the intra-group processing unit 22a as it is. The in-group processing unit 22a converts the data in the in-group standard format of the own group into the internal data representation format of the computer 12a and sends it to the application 21. (D) Intra-group perfect connection and inter-group standard method (Fig. 2d) In the case of intra-group perfect connection and inter-group standard connection method of Fig. 2d,
The intra-group processing unit 32 shown in FIG. 4b is used as the intra-group processing unit 22a shown in FIG. 3, and the inter-group processing unit 22 shown in FIG.
An inter-group processing unit 33 shown in FIG. 4c is used as b.

In this system, when data is communicated within a group, it is the same as the case described in (3) Complete coupling within a group and complete coupling between groups, and in the case of data communication across groups. Is the same as the case described in (2) intra-group standard and inter-group standard system. Although the data communication in the embodiment shown in FIG. 2 is performed as described above, when the computers 11a to 13c are divided into each group, the classification is performed by the degree of similarity of the internal data representation format of each computer,
Alternatively, they are classified according to the frequency of data representation format conversion in the network.

Here, the degree of similarity of the internal data is represented by some numerical value that influences the scale of the conversion mechanism, for example, the number of data expression formats that require conversion (this is a degree of dissimilarity). The frequency of format conversion is expressed by the degree of the data expression format conversion required in the network, for example, the probability that the data expression format conversion occurs.

Then, basically, those having a similar internal data expression format or those having a high frequency of data expression format conversion are classified into one group. In addition, as the standard data expression format within the group, the one having the most common data expression format in the group or the one in which the data expression format conversion is most frequently performed in the group is selected. The standard data representation format between groups will be selected in the same way as when selecting the standard data representation format within a group.

Generally, the degree of similarity of the internal data representation format is
If the group is organized with the essence, the scale of the conversion mechanism will
In addition, it is important to organize groups by emphasizing the frequency of format conversion.
If so, the speed of the overall conversion is more efficient. Next
First, in order to evaluate the effect of the method of the present invention, first,
The objective scale of degree and conversion scale is set as shown in (1) below.
Therefore, when the computer group is classified according to the frequency of format conversion,
Computer groups were classified according to the similarity of the internal data representation format.
In terms of the scale and speed of the conversion mechanism,
The effects of the conventional method and the method of the present invention are described in (2) and (3) below.
Described in. (1) Evaluation of conversion speed and conversion scale Now, N = {a₁, a₂,_...., a_n} Network, M
Conversion method in network N, a_iAs a calculator,
d_ijCalculator a_iAnd calculator a_jDifferent in internal representation between
Number of data representation formats, D_ijComputer by conversion method M
a_iAnd calculator a _jTotal number of conversions actually performed between
And

The conversion scale S _{M, N} in the conversion method M of the network N is defined by the following equation (1). In addition, when data representation format conversion occurs in the network N,
If the probability that it is a conversion between the computer a _i and the computer a _j is p _ij , the conversion speed F _{M, N} by the conversion method M of the network _N is determined by the following expression (2).

[0044]

[Equation 1]

The conversion scale is the total number of data expression format conversion mechanisms that must be prepared in the entire network, and the conversion speed is the data expression format conversion used for one data expression format conversion in the entire network. It is the average value of the number of mechanical parts. For example, assume that the internal data representation format of the _three computers a ₁ , a ₂ , and a ₃ is as follows.

Calculator / Data representation format Integer bit Long character string --------------------₁ big endian 16 ASCII II a₂ little endian 32 ASCII II₃ little endian 16 EBCDIC where conversion method M is a₂Is a standard method
(Calculator a₁And calculator a ₃In between, computer a₂Inside
Data conversion method via data representation format)
Then, the difference of internal data d_ijIs d₁₂= D₁₃= D_{twenty three}= 2, but the number of processes required for the actual conversion by the conversion method M
D_ijIs D₁₂= D_{twenty three}= 2D₁₃= D₁₂+ D_{twenty three}= 4.

Of course, conversion scale S _{M, N} and conversion speed F
_{Although M and N} are only rough evaluation scales, in order to simplify the problem, the method of the present invention is evaluated by the conversion scale S _{M, N} and the conversion speed F _{M, N} as described above. (2) When the computer group is classified according to the frequency of the conversion format. Network N consisting of _five computers a ₁ , a ₂ , ..., A ₅
think about.

Every two computers in the network N differ in the data representation format by one. That is, for all i and j (1 ≦ j ≦ n), d _ij = d _ji = 1. The probability distribution of data conversion is as follows. Table 1, a _i and the probability occurrence of data conversion _{a j p ij i / j a} 1 a 2 a 3 a 4 a 5 --------------------- −− a ₁ 0 0.5 0.1 0 0.1 a ₂ 0.5 0 0.1 0 0 a ₃ 0.1 0.1 0 0.1 0.1 a ₄ 0 0 0.1 0 0 a ₅ 0.1 0 0.1 0 0 The probability of data conversion p ₁₄ = 0 is calculated by the computer. This means that data conversion is not performed between a ₁ and the computer a.

Here, the network N is classified into the following three groups X, Y, Z, and X = {a ₁ , a ₂ , a ₃ } Y = {a ₄ } Z = {a ₅ } FIG. Full combination method, standard method,
The conversion scale S and the conversion speed F of each of the three hierarchical methods of the present invention will be compared.

In the figure, computers 41, 42, 43,
44 and 45 are computer a₁, a ₂, a₃, a_Four, a_FiveAgainst
Therefore, the numbers on the straight line connecting the computers are shown in Table 1.
Indicates the probability of data conversion. Note that the above comparison is performed
At this time, in the layering method of the present invention of FIG.
The data conversion is performed in the group by the complete connection method and by the group between the standard methods.
In addition, the standard data representation format within each group is calculated.
Calculator a₃, a_Four, a_FiveThe internal data representation format of
Data representation format is computer a₃Internal data representation format of
It (A) Perfect combination method (FIG. 5a) In the case of the perfect combination method, all i and j are always
And the total number of conversion processes D_ij= 1, so the conversion scale S_{1, N}And conversion speed F_{1, N}Is below
It becomes an expression.

S _{1, N} = 5 × 4 = 20 F _{1, N} = 1 × 0.5 +1 × 0.1 +1 × 0.1 +1 × 0.1 +1
× 0.1 + 1 × 0.1 = 1 (b) Standard method (Fig. 5b) If the standard data representation format is different from any internal data representation format of the computers a ₁ , a ₂ , a ₃ , a ₄ , a ₅ For i, j, the total number of conversion processes D _ij = 2 (however, when the standard data representation format is the internal data of a _k , the total number of conversion processes D for all i)
_ki = 1). Conversion scale S _{2, N} and conversion speed F _{2, N}
Is the following formula.

S _{2, N} = 5 × 2 = 10 F _{2, N} = 2 × (0.6 + 0.1 +0.1 +0.1 +0.1) = 2 (c) Method of the present invention (FIG. 5c) In this way, the data conversion is performed in the group by the complete coupling method and the group is performed by the standard method, and the standard data representation format within the group of each group is defined as the internal data representation format of the computers a ₃ , a ₄ and a ₅ , respectively. If the inter-standard data representation format is the internal data representation format of the computer a ₃ , the value of the total number of conversion processes D _ij is as shown in the table below.

Table 2, Value of total number of conversion processes D _ij i / j a ₁ a ₂ a ₃ a ₄ a ₅ −−−−−−−−−−−−−−−−−−−−−−−− a ₁ 0 1 1 2 2 a ₂ 1 0 1 2 2 a ₃ 1 1 0 1 1 a ₄ 2 2 1 0 2 a ₅ 2 2 1 2 0 Conversion scale S _{3, N} and conversion based on the values in the above table Speed F
_{When 3, N} is calculated, S _{3, N} = 2 +2 +2 +2 +2 = 10 F _{3, N} = 1 × (0.5 +0.1 +0.1 +0.1 +0.1) +2 × 0.
The smaller the value of 1 = 1.1 the conversion scale S _N , the smaller the scale of the conversion mechanism, and the smaller the value of the conversion speed F _N , the faster the conversion speed of the entire network.

Therefore, the layered scheme of the present invention of FIG. 5c achieves almost the same conversion speed as the fully-combined scheme of FIG. 5a with a smaller conversion scale than the standard scheme of FIG. 5b. Theoretically, the conversion speed F is the minimum in the fully coupled system, and the conversion scale S is the minimum in the standard system. The hierarchical system of the present invention can bring both the conversion speed F and the conversion scale S close to the minimum value. (3) When the computer groups are classified according to the similarity of the internal data representation format.

Consider a network N consisting of _five computers a ₁ , a ₂ , ..., A ₅ . It is assumed that any two computers in the network N uniformly perform data conversion. That is, for all i and j (1 ≦ j ≦ n), the probability of data conversion occurring p _ij = _1/5 C ₂ = 1/10 =
Set to 0.1. The number d _ij of types of data representation formats that must be converted between the computer a _i and the computer a _j follows the table below.

In Table 3, the (non) similarity of a _i and a _j d _ij i / j a ₁ a ₂ a ₃ a ₄ a ₅ −−−−−−−−−−−−−−−−−−−− −−−−− a ₁ 0 2 3 3 2 a ₂ 2 0 1 10 3 a ₃ 3 1 0 10 10 a ₄ 3 10 10 0 1 a ₅ 2 3 10 1 0 Grouping into groups X, Y and Z, X = {a ₁ } Y = {a ₂ , a ₃ } Z = {a ₄ , a ₅ } Full combination method, standard method as shown in FIGS. 6a, b and c ,
The conversion scale S and the conversion speed F of each of the three hierarchical methods of the present invention will be compared.

In the figure, computers 41, 42, 43,
44 and 45 are computer a₁, a ₂, a₃, a_Four, a_FiveAgainst
Therefore, the numbers on the straight line connecting the computers are shown in Table 3.
a_iAnd a_j(Non) similarity d of_ijIndicates. Note that the above comparison
In the layering method of the present invention of FIG. 6c,
Is the data conversion, and within the group is the complete combination method (even in the standard method)
The results will not change), the standard method will be used between groups, and
Calculating standard data representation formats for X, Y and Z groups respectively
a ₁, a₂, a_FiveThe internal data representation format of
Computer expression format₁The internal data representation format of. (A) Full combination method (Fig. 6a) In the case of the full combination method, all i and j are always
And the total number of conversion processes D_ij= 1, so the conversion scale S_{1, N}And conversion speed F_{1, N}Is below
It becomes an expression.

S _{1, N} = {(2 × 2 + 3 × 2) + (2 + 1
＋ 10 ＋ 3） ＋ (3 ＋ 1 ＋ 10 × 2） ＋ (3 ＋ 10 × 2 ＋ 1）
+ (2 +3 + 10 + 1)} / 2 = 90/2 = 45 F _{1, N} = S _{1, N} × 0.1 = 4.5 (b) Standard method (Fig. 6b) The internal data representation format of computer a ₁ is group standard data In terms of the expression format, the conversion scale S _{2, N} and the conversion speed F _{2, N} are as follows.

S _{2, N} = (0 + 2 + 3 + 3 + 2) × 2 = 20 F _{2, N} = 0.1 × {(2 + 3 + 3 + 2) + (2 + 5 + 5 +
4) + (3 + 5 + 6 + 5) + (3 + 5 + 6 + 5) + (2
+4 +5 +5)} = 8 (c) Method of the present invention (Fig. 6c) As described above, the data conversion is the complete combination method within the group, the standard method between the groups, and the group of groups X, Y, Z Assuming that the internal standard data representation format is the internal data representation format of the computers a ₁ , a ₂ and a _{5 and} the inter-group standard data representation format is the internal data representation format of the computer a ₁ , the total number of conversion processes D _ij is It looks like the table below.

Table 4, the value of the total number of conversion processes D _ij i / j a ₁ a ₂ a ₃ a ₄ a ₅ −−−−−−−−−−−−−−−−−−−−−−−− a ₁ 0 2 3 3 2 a ₂ 2 0 1 5 4 a ₃ 3 1 0 6 5 a ₄ 3 5 6 0 1 a ₅ 2 4 5 1 0 Conversion scale S _{3, N} and conversion based on the values in the above table Speed F
_{When 3, N} is calculated, S _{3, N} = (0 ＋2 +1 +1 +1) × 2 ＝ 12 F _{3, N} ＝ 0.1 × {(2 ＋3 ＋3 ＋2) ＋ (2 ＋1 ＋5 ＋
4) x 2 + (3 + 1 +6 +5) x 2} = 6.4 The smaller the conversion scale S _N , the smaller the scale of the conversion mechanism, and the smaller the conversion speed F _N , the conversion speed of the entire network. Means early.

Therefore, the layering scheme of the present invention in FIG. 6c has the smallest conversion scale S _N in FIGS. 6a, b, c,
It achieves a conversion speed F _N that is slower than the fully coupled scheme of FIG. 6a but faster than the standard scheme of FIG. 6b. On the contrary, in order to increase the conversion scale and increase the conversion speed, the hierarchization of the computer group may be changed.

In the above embodiment, the maximum number of computers is nine, the computers are divided into a plurality of groups, and each computer is arranged in each group. However, the present invention is not limited to this. The number of computers is not limited to the above-described embodiment, and the number of computers may be nine or more, and a plurality of groups may be provided in each group to form a multi-tier.

[0063]

As is clear from the above description,
In the present invention, between computers whose data representation formats are partially different, it is possible to perform high-speed communication between computers with the minimum required data conversion, and between computers that are in different computer groups, the data representation format is converted. However, since the number of computer groups is smaller than the number of different internal data formats, the conversion mechanism can be made compact as a whole.

Moreover, since program communication between different computer groups generally requires many changes other than the internal data representation format, it is rarer than program communication between the same computer groups, and data conversion is performed as described above. Even if it is performed twice as described above, the reduction in efficiency due to the conversion of the data representation format does not pose a problem. Further, according to the present invention, even if a computer having a new internal data representation format is added, the added computer is classified into some group,
The possibility of changing the inter-group conversion unit is low.

[Brief description of drawings]

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

FIG. 2 is a diagram showing an example of the present invention.

FIG. 3 is a diagram showing a data conversion mechanism of each computer according to the present invention.

FIG. 4 is a diagram showing configurations of an intra-group processing unit and an inter-group processing unit according to the present invention.

FIG. 5 is a diagram (part 1) comparing the effects of the present invention and the conventional example.
Is.

FIG. 6 is a diagram (part 2) comparing the effects of the present invention and the conventional example.
Is.

FIG. 7 is a diagram showing a conventional example.

[Explanation of symbols]

1, 11a, 11b, 11c, 12a, 12b, 12
c, 13a, 13b 13c Computers 2, 11, 12, 13 Computer group 3 Standard data representation format within group 4, 14 Standard data representation format between groups 22 Conversion control unit 22a, 31, 32 Intra-group processing unit 22b, 33, 34 Inter-group processing unit 22c Conversion control unit 22d Computer group management table 23 Communication control unit

Claims (8)

[Claims] 1. A plurality of computers (1) are connected via a network, and each computer (1) provides a means for calling a program of another computer (1) and a service to the other computer (1). In the data representation conversion method of the distributed processing system, in which the data representation formats are converted between computers having different internal data representation formats (1) and the data representation conversion method is used, the computers connected to the network (1 ) Is hierarchically divided into multiple computer groups (2), and each computer group (2) has
At least, an intra-group standard data representation format (3) that represents the computer group (2) is provided, and each computer (1) has data that is different from the computer (1) of another computer group (2) that has a different internal data representation format. When communicating, each computer group (2) is sent via the group standard data representation format (3).
Hierarchical data representation conversion method characterized by mutual data communication between them. 2. Each computer (1) is between the in-group standard data representation format (3) of its own computer group (2) and all in-group standard data representation format (3) of the other computer group (2). The hierarchical data representation conversion method according to claim 1, further comprising: 3. A plurality of computers (1) are connected via a network, and each computer (1) provides a means for calling a program of another computer (1) and a service to the other computer (1). In the data representation conversion method of the distributed processing system, in which the data representation formats are converted between computers having different internal data representation formats (1) and the data representation conversion method is used, the computers connected to the network (1 ) Is hierarchically divided into multiple computer groups (2), each computer group (2) is provided with a standard data representation format (3) within the group that represents the computer group (2), and , The inter-group standard data representation format (4) for the computers (1) to communicate between each computer group (2), and each computer (1) has another internal data representation format (2) ) Data communication with the computer (1) via the intergroup standard data representation format (4). , Hierarchical data representation conversion method and performing the mutual data communication between each computer group (2). 4. The computer according to claim 3, wherein each computer (1) has a conversion mechanism between the standard data representation format within a group (3) and the standard data representation format between groups of its own computer group (4). Hierarchical data representation conversion method. 5. A computer group (2) to which each computer (1) belongs, as well as the internal data representation format of that computer (1).
5. The hierarchical data representation conversion system according to claim 1, 2, 3 or 4, having a conversion mechanism between the standard data representation format within group (3). 6. A computer group (2) to which each computer (1) belongs and the internal data representation format of that computer (1)
5. The hierarchical data representation conversion method according to claim 1, 2, 3 or 4 having a conversion mechanism for converting between all types of data representation formats. 7. Hierarchical data according to claim 1, 2, 3, 4, 5 or 6, wherein computers (1) having similar internal data constitute one computer group (2). Expression conversion method. 8. A computer that requires frequent data communication.
7. The hierarchical data representation conversion system according to claim 1, 2, 3, 4, 5 or 6, wherein one computer group (2) is constituted by (1).