JP2005354364A - Distributed data processing and managing device, and its method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a distributed data processing and managing device which improves a system response time by eliminating the waste of a routing table, by adopting a routing table construction method different from a FINGER and reducing the number of times of transfers smaller than a Chord system in the case of an internet wide system having many computers, and to provide its method. <P>SOLUTION: The distributed data processing and managing device includes other distributed data processing and managing devices, and a communication processing unit which transmits and receives a message; and a route table processing unit for determining the distributed data processing and managing device to a transfer destination if it detects that the message is not data in charge of and reconfigurates the routing table. The i-th entry is based on an identifier which 2<SP>i</SP>is separated from an own identifier by a routing table construction of a Chord System. On the other hand, the routing table processing unit reconfigurates the routing able to the i-th entry from the own identifier based on the identifier separated by unit×base<SP>i</SP>as a reference. Unit=(identifier of a device which in turn appears in the right direction from the own identifier)-(own identifier). Base=2×unit<SP>(-1/160)</SP>. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a distributed data processing system to which a distributed hash table technique is applied, and more particularly, to a distributed data processing / management apparatus and a method therefor characterized by a configuration of a routing table designed to speed up the response of the system.

A system using a distributed hash table technique is known as a distributed data processing and management system configured by a group of computers connected continuously or temporarily to a network without depending on a specific server.
In such a system, a group of computers connected continuously or temporarily to a network are connected to each other using an application layer network on the basis of an identifier obtained by applying a hash function.

Data to be processed or managed is transferred on the application layer network by an identifier obtained by applying a hash function, and is processed or managed on the transfer destination computer.
In order to speed up the response of such a system, it is desirable to reduce the time required for transfer on the application layer network.

Here, The Chord Project of MIT University in the United States is well known as a distributed hash table technique applied to a distributed data processing and management system (see Non-Patent Documents 1 and 2).
The Chord system employs SHA-1 as a hash function and constitutes a 160-bit hash space.
The data to be processed or managed is assigned a 160-bit identifier by the hash function SHA-1.

Computers constituting the system are also recognized by a network identifier, specifically, a 160-bit identifier to which a hash function SHA-1 is applied to a set of an IP address and a port number.
Here, the 160-bit hash space is assumed to circulate as shown in FIG.
That is, when viewed from the 2 ¹⁶⁰ -1 identifier, the identifier advanced by 1 in the positive direction (arrow A) is 0, while the identifier advanced by 1 in the negative direction (arrow B) when viewed from the identifier 0 is 2 ¹⁶⁰ -1.

Data to be processed or managed is handled by one of the computers constituting the system based on the assigned identifier.
Further, when processing or managing the data, if there is no computer having the corresponding identifier, the computer having the next appearing identifier in charge in the positive direction as viewed from the data identifier is in charge.
And the computer which comprises a system communicates with each other by comprising an application network.

The application network at this time is configured as follows.
A computer constituting the system recognizes a computer having an identifier that appears next in the positive direction as viewed from its own identifier as its own SUCCESSOR.
Recognizing in this case indicates that the IP address and port number of SUCCESSOR are known (held) and that a message can be transmitted to SUCCESSOR.
Similarly, a computer constituting the system recognizes a computer having an identifier that appears next in a negative direction as viewed from its own identifier as its own PREDECESSOR.
Recognizing in this case means that the IP address and port number of the PREDECESSOR are known (held), and it is clear that the message can be sent to the PREDECESSOR.

On the application network configured as described above, when a computer with identifier X sends a message to a computer in charge of data with identifier Y, the message transmission is completed by repeating the transfer to SUCCESSOR. It is obvious to do.
However, when the number of computers constituting the system is N, the number of message transfers is N / 2 on average, and there is a problem that it takes too much time when N is a large number.
For this reason, the Chord system employs a routing table (route table) called FINGER for the above problem.

A computer (node) constituting the system has a route table composed of 160 entries from entry 0 to entry 159.
Here, the computer having an identifier X, node as the entry i (0 ≦ i ≦ 159) , X + 2 i mod2 160 identifier computer responsible for data (i.e., whether the computer just having an identifier of X + ^{2 i} mod2 ¹⁶⁰ If there is no such computer, the computer has an identifier that appears next in the positive direction as viewed from the identifier of X + 2 ⁱ mod2 ¹⁶⁰ ).

At this time, the identifier of the corresponding computer and the network identifier, that is, the IP address and the port number are stored in the route table of entry i.
When each computer has a routing table configured in this way, a process in which a computer having an identifier X transmits a message to a computer in charge of data having an identifier Y on an application network composed of these computers Is as follows.
First, the entry i searches its routing table, extracts an entry including an identifier that is closest to Y among identifiers that do not exceed Y in the positive direction, and transfers the message to a computer having that identifier.

Similarly, the transfer destination computer also searches its routing table, extracts an entry including an identifier that approaches Y most among identifiers that do not exceed Y in the positive direction, and sends a message to the computer having that identifier. Forward.
In subsequent computers, it is obvious that the message reaches the computer having the identifier Y by repeating the same processing.
Since the distance in the hash space is halved every time transfer is performed due to the configuration of the routing table, in a system employing FINGER, when the number of computers constituting the system is N, the message is the maximum log ( N) The destination is reached by performing the transfer.

The routing table itself can also be efficiently constructed by using the above characteristics.
The entry i is constructed recursively in the order from i = 0 to i = 159.
When i = 0, X + 2 ⁰ mod2 ¹⁶⁰ = X + 1 mod2 ¹⁶⁰ , and entry 0 is SUCCESSOR (a computer having an identifier that appears next in the positive direction as viewed from X).
Here, assuming that entry i-1 has already been obtained, to obtain entry i, a computer in charge of the data of identifier X + 2 ⁱ mod2 ¹⁶⁰ (X + 2 ⁱ mod2 ¹⁶⁰ just or a computer having such an identifier is used. If it does not exist, the message is transmitted to the computer of the identifier that appears next in the positive direction as seen from the X + 2 ⁱ mod2 ¹⁶⁰ computer.

However, at this time, the entry closest to the computer of the identifier X + 2 ⁱ mod2 ¹⁶⁰ from the identifiers that do not exceed the identifier X + 2 ⁱ mod2 ¹⁶⁰ in the positive direction among the entries 0 to i-1 that have already been obtained. That is, the message may be transferred to the computer of entry i-1.
By the transfer described above, the distance on the hash space is halved.
If the routing table is appropriately configured in the transfer destination computer, the distance is also halved for each transfer from the transfer destination.

For this reason, the computer in charge of the data of the identifier X + 2 ⁱ mod2 ¹⁶⁰ , which is the destination at the maximum of i transfers (X + 2 ⁱ mod2 ^160, or if there is no computer having such an identifier, from X + 2 ⁱ mod2 ¹⁶⁰ The message arrives at the next identifier computer in the positive direction.
In this way, entry i is obtained. Through the above processing, entry 0 to entry 159 can be efficiently configured.
By periodically repeating the method described above on each computer that constitutes the system, the routing table of each computer using FINGER is maintained appropriately.
http: // www. pdos. lcs. mit. edu / chord / Ion Stoica, Robert Morris, David Karger, M.M. Frans Kaashoek, and Hari Balakrishnan, Chord: A Scalable Peer-to-peer Lookup Service for Internet Applications, ACM SIGCOMM 2001, SanDeigo, CA, August 2001, pp. 149-160

As described above, the Chord system shown in Patent Documents 1 and 2 adopts the routing table construction method by FINGER, and when the number of computers constituting the system is N, the maximum is 1 og (N) times on the application network. By repeating log (N) / 2 transfer, a message can be transmitted between any two computers.
However, assuming that the Chord system is applied to the Internet and the number of computers constituting the system is about 1 million, log (10 ⁶ ) ≈20. Required on the network.

This is the number of transfers on the application network, and when considering the number of routers that pass through, this is a larger number of transfers.
Therefore, although the Chord system is a theoretically efficient system, when considering application to an actual operating environment, the number of times of transfer should be as small as possible due to response time constraints.
In addition, as shown below, the Chord system's FINGER routing table construction method has a large waste.

When the number of computers constituting the system is about 1 million (≈2 ²⁰ ), it is expected that the respective identifiers are uniformly scattered on the 160-bit hash space by the hash function SHA-1.
In this case, two adjacent identifier will be present in an average of 2 ¹⁴⁰ away.
That is, for a computer having the identifier X, the SUCCESSOR waits for an identifier of approximately X + 2 ¹⁴⁰ mod2 ¹⁶⁰ .

In this case, in the routing table construction method by FINGER, this means that all entries from entry 0 to entry 140 wait for the same entry.
For this reason, 160 entries are stored in the routing table, but most of them are wasted.
FINGER was able to halve the hash space distance to the destination for each transfer, but it was more efficient than FINGER for each transfer by devising a wasteful entry. It is possible to narrow the distance in the hash space to the destination well.

  The present invention has been made in view of such circumstances. By adopting a routing table configuration method different from FINGER, the routing table is not wasted, and the Internet-wide configuration including about 1 million computers. In the case of a system, the present invention provides a distributed data processing / management apparatus and method that realizes a smaller number of transfers than the Chord system and improves the response time of the system.

A distributed data processing / management apparatus according to the present invention is a distributed data processing / management apparatus having a message transfer function similar to that of a Chord system having a 160-bit hash space using a distributed hash table technology, A communication processing unit that is connected and transmits / receives a message to / from another distributed data processing / management device, and if it detects that the message received by the communication processing unit is not data that it is responsible for, should transfer the message to determine the other distributed data processing / management unit, and a routing table processor for reconstructing the routing table, the routing table configuration method Chord system, for the i-th entry, 2 ⁱ from its own identifier The routing table processing unit uses the identifier separated by unit × base ⁱ from the identifier as a reference. The routing table is reconfigured with respect to the entries.
unit = (identifier of the next device that appears in the forward direction from its own identifier)-(own identifier)
base = 2 × unit ^(-1/160)

  The distributed data processing / management apparatus of the present invention includes a routing table management unit in which the routing table indicating a message transmission route is stored, and the routing table processing unit performs routing in the routing table management unit at regular intervals. The table is reconfigured.

A distributed data processing / management method of the present invention is a distributed data processing method having a message transfer process similar to that of a Chord system having a 160-bit hash space using a distributed hash table technique, and is connected to a communication network. A communication process for transmitting / receiving a message to / from another distributed data processing / management apparatus, and when it is detected that the message received in the communication process is not the data handled by itself, A routing table processing process for reconfiguring a routing table to determine a distributed data processing / management device, and the Chord system routing table configuration method is 2 ⁱ away from its own identifier for the i-th entry In contrast to an identifier as a reference, in the routing table processing process, an identifier separated by unit × base ⁱ from its own identifier as a reference, The routing table is reconfigured for the i-th entry.
unit = (identifier of the next device that appears in the forward direction from its own identifier)-(own identifier)
base = 2 × unit ^(-1/160)

  The distributed data processing / management method of the present invention includes a routing table management unit in which the routing table indicating a message transmission route is stored, and the routing table processing process is performed at regular intervals, and the routing table management is performed. The route table is reconstructed in the section.

In the Chord system, the configuration of the routing table of a computer having the identifier X is the identifier of the computer in charge of data having the identifier “X + 2 ⁱ mod2 ¹⁶⁰ ” for the entry i.
On the other hand, in the present invention, the number of entries in the routing table is 160 entries as in the Chord system FINGER, but for the entry i, data having an identifier with a value obtained from the following equation (1) is used. Use the identifier of the computer in charge.
Identifier of entry i = X + unit × base ⁱ mod2 ¹⁶⁰ (1)
However,
unit = (identifier of X SUCCESSOR) −X
base = 2 × unit ^(-1/160)
And

In this embodiment, when the 160-bit hash space is divided 160 times by the same base number, the base value is selected so that “(X SUCCESSOR identifier) −X” is finally obtained. Yes.
By selecting base as described above, when i = 0, the above equation (1) becomes the identifier of SUCCESSOR, and entry 0 becomes SUCCESSOR.
Further, when i = 1, since base is 1 or more, it can be seen that entry 1 is an identifier that exceeds the identifier of SUCCESSOR in the positive direction. That is, entry 0 and entry 1 are different.

In the case of FINGER of the Chord system, the unit is set to 1 and the base is set to 2. Therefore, the apparatus has the same identifier in most entries, but in the present invention, the apparatus has the same identifier. It becomes possible to avoid that.
Here, only the target of each entry is different, and the other configuration is the same process as the route table (or route table) configuration method by FINGER in the Chord system, and the message transfer is also the same.

  As described above, according to the distributed data processing / management apparatus of the present invention, by devising the configuration method of the routing table, the Internet-wide system composed of about 1 million computers from the simulation results On the other hand, with the same amount of work as the conventional Chord system, the number of transfers is approximately half that of the Chord system, improving the response time of the system, and wasteful entry in the routing table compared to the Chord system. This makes it possible to effectively use the route table.

Hereinafter, an apparatus according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of a distributed data processing / management apparatus according to the embodiment. FIG. 1 shows a form in which a plurality of distributed data processing / management apparatuses are connected to each other via a communication network T.
In this figure, a distributed data processing / management apparatus according to the present embodiment is connected to a communication network T, and includes data to be transmitted / received to / from other distributed data processing / management apparatuses. A communication processing unit 2 for detecting whether or not a message received by the communication processing unit 2 includes data handled by the communication processing unit 2, and a data processing unit 3 for performing processing when the message is handled by the communication processing unit 2. A data management unit 4 for reading or writing data to be read, and another distributed data processing / management device to which the message is to be transferred when the message received by the communication processing unit 2 is not the data handled by itself. In addition, a route table processing unit 6 for periodically reconfiguring the route table management unit 5 in order to keep the route table appropriate, and a route table processing unit 6 Thus it has a routing table managing unit 5 for storing the periodically reconfigured routing table, the.
As described above, the distributed data processing / management apparatus is connected to the communication network T via the communication processing unit 2 and uses the message distribution function provided by the communication network T to perform other distributed data processing / Send and receive messages between management devices.

Next, the routing table stored in the routing table management unit 6 will be described with reference to FIG. FIG. 2 shows an example of the data structure of the routing table stored in the routing table management unit 6.
The routing table is composed of 160 entries from entry 0 to entry 159.
Each entry includes an entry number, an identifier of the distributed data processing / management apparatus, an IP address of the distributed data processing / management apparatus, and a port number of the distributed data processing / management apparatus.
The routing table further has two special entries for managing SUCCESSOR and PREDECESSOR, and the configuration of these entries is the same.

Next, with reference to FIG. 3, a message transmitted / received between the distributed data processing / management apparatuses will be described. FIG. 3 shows an example of the format of the message.
As shown in FIG. 3, the message includes a transmission destination identifier indicating a transmission destination of the message, a command indicating a processing method of data carried by the message, data to be transmitted, an IP address of the transmission source device, and a transmission source device. Each item has a port number.

Next, with reference to FIG. 1 and FIG. 4, the operation of the data reception process of the distributed data processing / management apparatus 1 of the present embodiment will be described. FIG. 4 is a flowchart showing the flow of data reception processing of the distributed data processing / management apparatus 1 of this embodiment.
After connecting to the communication network T, the communication processing unit 2 receives a message from another distributed data processing / management apparatus (step S1).
Then, the communication processing unit 2 looks at the transmission destination identifier of the received message and detects whether or not it includes data handled by itself, that is, determines whether or not the data is to be handled by itself.

That is, the communication processing unit 2 is data to be handled by itself when the message destination identifier is in the positive direction as viewed from the PREDECESSOR identifier and in the negative direction as viewed from its own identifier. If any of the above-mentioned conditions is not satisfied, it is detected that the data is not to be handled by itself (step S2).
The communication processing unit 2 outputs this message to the data processing unit 3 when determining that the data is data to be handled by the communication processing unit 2 itself.

As a result, the data processing unit 3 performs data processing on the included data in accordance with the command indicated in the message.
At that time, the data processing unit 3 writes data to the data management unit 4 or reads data as necessary.
Further, the data processing unit 3 responds to the distributed data processing / management apparatus of the transmission source using the message transmission source IP address and the transmission source port number as necessary (step S3). .

On the other hand, when the communication processing unit 2 detects from the transmission destination identifier of the received message that this message does not include data to be handled by itself, the communication processing unit 2 sends this message to another distributed data processing / management apparatus. Attempt to transfer.
At this time, the communication processing unit 2 sends the message from the routing table stored in the routing table management unit 6 via the routing table processing unit 3 to the message that does not exceed the message destination identifier in the positive direction. An entry having an identifier approaching the transmission destination identifier included in is searched (step S4).

Then, the routing table processing unit 4 reads the IP address and port number of the distributed data processing / management apparatus to be transferred from the entry extracted from the routing table, and determines the distributed data processing / management apparatus as the message transfer destination. Determine (step S5).
As a result, the communication processing unit 2 transfers the message to the distributed data processing / management apparatus indicated by the obtained IP address and port number (step S6).

Next, the operation of the data transmission process of the distributed data processing / management apparatus 1 according to this embodiment will be described with reference to FIGS. FIG. 5 is a flowchart showing the flow of data transmission processing of the distributed data processing / management apparatus 1 of this embodiment.
The communication processing unit 2 applies a hash function SHA-1 to obtain a predetermined numerical value when it becomes necessary to transmit a message to another distributed data processing / management apparatus for data processing or management. The message shown in FIG. 3 is generated, which includes the transmission destination identifier, the command indicating the data processing method, the data, the own IP address that is the transmission source, and the own port number (step S11).

Next, the routing table processing unit 5 searches the routing table stored in the routing table management unit 6 and selects the identifier closest to the destination identifier from those that do not exceed the message destination identifier in the positive direction. An entry having “” is extracted (step S12).
Then, the routing table processing unit 5 reads the IP address and port number of the distributed data processing / management device to be transferred from the extracted entry, and determines the transmission destination distributed data processing / management device (step S14). .
Thereby, the communication processing unit 2 transmits a message to the distributed data processing / management apparatus indicated by the IP address and port number obtained as described above (step S14).

Next, with reference to FIG. 1 and FIG. 6, the operation of the routing table reconfiguration process of the distributed data processing / management apparatus 1 of the present embodiment will be described. FIG. 6 is a flowchart showing the flow of the routing table reconfiguration process of the distributed data processing / management apparatus 1 of this embodiment.
The routing table processing unit 5 sets the target identifier I for obtaining the identifier of another distributed data processing / management apparatus to be set in the entry i to 0 and performs initialization (step S21).
Next, the routing table processing unit 5 calculates the target identifier I.

Specifically, the target identifier is calculated using the following equation.
Target identifier = X + unit x base ⁱ mod 2 ¹⁶⁰
Where X = own identifier
Unit = SUCCESSOR identifier-X
Base = 2 × unit ^(−1/160)
It is.

Also, depending on the program processing system, the power of calculation for huge number such more than 2 ¹⁰⁰ can be difficult.
For example, when the Java (registered trademark) language is the above, in such a case, the following modified expression is used for the calculation of the base.
Base = 2 × unit ^(−1/160) = 2 × 2 ^{( −log (unit) / 160)}
Here, log (unit) may be approximately the bit length of unit.

there,
Base = 2 × 2 ^{(-(unit bit length) / 160)}
It becomes.
And if it is Java (registered trademark) language, unit is BigInteger,
Base = 2 × Math.pow (2, − (unit.bitLength ()) / 160)
It becomes possible to calculate.
The route table processing unit 5 obtains the target identifier I by the arithmetic processing as described above (step S22).

Next, the routing table processing unit 5 determines whether or not the information on the distributed data processing / management apparatus to be stored in the entry i is known from the obtained target identifier.
At this time, the routing table processing unit 5 determines that it is always known when I = 0, while it is known when I ≧ 1 does not exceed the identifier of the entry i−1 in the positive direction. If it exceeds, it is determined that it is not known (step S23).
Then, the routing table management unit 5 notifies the communication processing unit 2 of the determination result when the information on the distributed data processing / management apparatus related to the target identifier is not known.
As a result, the communication processing unit 2 receives the target identifier obtained as the transmission destination identifier, a command for prompting a response from the distributed data processing / management apparatus in charge, null data as its data, its own IP address, and its own A message having a port number is generated (step S24).

Next, the communication processing unit 2 transmits the created message to the IP address and port number of the distributed data processing / management apparatus of the entry i-1.
As a result, in the distributed data processing / management apparatus of the transmission destination, the message transfer is repeated to the distributed data processing / management apparatus that determines that it is in charge as in the message reception process described above. Do.
Then, in the distributed data processing / management apparatus to which the message finally arrives, information about itself, that is, the response of the identifier, the IP address, and the port number is sent to the message transmission source (step S25). ).

Next, the communication processing unit 2 (distributed data processing / management apparatus of the transmission source) receives a response from the distributed data processing / management apparatus terminal to which the message finally arrives (step S26).
When the communication processing unit 2 determines that the distributed data processing / management device related to the target identifier is known or receives a response, the communication processing unit 2 reconfigures the routing table with respect to the routing table processing unit 2. Instruct.
As a result, the routing table processing unit 5 determines that the distributed data processing / management apparatus related to the target identifier is not known, and receives a response, the route table entry i stored in the routing table management unit 6 Is updated (reconfigured).

At this time, when receiving a response, the routing table processing unit 5 updates the entry i with the identifier, IP address, and port number extracted from the response message.
On the other hand, when it is determined that the distributed data processing / management apparatus related to the target identifier is known, the routing table processing unit 5 updates the entry i with the same identifier, IP address, and port number as the entry i-1. (Step S27).
Then, the routing table processing unit 5 detects whether or not i is 160 or more. When i is 160 or more, the routing table processing unit 5 ends the routing table reconstruction process, and i is less than 160. If it is detected, the process proceeds to step S9 (step S28).

Next, the route table processing unit 5 increments i by 1 and returns the process to step S22 (step S29).
The above-described processing is a series of procedures for reconfiguring the routing table.
Further, the distributed data processing / management apparatus 1 repeats the above-described routing table reconstruction processing periodically (eg, every minute) in a certain cycle, and always keeps the routing table appropriately.

Next, simulation results of the present embodiment will be described with reference to FIGS. 7 and 8 are tables and graphs showing simulation results for a system composed of 1 million devices.
The columns in the table shown in FIG. 7 correspond to the number of devices used in the simulation, the average number of transfers using Chord corresponding to the number of devices, the average number of transfers in the present invention corresponding to the number of devices, and the number of devices, respectively. The average number of transfers in the present invention is divided by the average number of transfers in Chord.

FIG. 8 is a graph plotting the average number of transfers in Chord and the average number of transfers in the present invention, the horizontal axis is the number of devices used in the simulation, and the vertical axis is the total number of average points. Show.
As is clear from FIGS. 7 and 8, the simulation results show that the present invention realizes about half the number of transfers compared to the Chord system, and the number of transfers can be reduced.

  The program for realizing the functions of the distributed data processing / management apparatus in FIG. 1 is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system and executed. Thus, distributed data processing / management in message transmission / reception may be performed. The “computer system” here includes an OS and hardware such as peripheral devices. The “computer system” includes a WWW system provided with a homepage providing environment (or display environment). The “computer-readable recording medium” refers to a storage device such as a flexible medium, a magneto-optical disk, a portable medium such as a ROM and a CD-ROM, and a hard disk incorporated in a computer system. Further, the “computer-readable recording medium” refers to a volatile memory (RAM) in a computer system that becomes a server or a client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. In addition, those holding programs for a certain period of time are also included.

  The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

2 shows an internal configuration of a distributed data processing / management apparatus and a state in which a plurality of distributed data processing / management apparatuses 1 are connected to a communication network T. It is a conceptual diagram which shows the structural example of the routing table stored in the routing table management part of FIG. 2 shows an example of a format of a message transmitted / received between distributed data processing / management apparatuses. It is the flowchart which showed the procedure regarding the message reception of a distributed data processing / management apparatus. It is the flowchart which showed the procedure regarding the message transmission of a distributed data processing / management apparatus. It is the flowchart which showed the procedure of the routing table reconstruction of a distributed data processing / management apparatus. It is a table | surface which shows a response | compatibility with the number of apparatuses used and the said average frequency | count in the simulation result which calculates | requires the average transmission frequency | count about each of Chord system and this invention. It is a graph which shows a response | compatibility with the number of apparatuses to be used, and the said average frequency in the simulation result which calculates | requires the average frequency | count of transfer about each of the Chord system and this invention. It is a conceptual diagram explaining the circularity of the identifier in a hash space (an entry moves sequentially clockwise).

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Distributed data processing / management apparatus 2 ... Communication processing part 3 ... Data processing part 4 ... Data management part 5 ... Routing table processing part 6 ... Routing table management part T ... Communication network

Claims (4)

A distributed data processing / management apparatus having a message transfer function similar to that of a Chord system having a 160-bit hash space using a distributed hash table technique,
A communication processing unit connected to a communication network and transmitting / receiving messages to / from other distributed data processing / management devices;
When the communication processing unit detects that the received message is not its own data, a routing table process for reconfiguring the routing table to determine another distributed data processing / management device to which the message is to be transferred And
Whereas the Chord system routing table construction method is based on an identifier that is 2 ⁱ away from its own identifier for the i-th entry,
The distributed data processing / management apparatus, wherein the routing table processing unit reconfigures a routing table for the i-th entry with reference to an identifier separated by unit × base ⁱ from its own identifier.
unit = (identifier of the next device that appears in the forward direction from its own identifier)-(own identifier)
base = 2 × unit ^(-1/160) A routing table management unit in which the routing table indicating the transmission route of the message is stored;
2. The distributed data processing / management apparatus according to claim 1, wherein the routing table processing unit reconfigures the routing table in the routing table management unit at regular intervals. A distributed data processing / management method having a message transfer process similar to that of a Chord system having a 160-bit hash space using a distributed hash table technique,
A communication process connected to a communication network and transmitting / receiving messages to / from other distributed data processing / management devices;
A route table processing step for reconfiguring a route table to determine another distributed data processing / management device to which the message is to be transferred when it is detected that the received message is not the data handled by the communication processing step. And
Whereas the Chord system routing table construction method is based on an identifier that is 2 ⁱ away from its own identifier for the i-th entry,
A distributed data processing / management method characterized in that in the routing table processing step, a routing table is reconfigured for an i-th entry with reference to an identifier separated by unit × base ⁱ from its own identifier.
unit = (identifier of the next device that appears in the forward direction from its own identifier)-(own identifier)
base = 2 × unit ^(-1/160) A routing table management unit in which the routing table indicating the transmission route of the message is stored;
4. The distributed data processing / management method according to claim 3, wherein the routing table processing step is performed at regular intervals, and the routing table is reconfigured in the routing table management unit.

Images