JP5508346B2 - Distributed data management system, distributed data management method, and distributed data management program - Google Patents

Description

  The present invention relates to a distributed data management system, a distributed data management method, and a distributed data management program for managing data distributed to a plurality of servers.

  When performing large-scale data management like a database, a distributed data management system is used to connect a plurality of servers to obtain high processing performance. The distributed data management system is a distributed system in which a plurality of servers operate in cooperation, and each server shares and manages data managed by the entire system. Such a distributed data management system is used in fields requiring high real-time properties such as medical care and communication, and is determined for each client application in response to a data acquisition request from a client using the distributed data management system. It is desirable to return data within a certain time.

  As a technique for managing data by distributing it to a plurality of servers, there is consistent hashing (see Non-Patent Document 1). As shown in FIG. 7, consistent hashing assigns an identifier to each of a plurality of nodes 130 (130A to 130F), which are servers that implement a distributed data management program, and data managed by the node 130. The range of identifiers of data managed by the node 130 is determined between the identifier of the node 130 and the identifier of the node 130 having the largest identifier among the nodes 130 having identifiers smaller than the node 130. Technology. As shown in FIG. 7, the node 130 and the data arranged in the order of the identifier values are called ID space in the consistent hashing. Here, since the node 130 and the identifier of the data are determined by a method of calculating a random value such as a hash function, the amount of data handled by each node 130 is stochastically equal and distributed data management In the system, it is possible to perform load sharing close to equality for each node 130.

  Each node 130 holds an address table in which identifiers of all other nodes 130 are associated with addresses (such as IP addresses) of the nodes 130. In a distributed data management system using consistent hashing, even if the number of nodes 130 constituting the system increases or decreases, only a part of the data is reassigned among a plurality of nodes 130. From the viewpoint of the entire distributed data management system, only certain data movement occurs regardless of the number of nodes 130.

D. Karger, E. Lehman, T. Leighton, M. Levine, D. Lewin, and R. Panigrahy, “Consistent Hashing and Random Trees: Distributed Caching Protocols for Relieving Hot Spots on the World Wide Web,” in Proceedings of the 29th ACM Symposium on Theory of Computing (STOC'97), pp.654-663, May 1997.

  Here, Non-Patent Document 1 does not disclose a specific method of data transfer when the number of nodes 130 is dynamically changed. The simplest method for data transfer is to process all data transmission (sending to a client that requires data) / storage (storage at the node in charge of data) request when the number of nodes 130 is changed. Can be considered, and all the data is transferred from the previous responsible node 130 to the new responsible node 130, and then the data transmission / storage request processing is resumed.

  However, in fields that require high real-time properties such as medical care and communication, the method of transferring all data by stopping the data transmission / storage request in this way takes a long time to transfer the data, Application of data transmission / storage request is difficult because it is delayed.

  Therefore, when data is actually needed on the new node 130, a method of transferring this data from the previous node 130 to the new node 130 and performing processing can be considered. In this method, data transfer occurs sporadically compared to the method of transferring all data in advance, so data can be transferred while accepting data transmission / storage requests, which has an impact on real-time performance. Few. On the other hand, the time until the data transfer is completed is long, and the data transfer will not be completed until an “actual data is required” opportunity such as a data transmission / storage request occurs for all data. .

  In such a data transfer method, when a data transmission / storage request arrives at the node 130 newly added to the distributed data management system and it is confirmed that there is no data to be processed on the node 130, It is conceivable to search for which node 130 is the previous responsible node 130.

  In general, as shown in FIG. 8A, the previous responsible node 130A is likely to be a node 130 having an identifier larger than the newly added node 130G and having the closest identifier value. Therefore, it is conceivable that the new node 130G makes an inquiry about the presence / absence of data to the node 130A. As described above, data transfer from adjacent nodes 130 in the ID space is established when the number of nodes 130 added at a time is small (for example, only one).

  On the other hand, when the number of nodes 130 added at a time is large, as shown in FIG. 8B, one or more other new nodes are placed between the new responsible node 130G and the previous responsible node 130A. There may be a node 130 (130H, 130I, 130J). In such a case, even if a new node 130G that has simply received a data transmission / storage request makes an inquiry about data to the adjacent node 130H, the inquiry destination node 130H has just been added and has no data. Therefore, the new responsible node 130G cannot acquire desired data.

  Here, as a result of making an inquiry to the adjacent node 130, if desired data cannot be acquired, a method of making an inquiry to the adjacent nodes 130 (130I, 130J,...) In order is conceivable. In this method, a plurality of hops (N hops) are required until actual data is found in proportion to the number N of nodes 130 added adjacently between the new assigned node 130G and the previous assigned node 130A. Therefore, the real-time property is impaired in proportion to the value of N.

  As described above, since the identifier of the node 130 is randomly given by a hash function or the like, the value of N is not constant, but the ratio of the number of added nodes 130 to the number of all nodes 130 before the addition is used. Increase proportionally.

  As a technique that does not impair the real-time property, there is a method that inquires all other nodes 130 when the new node 130G that has received the data transmission / storage request does not have the desired data. Although this method does not require a plurality of hops for searching for data, an inquiry is made to the entire system, so the processing load increases in proportion to the number of all nodes 130 and the scalability is significantly impaired.

  The present invention was created in view of the circumstances described above, and even when a plurality of servers are added, distributed data management capable of transferring data in real time and without losing scalability. It is an object to provide a system, a distributed data management method, and a distributed data management program.

  In order to solve the above problems, a distributed data management system of the present invention is a distributed data management system in which a plurality of servers distribute and manage data, and the server stores data associated with an identifier of the data. A first storage unit, a second storage unit storing an address table in which identifiers and addresses of the plurality of servers are associated, and a new generation of the address table when the plurality of servers are increased or decreased Including the address table generation unit to be stored in the second storage unit together with the generation time of the address table, the data identifier transmitted from the client, and the generation time of the data. A request signal acquisition unit for acquiring a request signal to be requested, an identifier of the data included in the acquired request signal, and a data stored in the second storage unit. Using the identifier of the plurality of servers in the address table, and a server identification unit that identifies the server in charge of the data, and the latest address table address stored in the second storage unit A request signal transfer unit that transfers the request signal to the specified server in charge, and refers to the first storage unit based on an identifier of the data included in the request signal that requests transmission of the data, Read the corresponding data and send it to the client, or refer to the first storage unit based on the identifier of the data included in the request signal requesting storage of the data, and A response unit that determines the success or failure of the storage and transmits a storage success / failure signal indicating the determination result to the client, and a data corresponding to the identifier of the data included in the request signal. Data is not stored in the first storage unit, among the address tables stored in the second storage unit, the generation time of the address table is greater than the generation time of the data included in the request signal Using the oldest and latest address table, the data and the identifier of the data are transferred between the previous server specification unit that specifies the server in front of the data and the specified server in front. And a data updating unit for storing the transferred data and an identifier of the data in the first storage unit.

  According to such a configuration, an address table having an address table generation time is generated for each increase / decrease of the server and stored in the second storage unit, and desired data is stored in the data table of the first storage unit of the server in charge. If not, the previous responsible server is identified based on the past address table selected based on the data generation time and the address table generation time, and the data is transferred from the identified previous responsible server to the responsible server. Even when a plurality of servers are added, data can be transferred in real time and without loss of scalability.

  In the distributed data management system described above, the first storage unit stores the data, an identifier of the data, and a generation time of the data in association with each other, and the data update unit transmits the transferred data And the time when the data update unit acquires the transferred data and the identifier of the data when the identifier of the data is stored in the first storage unit, or the transferred data and the identifier of the data Is stored in the first storage unit as the generation time of the data, and the response unit stores the data in the data or the storage success / failure signal. It is desirable to give the generation time and transmit to the client.

  According to such a configuration, when data is transferred from the previous server in charge to the server in charge, the time at which the server in charge acquires the data or the time at which the data is stored in the data table of the first storage unit of the server in charge is generated. Since the time is stored in the first storage unit and transmitted to the client, it is possible to cause the client to generate a request signal based on a new data generation time after data transfer.

  Further, the distributed data management method of the present invention is a distributed data management method in which a plurality of servers distribute and manage data, and the server stores the data and the identifier of the data in association with each other. A storage unit and a second storage unit that stores an address table in which identifiers and addresses of the plurality of servers are associated, and when the number of the servers increases or decreases, the address table is newly generated. It is stored in the second storage unit together with the generation time of the address table, and the one server includes a data identifier and a generation time of the data transmitted from the client, and transmits or stores the data Obtaining a request signal for requesting, an identifier of the data included in the acquired request signal, and the latest stored in the second storage unit Identifying the server in charge of the data based on the identifiers of the plurality of servers in the address table, and the one address of the latest address table stored in the second storage unit by the one server And transferring the request signal to the specified responsible server using the first storage unit based on the identifier of the data included in the request signal for requesting transmission of the data by the responsible server The corresponding data is read out and transmitted to the client, or the first storage unit is referenced based on the identifier of the data included in the request signal requesting storage of the data. Determining whether the data is stored successfully and transmitting a storage success / failure signal indicating a determination result to the client. When the data corresponding to the identifier of the data to be stored is not stored in the first storage unit, the responsible server generates the address table from among the address tables stored in the second storage unit. Identifying a previous server in charge of the data using an address table whose time is older than the generation time of the data included in the request signal, and the server in charge is connected to the previous server in charge. Transferring the data and the identifier of the data between, and the responsible server storing the transferred data and the identifier of the data in the first storage unit, To do.

  In the distributed data management method described above, the first storage unit stores the data and the identifier of the data and the generation time of the data in association with each other, and the responsible server stores the transferred data and When storing the identifier of the data in the first storage unit, the time when the server in charge acquired the transferred data and the identifier of the data, or the transferred data and the identifier of the data The time stored in the first storage unit by the server in charge is stored in the first storage unit as the generation time of the data, and the server in charge sets the generation time of the data in the data or the storage success / failure signal. It is desirable to give and send to the client.

  The present invention can also be embodied as a distributed data management program for causing a computer to function as each functional unit of the server of the distributed data management system.

  According to the present invention, even when a plurality of servers are added, data can be transferred in real time and without loss of scalability.

It is a figure which shows typically the distributed data management system which concerns on embodiment of this invention. It is a block diagram which shows the structure of the server of FIG. It is a figure which shows an example of a data table. It is a figure which shows an example of an address table. It is a figure for demonstrating the operation example of the distributed data management system which concerns on embodiment of this invention, Comprising: It is a figure which shows ID space. It is a flowchart for demonstrating the operation example of the distributed data management system which concerns on embodiment of this invention. It is a figure which shows ID space in a prior art. In the ID space in the prior art, it is a figure for demonstrating the case where a node (server) is added, Comprising: (a) is a figure which shows ID space where one node was added, (b) is a several node It is a figure which shows ID space to which is added.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings as appropriate. As shown in FIG. 1, a distributed data management system 1 according to an embodiment of the present invention is a system that obtains a request signal from a client 10 via a load balancer 20 and responds, and includes a plurality of servers 30 (30A , 30B, ..., 30N). The plurality of servers 30 manage data in a distributed manner, and are connected to be communicable with each other via a network (not shown).

<Client and load balancer>
The client 10 stores a data identifier and a generation time of the data in association with each other. The client 10 requests transmission of data stored in the distributed data management system 1 to the client 10 or storage in the distributed data management system 1 in accordance with the operation of the input device (keyboard, mouse, etc.) by the user. A signal is generated, and the generated request signal is transmitted to the distributed data management system 1. Here, the request signal includes an identifier (data identifier) of data to be transmitted or stored and a generation time (data generation time). The load balancer 20 randomly selects an arbitrary server 30 and transfers the request signal transmitted from the client 10 to the selected server 30. The client 10 first generates a request signal for requesting storage of data that does not include the data generation time, transmits the request signal to the distributed data management system 1, and is added to the response signal from the server 30 as described later. The transmitted data generation time is acquired, the acquired data generation time is stored, and thereafter, the request signal can be generated using the data identifier and the data generation time as arguments.

<Server>
The plurality of servers 30 respectively manage data handled by the server 30 among data to be managed by the entire distributed data management system 1. The server 30 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read-Only Memory), an input / output circuit, and the like. As shown in FIG. A storage unit 31-1, a second storage unit 31-2, an address table generation unit 32, a request signal acquisition unit 33, a responsible server identification unit 34, a request signal transfer unit 35, a response unit 36, A front server identification unit 37, a data transfer unit 38, and a data update unit 39 are provided.

<First storage unit>
A data table 31a shown in FIG. 3 is stored in the first storage unit 31-1. The data table 31a is a table in which data handled by the server 30, an identifier of the data (data identifier), and a generation time (data generation time) of the data are associated with each other.

<Second storage unit>
The second storage unit 31-2 stores an address table 31b shown in FIG. The address table 31b is a table in which identifiers (server identifiers) of all servers 30 constituting the distributed data management system 1 are associated with addresses (such as IP addresses) of the servers 30, and the generation time of the address table 31b Stored for each (address table generation time).

<Address table generator>
The address table generation unit 32 newly generates an address table 31b when the number of servers 30 constituting the distributed data management system 1 increases or decreases, and stores it in the second storage unit 31-2 together with the address table generation time of the new address table 31b. Remember me. That is, the address table 31b stored in the second storage unit 31-2 is stored as a history without being deleted even if the server 30 increases or decreases and a new address table 31b is generated. Increase. In addition, when the number of servers 30 is increased or decreased, for example, a technique for assigning addresses to the added (added) servers 30 or sharing the newly added or decreased addresses with the servers 30 is known. This technique can be employed, and detailed description thereof is omitted here.

<Request signal acquisition unit>
The request signal acquisition unit 33 acquires the request signal transmitted from the client 10 via the load balancer 20 and outputs the request signal to the assigned server specifying unit 34.

<Responsible server identification part>
The assigned server specifying unit 34 acquires the request signal output from the request signal acquiring unit 33, the data identifier included in the acquired request signal, and the latest address table stored in the second storage unit 31-2. Based on the server identifier 31b, the server in charge of data is specified. Here, the assigned server specifying unit 34 can specify the assigned server using an existing technique such as consistent hashing. When the server 30 is identified as the responsible server, the responsible server identifying unit 34 outputs a request signal to the response unit 36. When the other server 30 is identified as the responsible server, the responsible server identifying unit 34 The server specifying unit 34 reads the address of the server in charge from the address table 31b stored in the second storage unit 31-2, and outputs the request signal and the read address to the request signal transfer unit 35.

<Request signal transfer unit>
The request signal transfer unit 35 acquires the request signal and the address output from the assigned server specifying unit 34 and transfers the request signal to the response unit 36 of the assigned server using the address.

<Response section>
When the own server 30 is the responsible server, the response unit 36 transfers the request signal output from the responsible server specifying unit 34 of the own server 30 or the request signal transfer unit 35 of the other server 30 that is a non-responsible server. Obtained request signal. When the request signal is a request for data transmission, the response unit 36 uses the data table 31a stored in the first storage unit 31-1 based on the data identifier included in the acquired request signal. The data corresponding to the data identifier is read out and transmitted to the client 10. When the request signal is a request for data storage, the response unit 36 stores the data table stored in the first storage unit 31-1 based on the data identifier included in the acquired request signal. Referring to 31a, it is determined whether or not the data corresponding to the data identifier is stored (stored) in the data table 31a, and a storage success / failure signal indicating the determination result is generated and transmitted to the client 10.

  In some cases, the server 30 that is the server in charge has just been added to the distributed data management system 1 and desired data is not stored in the data table 31a. In this case, the response unit 36 outputs a request signal to the previous server identification unit 37.

<Previous server specific department>
The previous server identification unit 37 acquires the request signal output from the response unit 36, and the address table 31b stored in the second storage unit 31-2 based on the data generation time included in the acquired request signal. , The address table 31b that is older than the data generation time and has the latest address table generation time is specified. The previous server specifying unit 37 specifies the server in front of the data (the previous server) based on the data identifier included in the acquired request signal and the server identifier of the specified address table 31b. Here, similarly to the assigned server specifying unit 34, the previous assigned server specifying unit 37 can specify the previous assigned server using a technique such as existing consistent hashing. Since the address table 31b that is older than the data generation time and has the latest address table generation time corresponds to the address table before the update of the node in charge of the desired data, the address table 31b can be used before updating the node in charge ( When the responsible node has not been updated, the responsible node (at the first storage of data in the distributed data management system 1), that is, the previous responsible node can be identified. The previous server identification unit 37 reads the address of the own server 30 and the address of the previous server from the address table 31b stored in the second storage unit 31-2, and sends the request signal and the read address to the data transfer unit. 38.

<Data transfer unit>
The data transfer unit 38 whose own server 30 is a responsible server and does not store data acquires the request signal and address output from the previous responsible server specifying unit 37, and based on the acquired request signal and address, Data is transferred to and from the data transfer unit 38 of the previous server, and the data and the data identifier of the data are acquired from the previous server and output to the data update unit 39.

<Data update unit>
The data update unit 39 acquires the data and data identifier output from the data transfer unit 38, and stores the acquired data in the data table 31a of the first storage unit 31-1.

≪Data transfer≫
Here, when the desired data is not stored in the data table 31a of the first storage unit 31-1 of the own server 30, the data transfer unit 38, the data update unit 39, and the response of the own server 30 and the previous server The function regarding the data transfer of the part 36 is demonstrated in detail.

  The data transfer unit 38 which is the current server 30 (the new server) and does not store the data sends the request signal and the data transfer request signal including the address of the server 30 to the address of the previous server. And transmitted to the data transfer unit 38 of the previous server.

  The data transfer unit 38 of the previous server in charge acquires the data transfer request signal transmitted by the data transfer unit 38 of the new server in charge, and the data identifier of the request signal included in the acquired data transfer request signal is the data update unit 39. Output to.

  The data update unit 39 of the previous server in charge acquires the data identifier output from the data transfer unit 38, and refers to the data table 31a stored in the first storage unit 31-1 based on the acquired data identifier. The data corresponding to the data identifier is read, and the data identifier and the read data are output to the data transfer unit 38. Here, the data update unit 39 deletes the read data and the data identifier of the data (and the data generation time of the data) from the data table 31a.

  The data transfer unit 38 of the previous server in charge acquires the data and data identifier output from the data update unit 39, and uses the data and data identifier of the new server in charge of the acquired data and data identifier. 38.

  The data transfer unit 38 of the new server in charge acquires the data and the data identifier transmitted by the data transfer unit 38 of the previous server in charge, and outputs the acquired data and data identifier to the data update unit 39.

  The data update unit 39 of the new responsible server acquires the data and data identifier output from the data transfer unit 38, and stores the acquired data and data identifier in the data table 31a of the first storage unit 31-1. The update notification signal is output to the response unit 36. Here, the data update unit 39 of the new responsible server is transferred from the previous responsible server when storing the data and data identifier transferred from the previous responsible server in the data table 31a of the first storage unit 31-1. The time when the data update unit 39 acquired the data and the data identifier received, or the time when the data update unit 39 stores the data and data identifier transferred from the previous server in the data table 31a. The data generation time is stored in the data table 31a.

  The response unit 36 of the new responsible server acquires the update notification signal output from the data update unit 39, and transmits data or a storage success / failure signal to the client 10 in accordance with the acquired update notification signal. At this time, the response unit 36 of the new responsible server reads out the data generation time stored in the data table 31a of the first storage unit 31-1 based on the data identification signal included in the request signal, and the data or storage success / failure signal The data generation time can be given to the client 10 and transmitted to the client 10.

  When the client 10 acquires the data or the storage success / failure signal transmitted by the response unit 36 and the data generation time, the client 10 updates the data generation time stored in association with the data identifier based on the acquired data generation time.

  If the data update unit 39 of the previous server in charge cannot read out the desired data (when there is no desired data in the data table 31a of the previous server in charge), the data update unit 39 of the previous server in charge A signal indicating that there is no data is output to the data transfer unit 38, and the data transfer unit 38 of the previous server in charge transmits a signal indicating that there is no data to the data transfer unit 38 of the new server. The data transfer unit 38 of the new responsible server acquires a signal indicating that there is no data, and outputs the acquired signal to the response unit 36 via the data update unit 39 (the response unit without using the data update unit 39). It is also possible to output directly to 36). The response unit 36 of the new responsible server generates a storage success / failure signal indicating that there is no data, and transmits it to the client 10.

<Operation example>
Subsequently, an operation example of the distributed data management system 1 according to the embodiment of the present invention will be described with reference to FIGS. 5 and 6 (see FIGS. 1 to 4 as appropriate). In such an operation example, as shown in FIG. 5, new servers 30G, 30H, 30I, and 30J are added between the servers 30F and 30A with respect to the distributed data management system 1 configured by the servers 30A to 30F. It is assumed that the server in charge of data is changed from the server 30A to the server 30H by adding the servers 30G to 30J.

  First, the client 10 transmits a request signal for certain data to the server 30F via the load balancer 20. When the request signal acquisition unit 33 of the server 30F acquires the request signal (step S1), the server identification unit 34 of the server 30F displays all the servers stored in the data identifier included in the request signal and the latest address table 31b. The server in charge of data is specified based on the 30 server identifiers.

  In this operation example, since the server in charge is not the server 30F but the server 30H (No in step S2), the request signal transfer unit 35 of the server 30F transmits the request signal to the response unit 36 of the server 30H that is the server in charge. (Step S3). If the server 30F is a responsible server (Yes in step S2), the responsible server specifying unit 34 of the server 30F outputs a request signal to the response unit 36 of the server 30F, and the flow proceeds to step S4. .

  After execution of step S3, the response unit 36 of the server 30H that is the server in charge acquires the request signal, and based on the data identifier included in the acquired request signal, the data table 31a of the first storage unit 31-1. Search data from. In the present operation example, since the desired data is not stored in the data table 31a of the server 30H (No in step S4), the previous server specifying unit 37 is older and newer than the data generation time included in the request signal. Based on the server identifier stored in the address table 31b having the address table generation time and the data identifier included in the request signal, the server in charge of data is specified (step S5). If desired data is stored in the data table 31a of the server 30H (Yes in step S4), the flow proceeds to step S7.

  After execution of step S5, the data transfer unit 38 and the data update unit 39 of the server 30H that is the server in charge, and the data transfer unit 38 and the data update unit 39 of the server 30A that is the previous server perform data transfer. The data updating unit 39 of the server 30H as the server in charge updates the data generation time (step S6). By step S6, the data and the data identifier are stored in the data table 31a of the first storage unit 31-1, and the time when the data and the data identifier are acquired or the stored time is stored as the data generation time.

  Subsequently, the response unit 36 of the server 30H, which is the server in charge, responds to the client (step S7). When the request signal is a request for data transmission, the response unit 36 transmits the data and the data generation time to the client 10, and when the request signal is a request for data storage, a response is made. The unit 36 transmits the storage success / failure signal and the data generation time to the client 10.

  The distributed data management system 1 according to the embodiment of the present invention generates an address table 31b having an address table generation time for each increase / decrease of the server 30 and stores the generated address table 31b in the second storage unit 31-2. If the desired data is not stored in the data table of the storage unit 31-1, the previous server is identified based on the past address table 31b selected based on the data generation time and the address table generation time. Since the data is transferred from the specified previous responsible server to the new responsible server, even if a plurality of servers 30 are added, the adjacent servers 30 are inquired about the presence of data in order, or all the servers 30 are inquired. There is no need to inquire about the presence or absence of data, and only the previous server can check the presence or absence of data in one hop. It is possible to transfer data without compromising.

  Further, the distributed data management system 1 according to the embodiment of the present invention is configured such that when the data is transferred from the previous server in charge to the new server, the time when the new server acquires data or the first storage unit of the new server Since the time when the data is stored in the data table 31a of 31-1 is stored in the data table 31a as the data generation time and transmitted to the client 10, a request is made to the client 10 based on the new data generation time after the data transfer. A signal can be generated.

  As mentioned above, although embodiment of this invention was described with reference to drawings, this invention is not limited to the said embodiment, In the range which does not deviate from the summary of this invention, it can change suitably. For example, the present invention can be embodied as a distributed data management program for causing a computer to function as the server 30. Further, the server 30 and the data identifier determination method are not limited to those using a hash function, and the address of the server 30 is not limited to an IP address. Each of the client 10 and the server 30 may be realized by a single computer or may be realized by a plurality of computers.

1 Distributed data management system 30 Server (node)
31-1 1st memory | storage part 31-2 2nd memory | storage part 32 Address table production | generation part 33 Request signal acquisition part 34 Responsible server specific | specification part 35 Request signal transfer part 36 Response part 37 Previous charge server specific part 38 Data transfer part 39 Data update unit

Claims (6)

A distributed data management system in which multiple servers distribute and manage data,
The server
A first storage unit that stores data and an identifier of the data in association with each other;
A second storage unit storing an address table in which identifiers and addresses of the plurality of servers are associated;
An address table generation unit that newly generates the address table and stores it in the second storage unit together with the generation time of the address table when the number of servers increases or decreases;
A request signal acquisition unit that receives a request signal for requesting transmission or storage of the data, including an identifier of the data transmitted from the client and a generation time of the data;
The server in charge of the data is identified based on the identifier of the data included in the acquired request signal and the identifiers of the plurality of servers in the latest address table stored in the second storage unit The server identification department in charge,
A request signal transfer unit that transfers the request signal to the specified server in charge using the address of the latest address table stored in the second storage unit;
Based on an identifier of the data included in the request signal for requesting transmission of the data, the first storage unit is referred to, the corresponding data is read and transmitted to the client, or the storage of the data is requested. A response that refers to the first storage unit based on the identifier of the data included in the request signal, determines whether the corresponding data is stored successfully, and transmits a storage success / failure signal indicating the determination result to the client And
When the data corresponding to the identifier of the data included in the request signal is not stored in the first storage unit, the generation time of the address table among the address tables stored in the second storage unit Using a latest address table that is older than the generation time of the data included in the request signal, and a previous responsible server identifying unit that identifies the responsible server before the data;
A data transfer unit that transfers the data and the identifier of the data to the identified previous server in charge;
A data updating unit for storing the transferred data and an identifier of the data in the first storage unit;
A distributed data management system comprising: The first storage unit stores the data, the identifier of the data, and the generation time of the data in association with each other.
The data update unit, when storing the transferred data and the identifier of the data in the first storage unit, the time when the data update unit acquired the transferred data and the identifier of the data, or , The time at which the data update unit stores the transferred data and the identifier of the data in the first storage unit is stored in the first storage unit as the generation time of the data,
The distributed data management system according to claim 1, wherein the response unit assigns a generation time of the data to the data or the storage success / failure signal and transmits the data to the client. A distributed data management method in which multiple servers distribute and manage data,
The server includes a first storage unit that stores data and an identifier of the data in association with each other; a second storage unit that stores an address table in which identifiers and addresses of the plurality of servers are associated; And when the number of servers increases or decreases, the address table is newly generated and stored together with the generation time of the address table in the second storage unit,
One of the servers obtains a request signal for transmitting or storing the data, including the identifier of the data and the generation time of the data, transmitted from the client;
One of the servers is based on the identifier of the data included in the acquired request signal and the identifiers of the plurality of servers in the latest address table stored in the second storage unit, Identifying the server responsible for the data;
The one server transfers the request signal to the identified responsible server using the address of the latest address table stored in the second storage unit;
The responsible server refers to the first storage unit based on the identifier of the data included in the request signal for requesting transmission of the data, reads the corresponding data and transmits the data to the client, or Based on the identifier of the data included in the request signal for requesting data storage, the first storage unit is referred to, and a storage success / failure signal indicating a determination result by determining success / failure of storage of the corresponding data is indicated. Sending to the client;
Including
When the data corresponding to the identifier of the data included in the request signal is not stored in the first storage unit, the server in charge is associated with the address table stored in the second storage unit. Identifying a server in front of the data by using an address table whose address table generation time is older and newer than the data generation time included in the request signal;
The responsible server transfers the data and the identifier of the data to and from the previous responsible server;
The responsible server storing the transferred data and the identifier of the data in the first storage unit;
A distributed data management method comprising: The first storage unit stores the data, the identifier of the data, and the generation time of the data in association with each other.
When the responsible server stores the transferred data and the identifier of the data in the first storage unit, the time when the transferred server acquired the transferred data and the identifier of the data, or the transfer The time when the server in charge stores the data and the identifier of the data in the first storage unit is stored in the first storage unit as the generation time of the data,
The distributed data management method according to claim 3, wherein the responsible server gives the generation time of the data to the data or the storage success / failure signal and transmits the data to the client. A distributed data management program used when a plurality of computers distribute and manage data,
A computer comprising: a first storage unit storing data and an identifier of the data in association with each other; and a second storage unit storing an address table in which identifiers and addresses of the plurality of servers are associated with each other. ,
An address table generator that newly generates the address table and stores the address table in the second storage unit together with the generation time of the address table when the plurality of servers increase or decrease;
A request signal acquisition unit for acquiring a request signal for requesting transmission or storage of the data, including a data identifier and a generation time of the data, transmitted from the client;
The server in charge of the data is identified based on the identifier of the data included in the acquired request signal and the identifiers of the plurality of servers in the latest address table stored in the second storage unit Responsible server identification department,
A request signal transfer unit that transfers the request signal to the specified server in charge using the address of the latest address table stored in the second storage unit;
Based on an identifier of the data included in the request signal for requesting transmission of the data, the first storage unit is referred to, the corresponding data is read and transmitted to the client, or the storage of the data is requested. A response that refers to the first storage unit based on the identifier of the data included in the request signal, determines whether the corresponding data is stored successfully, and transmits a storage success / failure signal indicating the determination result to the client Part,
When the data corresponding to the identifier of the data included in the request signal is not stored in the first storage unit, the generation time of the address table among the address tables stored in the second storage unit Using a latest address table that is older than the generation time of the data included in the request signal and using the latest address server specifying unit that specifies the server in front of the data,
A data transfer unit that transfers the data and an identifier of the data to the identified previous server in charge; and
A data update unit for storing the transferred data and the identifier of the data in the first storage unit;
Distributed data management program to function as The first storage unit stores the data, the identifier of the data, and the generation time of the data in association with each other.
The data update unit, when storing the transferred data and the identifier of the data in the first storage unit, the time when the data update unit acquired the transferred data and the identifier of the data, or , The time at which the data update unit stores the transferred data and the identifier of the data in the first storage unit is stored in the first storage unit as the generation time of the data,
The distributed data management program according to claim 5, wherein the response unit assigns a generation time of the data to the data or the storage success / failure signal and transmits the data to the client.

Images