JP5533453B2 - Tree network reconfiguration device and program, and tree network reconfiguration method - Google Patents

Description

The present invention relates to a tree-like network reconfiguration device, a program, and a tree-like network reconfiguration method . For example, in a system in which a plurality of distributed content distribution servers are logically arranged in a tree shape, nodes in the middle of the tree (However, a node means a content distribution server) It can be applied to a response when a failure occurs.

  There is a content distribution system in which a plurality of content distribution servers (hereinafter referred to as distribution servers) that distribute content (for example, video content) to user terminals are distributed. In such a content distribution system, a plurality of distribution servers described above are logically arranged in a tree shape, and each content is distributed by branching and distributing content from the root node to the end node of the tree. There is a system in which the same content is deployed on a server, and each distribution server distributes the content to user terminals accommodated by itself.

  Here, when a failure occurs in a certain intermediate node on the tree while delivering the content from the root node of the tree to the end node, the content cannot be deployed to a node ahead of the intermediate node. Therefore, it is necessary to reconfigure the tree so that the content can be deployed to all nodes without going through the intermediate node where the failure has occurred. Embodiment 4 of Patent Document 1 describes a tree reconstruction method. In this reconfiguration method, for each node downstream of the intermediate node in which a failure has occurred, another node that accommodates the node is determined. Other nodes to be accommodated are determined from the end nodes on the tree. When there are a plurality of downstream nodes connected to the failed intermediate node, other nodes accommodating the respective downstream nodes are determined, and the other nodes accommodated by the downstream nodes may be different.

JP 2005-25622 A

  However, if a failure occurs in an intermediate node during content delivery, the content may not have been delivered to the end node that accommodated the downstream node of the failed node. Even if the configured tree is applied, there is a possibility that the content cannot be delivered to a node downstream of the failed node.

  Even if the content is delivered to the end node that has accommodated the downstream node of the failed node, the node that was the end node before reconfiguration performs the downstream delivery as an intermediate node. Many nodes have a larger number of necessary hops than before reconstruction, and it takes a considerable amount of time to deploy content to all nodes on the tree. In particular, when the delivery capacity of the replaced node is low, the problem of the deployment time becomes more serious.

Therefore, there is a demand for a tree-like network reconfiguration apparatus and program , and a tree-like network reconfiguration method that can quickly distribute delivery data to all nodes on the tree .

According to the first aspect of the present invention, a tree configuration is reconfigured from a network in which a plurality of servers arranged in a distributed manner are logically connected in a tree shape to a configuration in which one server is excluded based on the appearance of a predetermined event. In the tree-like network reconfiguration apparatus, the occurrence of the predetermined event means that the first data is distributed from the server at the root node of the tree-like network to all the servers in the tree-like network along the tree structure. and it is to failure of one server occurring in time are, (1) and candidate selection means for selecting a candidate server replaces a candidate to replace the server to be excluded, (2) replacing each candidate server And candidate specification information fetching means for fetching information on the specifications of the server, and (3) specification information of the replacement candidate server And (4) a failed server , a replacement server determining means for determining a server that replaces the server to be excluded, on the condition that the replacement candidate server can secure the performance capability of the server to be excluded. A delivery node confirmation means for determining whether the first data has already been delivered and deployed to the downstream server; and (5) the first data is delivered and deployed to the downstream server of the failed server. If not, the reconstructed tree replaced by the server determined by the replacement server determining means is determined as a tree applied to the delivery of the first data to the undelivered server, and a failure occurs. If the first data has already been delivered and deployed to a server downstream of the server that has been reconfigured, the tree before reconfiguration is transferred to the server that has not been delivered. In the distribution of new second data that will occur in the future, the reconstructed tree replaced by the server determined by the replacement server determination means will be applied to the tree to be applied. And an application tree deciding means for deciding.

The second of the present invention, a network in which a plurality of servers that are distributed are connected logically to the tree-like, based on the occurrence of a predetermined event, to reconstruct the tree structure to the excluded configure one server A tree-like network reconfiguration program for operating a computer mounted on a tree-like network reconfiguration device, wherein the occurrence of the predetermined event is a tree-like configuration along a tree configuration from a server at the root node of the tree-like network. A failure occurred in one of the servers while delivering the first data to all the servers in the network, and the computer was replaced with (1) a candidate to replace the server to be excluded. Candidate selection means for selecting a replacement candidate server, and (2) each of the replacement candidate servers (3) Excluded on condition that the replacement candidate server can secure the performance capability of the server to be excluded, as one condition, in view of the specification information of the replacement candidate server. Replacement server determining means for determining a server to replace the target server ; and (4) a delivered node confirmation means for determining whether the first data has already been delivered and deployed to a server downstream of the failed server. (5) If the first data is not delivered or deployed to a server downstream of the failed server, the reconstructed tree replaced by the server determined by the replacement server determination unit is not stored. A tree to be applied to the delivery of the first data to the delivery server is determined, and the server downstream of the failed server Is already distributed and deployed, the tree before reconfiguration is determined as the tree to be applied to the delivery of the first data to the undelivered server, and a new In the delivery of the second data, the replacement server determination unit functions as an application tree determination unit that determines a tree to which the reconstructed tree replaced by the server determined is applied.
According to a third aspect of the present invention, a tree configuration is reconfigured from a network in which a plurality of servers arranged in a distributed manner are logically connected in a tree shape to a configuration in which one server is excluded based on the appearance of a predetermined event. In the tree-shaped network reconfiguration method in the tree-shaped network reconfiguration device, the occurrence of the predetermined event is a first event from the server at the root node of the tree-shaped network to all servers in the tree-shaped network along the tree configuration. A failure has occurred in one of the servers during delivery to deploy data, and the tree-shaped network reconfiguring device determines candidate selection means, candidate specification information fetching means, and replacement server determination. Means, a delivered node confirmation means, and an application tree determination means, (1) a candidate that replaces the server to be excluded A candidate selection step executed by the candidate selection means for selecting a replacement candidate server; (2) candidate specification information executed by the candidate specification information acquisition means for acquiring the specification information of the server for each of the replacement candidate servers; determining a capture step, and (3) replacing viewed from the specification information of the candidate servers, that the the replacement candidate server can ensure more performance capability of the server to be excluded as a condition, replace the server exclusion target server A replacement server determining step executed by the replacement server determining means; and (4) the delivered node confirmation means for determining whether the first data has already been delivered and deployed to a server downstream of the failed server. A delivery node confirmation step to be executed, and (5) a server downstream of the failed server If the first data is not distributed or deployed to the server, the reconstructed tree replaced by the server determined by the replacement server determining means is used as the first data to the undelivered server. When the tree to be applied to the delivery is determined and the first data is already delivered and deployed to the server downstream of the failed server, the tree before reconfiguration is sent to the undelivered server. The tree to be applied to the delivery of the first data is determined, and the new second data to be generated in the future applies the reconstructed tree replaced by the server determined by the replacement server determining means. And an application tree determining step executed by the application tree determining means for determining a tree.

According to the present invention, it is possible to realize a tree-like network reconfiguration device and program , and a tree-like network reconfiguration method that can quickly distribute delivery data to all nodes on the tree.

It is a block diagram which shows the whole structure of the content delivery system of 1st Embodiment. It is explanatory drawing which shows the logical connection relation of the some delivery server in 1st Embodiment. It is a functional block diagram which shows the internal structure of the delivery server in 1st Embodiment. It is a functional block diagram which shows the internal structure of the content management server in 1st Embodiment. It is explanatory drawing which shows the structure of the information for evaluating the replacement node candidate in the delivery information database in 1st Embodiment. It is a flowchart which shows the delivery instruction | indication operation | movement which the content management server in 1st Embodiment performs. FIG. 9 is a sequence diagram illustrating a processing flow when content delivery is normally performed in response to a delivery instruction from an administrator in the first embodiment. FIG. 9 is a sequence diagram illustrating a processing flow when it is necessary to replace a delivery destination of content delivery that is to be started in response to a delivery instruction from an administrator for a faulty delivery server in the first embodiment. . It is a flowchart which shows the reconstruction process of the tree which the content management server in 1st Embodiment performs. It is a sequence diagram which shows the process which measures the distance of the replacement candidate of a failure node and the upstream node directly connected to a failure node in 1st Embodiment. It is explanatory drawing which shows the structure after the reconstruction of the tree in 1st Embodiment with the structure after the reconstruction of the tree in the past. It is a flowchart which shows the reconstruction process of the tree which the content management server in 2nd Embodiment performs. It is explanatory drawing which shows the structure of the information for evaluating the replacement node candidate in the delivery information database in 2nd Embodiment.

(A) First Embodiment Hereinafter, a first embodiment of a tree-like network reconfiguration device and program and a tree-like network reconfiguration method according to the present invention will be described with reference to the drawings. The tree-like network system of the first embodiment is a content distribution system.

(A-1) Configuration of the First Embodiment FIG. 1 is a block diagram showing the overall configuration of the content distribution system 1 of the first embodiment.

  In FIG. 1, a content distribution system 1 according to the first embodiment includes a plurality of distribution servers 2-0 to 2-N, a content management server 3, and a distribution information database (distribution information DB) 4 via a network 5. Are connected to communicate.

  The network 5 may be a single network (for example, an IP network), or may be a combination of a plurality of networks. Furthermore, the network 5 may have any connection form such as a star shape, a ring shape, or a mesh shape.

  In the content distribution system 1 of the first embodiment, the logical connection relationship between the plurality of distribution servers 2-0 to 2-N is determined regardless of the configuration of the network 5. That is, in the case of the content distribution system 1 of the first embodiment, a plurality of distribution servers (referred to as nodes as appropriate) 2-0 to 2-N are logically connected in a tree form as shown in FIG. (FIG. 2 shows a case where there are 12 distribution servers). However, the content management server 3 and the tree database 4 are provided apart from the tree network. Here, the tree-like network is gradually branched from one root node 2-0, and the content distribution route to the end nodes 2-5 and 2-7 to 2-11 is a single route. Network.

  The root node 2-0 is a node that only transmits content to predetermined next (downstream) nodes (2-1 and 2-2 in the example of FIG. 2). Each of the end nodes 2-5 and 2-7 to 2-11 is a node that only receives the content transmitted from the previous (upstream) node determined in advance. The intermediate nodes 2-1 to 2-4 and 2-6 receive the content transmitted from the previous (upstream) node determined in advance and transmit the content to the next (downstream) node determined in advance. It is a node.

  In this specification, “delivery” refers to transmitting content from a distribution server to another distribution server in order to deploy the content to each distribution server, while “distribution” refers to the distribution server being deployed to itself. Suppose that the transmitted content is transmitted to the user terminal.

  Each of the distribution servers 2-0 to 2-N is a server that distributes the distributed content to itself and distributes the content to the user terminals assigned to the distribution server. Each of the distribution servers 2-0 to 2-N is distributed in each region in consideration of the position and density of the user terminal.

  FIG. 3 is a functional block diagram showing the internal configuration of the distribution server 2-n (n is 0 to N), and particularly shows the internal configuration of the intermediate node.

  The distribution server 2-n includes a communication unit 10, a delivery content receiving unit 11, a content storage unit 12, a delivery content transmission unit 13, a delivery control unit 14, a content distribution unit 15, and a distribution control unit 16.

  The communication unit 10 performs communication (transmission / reception) with the network 5 side. If the network 5 is, for example, an IP network, the communication unit 10 is a TCP / IP transmission / reception unit. In FIG. 3, the communication unit that distributes the content between the distribution servers and the communication unit that distributes the content to the user terminal are shown in common. However, these communication units may be provided separately. .

  The delivery content receiving unit 11 receives the content given via the communication unit 10 delivered (transmitted) from the upstream node, and stores it in the content storage unit 12.

  The content storage unit 12 stores the delivered content. Here, a storage unit for storing content for distribution to the user terminal and a storage unit for temporarily storing the content delivered from the upstream node until it is transmitted to the downstream node may be provided separately. In the case of such two storage unit configurations, transfer from the latter storage unit to the former storage unit is also executed as appropriate.

  The delivery content transmission unit 13 transmits (delivery) the content stored in the content storage unit 12 to the downstream node via the communication unit 10.

  The delivery control unit 14 executes various control processes related to content delivery. For example, when the delivery content receiving unit 11 finishes receiving the content, the delivery control unit 14 notifies the delivery source delivery server via the communication unit 10 that the reception has been completed normally, and the reception is in progress. If it fails, a reception error is notified to the delivery server of the delivery source via the communication unit 10. Also, for example, the delivery control unit 14 activates content delivery to the downstream node by the delivery content transmission unit 13 when a delivery instruction is given from the content management server 3 via the communication unit 10. Further, for example, when the delivery control unit 14 receives a content reception end notification or a reception error from the downstream node via the communication unit 10, the delivery control unit 14 delivers the content according to the delivery instruction to the content management server 3 via the communication unit 10. Is returned and an error occurrence is returned. Further, for example, when the delivery control unit 14 recognizes the occurrence of a failure in the downstream node while trying to start delivery of content to the downstream node or during content delivery, a failure or error has occurred in the content management server 3 via the communication unit 10. It is a notification of occurrence.

  First, information on downstream nodes that deliver content may be included in the delivery instruction. Secondly, the downstream node information is not included in the delivery instruction, and the delivery control unit 14 may hold the downstream node information therein and take out the information at the time of the delivery instruction. Third, if the downstream node information is included in the delivery instruction, the downstream node information stored therein is stored in the delivery instruction if the downstream node information is not included in the delivery instruction. It may be used. After the tree reconfiguration described later, in the first and third methods, it is possible to perform delivery to the downstream node after reconfiguration by including information on the downstream node in the delivery instruction. In the case of the second method, prior to the delivery instruction, information on the downstream node after reconfiguration is given and held by the delivery control unit 14, and then the delivery instruction is given.

  The content distribution unit 15 distributes the content stored in the content storage unit 12 to the user terminal via the communication unit 10. The content distribution unit 15 may be shared with the delivery content transmission unit 13.

  The distribution control unit 16 controls content distribution to the user terminal. Here, the content distribution to the user terminals can be performed simultaneously or sequentially to all user terminals managed by the distribution server 2-n at a predetermined time or according to an instruction from a higher-level device (not shown). There may be individual distribution to the user terminal that requested the distribution.

  FIG. 3 shows the internal configuration of the distribution server 2-n serving as an intermediate node as described above. The distribution server 2-0 serving as the root node may not be configured to accept delivery content from the network. In this case, the content is stored in the content storage unit 12 by an administrator's file setting operation or the like. Also, the distribution server 2-0 serving as the root node has the same configuration as that in FIG. 3, and takes in content transmitted from an external device different from the distribution server from the network 5, and stores the content in the content storage unit 12. You may make it store. As an end node, there may be an end node that cannot be an intermediate node in addition to an end node that can be an intermediate node. An end node that can be an intermediate node has the configuration shown in FIG. An end node that cannot be an intermediate node may be one that does not have a configuration for delivering content to downstream nodes.

  The content management server 3 is a server that manages and controls the distribution of content in order to deploy the content to all the distribution servers 2-0 to 2-N (excluding those in which a failure has occurred). The content management server 3 may also manage and control the distribution of content to the user terminal.

  FIG. 4 is a functional block diagram showing the internal configuration of the content management server 3. Even if some functions of the content management server 3 are realized by the CPU and a program executed by the CPU, the internal configuration of the content management server 3 can be functionally shown in FIG.

  In FIG. 4, the content management server 3 includes a communication unit 20, an administrator interface unit 21, a tree basic configuration storage unit 22, a failure node storage unit 23, a tree reconfiguration unit 24, and a delivery instruction control unit 25.

  The communication unit 20 performs communication (transmission / reception) with the network 5 side. If the network 5 is, for example, an IP network, the communication unit 20 is a TCP / IP transmission / reception unit.

  The administrator interface unit 21 includes a keyboard, a mouse, a display, and the like. The administrator interface unit 21 captures an operation input (for example, a delivery instruction) by the administrator, and informs the administrator of a processing result in the content management server 3 and a guidance message. It is something to do. In FIG. 4, the administrator directly operates the content management server 3. However, the administrator operates an administrator terminal such as a personal computer, and communication between the administrator terminal and the content management server 3 The administrator may give an instruction to the content management server 3.

  The tree basic configuration storage unit 22 stores basic tree configuration information as shown in FIG. 2 of all the distribution servers 2-0 to 2-N. It is read by the instruction control unit 25. Here, the basic tree configuration refers to a tree configuration used for content delivery when no failure occurs in any of the distribution servers. As will be described later, when a failure occurs in any of the distribution servers, tree reconfiguration is executed. In this case, when the failed node is recovered, the basic tree configuration stored in the tree basic configuration storage unit 22 may be reapplied, and the recovered node is added to the reconfigured tree. It may be added according to a predetermined rule and overwritten in the tree basic configuration storage unit 22 as a new basic tree configuration. In the following description, it is assumed that the basic tree configuration stored in the tree basic configuration storage unit 22 is applied again when the failed node is recovered.

  The failure node storage unit 23 stores information on a distribution server in which a failure has occurred (server identification information, failure case time, etc.). The storage information is accessed (written, written) by the delivery instruction control unit 25. Read, erase, etc.). When information on a distribution server in which a failure has occurred is stored in the failure node storage unit 23, information on a distribution server that replaces the distribution server in which the failure has occurred, which is determined in a tree reconfiguration process described later, is also stored. It is made so that.

  The failure of the distribution server can be detected when the upstream distribution server attempts to deliver the content. The content management server 3 can also periodically check the status of each distribution server and detect the occurrence of a failure.

  When a failure occurs in any of the distribution servers, that is, when there is a distribution server in which content is not distributed in the basic tree configuration, the tree reconfiguration unit 24 controls the tree under the control of the distribution instruction control unit 25. Is reconstructed. The tree reconfiguration process is executed when it is detected that any of the distribution servers has failed. The reconfiguration method by the tree reconfiguration unit 24 will be clarified in the operation description to be described later.

  The delivery instruction control unit 25 receives a delivery instruction from the administrator, and confirms the delivery status between the delivery server of the delivery source and the delivery destination related to the delivery instruction, the presence / absence of a failure of the delivery server of the delivery source and the delivery destination, and the like. The content delivery instruction is given to the distribution server on the basic configuration tree or the distribution server on the reconfigured tree via the communication unit 20. The delivery instruction control unit 25 may control delivery of a plurality of types of content in parallel. The delivery instruction control unit 25 manages the time when each content is deployed to the root node 2-0.

  Although the detailed configuration is not shown, the delivery information database 4 stores at least two types of information that the content management server 3 refers to during processing. The first information is delivery status information indicating whether the content is undelivered or delivered for each delivery server. When there are a plurality of types of content related to delivery, information on the delivery status is stored for each content. The second information includes a replacement candidate that can be a node that replaces the intermediate node and delivers content to the downstream node of the intermediate node when a failure occurs in the intermediate node in the basic tree configuration, and the candidate Is information for evaluating (candidate evaluation information).

  Although FIG. 1 shows that the delivery information database 4 is also an element of the network 5, the delivery information database 4 may be connected to the content management server 3 by a dedicated line. It may be directly connected to the server 3 with a cable or the like. In such a case, the content management server 3 includes a component for accessing the delivery information database 4 in addition to the communication unit 20.

  FIG. 5 is an explanatory diagram showing a configuration example of the above-described candidate evaluation information (second information) in the delivery information database 4, and shows information of some distribution servers that can be replacement candidates.

  In FIG. 5, the candidate evaluation information has a row (record) for each piece of replacement candidate identification information, and each row includes information on the specification (processing capability provided as a server) of the replacement candidate server and a distance. With information.

  The specification information includes, for example, (a) the number of downstream nodes that can be accommodated (the number of paths that can be connected to the downstream nodes), (b) the communication bandwidth in each path to the downstream nodes, and (c) the number of contents that can be distributed (current The number of distributions that can be increased from the state) is described. The distance information is initially blank, and is measured and written in as described later when a failure node occurs and a node in that row becomes a replacement candidate.

  In the example of FIG. 5, the distribution server 2-7 is a server that can be a replacement candidate, and the distribution server 2-7 can distribute the content to a maximum of five downstream nodes, and the five downstream nodes 10 [GB / s], 10 [GB / s], 10 [GB / s], 10 [GB / s], and 10 [GB / s] can be secured from the current path, respectively. Distribution to many user terminals is possible. “8.1” is described as the distance, which is written after the distance measurement processing is performed as described later.

  In the case of the first embodiment, the node replaced with the failed node takes over not only the content delivery to the downstream node but also the content distribution to the user terminal assigned to the failed node. However, the node replaced with the failure node may take over only the content delivery to the downstream node, and the content delivery to the user terminal assigned to the failure node may be determined according to another rule. For example, a node that is close to the faulty node (or a node that has a small number of hops to the faulty node in the basic configuration) and that has a margin for content distribution is distributed to the user terminals assigned to the faulty node You may make it take over. In such a case, the number of content deliverables in (c) is excluded from the candidate evaluation information.

(A-2) Operation of First Embodiment Next, the operation of the content distribution system 1 according to the first embodiment will be described. In the following, operations related to tree reconstruction will be mainly described.

  The administrator instructs the content management server 3 to deliver content by specifying two directly connected distribution servers (or paths connecting these distribution servers) in the basic tree configuration. At this time, the content management server 3 executes the operation for determining the application tree shown in FIG.

  If there are a plurality of types of content related to delivery, the administrator designates the type of content and instructs delivery. For example, the administrator displays a basic tree configuration on the display, and designates (for example, clicks) any one of the displayed tree configurations on the display screen, whereby the path (2 Specify one distribution server). When displaying the basic tree structure on the display, the delivery status stored in the delivery information database 4 is taken out for all delivery servers, and the delivery server to which the content has not been delivered and the delivery server to which the delivery has already been made. It is also possible to display by distinguishing (for example, changing the display color) so that a content delivery instruction can be issued efficiently.

  When the processing shown in FIG. 6 is started, the content management server 3 first extracts the delivery status of the content related to the delivery instruction to each delivery server from the delivery information database 4 and obtains failure information from the built-in failure node storage unit 23. Remove (step 100).

  Then, the content management server 3 determines whether or not a failure node exists (step 101). If there is no failure node, the content management server 3 decides to deliver the content according to the basic tree configuration, and applies the delivery instruction instructed by the administrator to the basic tree configuration, and the delivery source Distribution server is determined (step 102), and a delivery instruction is given to the delivery server of the delivery source (step 103).

  If there is a failure node, the content management server 3 compares the failure detection time with the time when the content is deployed to the root node 2-0, thereby starting a series of content delivery operations related to the current delivery instruction. After that, it is determined whether or not a failure has occurred (step 104).

  If a failure occurs before a series of content delivery operations related to the current delivery instruction is started, the content management server 3 decides to deliver the content according to the reconfigured tree structure, and the administrator The delivery instruction instructed from is applied to the reconfigured tree structure to determine the delivery server of the delivery source (step 105), and the delivery instruction is given to the delivery server of the delivery source (step 103). That is, if a failure node has occurred before the content is deployed to the root node 2-0, the reconfigured tree configuration is applied. When step 105 is executed, a tree reconfiguration process may be performed in parallel.

  On the other hand, by comparing the failure detection time with the time when the content is deployed to the root node 2-0, it is determined that a failure has occurred in any of the nodes during delivery using the basic tree structure. If so, the content management server 3 determines whether or not the content delivery to the downstream node of the failed node has been completed (step 106). When the content delivery to the downstream node of the faulty node has been completed, the content management server 3 decides to deliver the content according to the basic tree structure, and sends the delivery instruction instructed by the administrator to the basic instruction. A delivery server as a delivery source is determined by applying to the tree configuration (step 102), and a delivery instruction is given to the delivery server as the delivery source (step 103). On the other hand, if the content delivery to the downstream node of the failed node has not been made, the content management server 3 decides to deliver the content according to the reconfigured tree structure, and the delivery instruction instructed by the administrator Is applied to the reconfigured tree structure to determine a delivery server as a delivery source (step 105), and a delivery instruction is given to the delivery server as a delivery source (step 103).

  The content management server 3 executes the processing shown in FIG. 6 and always applies the reconstructed tree configuration when a failure node occurs, and considers the failure node in view of the delivery status of the content to the entire tree. It is determined whether the basic tree structure or the reconstructed tree structure is applied even after the occurrence of the above. Therefore, it is possible to prevent the occurrence of a node to which no content is delivered, as compared with a case where it is uniformly decided to apply the reconfigured tree structure.

  In the above, the case where the administrator gives a delivery instruction specifying the delivery server of the delivery source and the delivery destination to the content management server 3 has been described. However, the delivery instruction to all nodes is given to the content management server 3. Anyway. The content management server 3 to which a delivery instruction to all nodes is given takes out one path at a time from the path on the root node side in the basic tree configuration, and relates to the upstream and downstream distribution servers related to the extracted path. A delivery instruction is formed, and the processing when the delivery instruction shown in FIG. 6 is input may be executed.

  FIG. 7 is a sequence diagram showing a processing flow when content delivery is normally performed in response to a delivery instruction from the administrator.

  When the administrator 6 issues a content delivery instruction to the content management server 3 (step 200), the content management server 3 retrieves the delivery status from the delivery information database 4 (step 201; see step 100 in FIG. 6). Through the processing of FIG. 6 described above, the delivery server (upstream) delivery server 2-U and the delivery destination (downstream) delivery server 2-L are determined, and the downstream delivery server 2-U is downstream. Instruct the delivery of the content to 2-L (step 202).

  In accordance with this instruction, the upstream delivery server 2-U delivers the content to the downstream delivery server 2-L (step 203). When the downstream distribution server 2-L receives the content normally, the downstream distribution server 2-L notifies the upstream (delivery source) distribution server 2-U that the reception has ended normally (step 204). At this time, the distribution server 2-U returns to the content management server 3 that the delivery according to the delivery instruction has been executed (step 205).

  Thereby, the content management server 3 gives information indicating that the content has been delivered to the delivery server 2-L to the delivery information database 4, and updates the delivery status of the delivery information database 4 for the delivery server 2-L to delivery completed. (Step 206). The content management server 3 receives the storage response from the delivery information database 4 and notifies the manager 6 that the delivery of the given delivery instruction has been completed normally (step 207).

  By repeatedly performing the processing shown in FIG. 7 by changing the combination of the delivery server and the delivery server, contents can be deployed to all nodes (distribution servers) in the tree shape.

  FIG. 8 is a sequence diagram showing a processing flow when it is necessary to replace a delivery destination of content delivery to be started in response to a delivery instruction from an administrator for a faulty delivery server. The same processes as those in FIG.

  In response to the delivery instruction from the administrator, the content management server 3 instructs the upstream delivery server 2-U to deliver the content to the downstream delivery server 2-L, and the upstream delivery server 2-U The processing flow until delivery of content to the distribution server 2-L is the same as in FIG. 7 (steps 200 to 203).

  If a failure has occurred in the downstream delivery server 2-L (step 250), the upstream delivery server 2-U cannot deliver the content even if it tries to deliver the content to the downstream delivery server 2-L. The upstream distribution server 2-U cannot establish a session with the downstream distribution server 2-L, or the downstream distribution server 2-L does not receive a notification that the reception has been normally completed. It detects that a failure has occurred in the downstream delivery server 2-L, and returns a delivery error to the content management server 3 (step 251).

  At this time, the content management server 3 stores therein failure information indicating that a failure has occurred in the downstream delivery server 2-L (step 252), and executes a tree reconfiguration process as described later. (Step 253).

  When the distribution server 2-Lnew to be replaced with the failed node 2-L is determined by the tree reconfiguration process, the content management server 3 delivers the content to the replacement distribution server 2-Lnew to the upstream distribution server 2-U. Instruct (step 254). The processing of the upstream distribution server 2-U and replacement distribution server 2-Lnew in response to such a delivery instruction is the same as in the case of FIG.

  That is, the upstream delivery server 2-U delivers the content to the replacement delivery server 2-Lnew in accordance with this delivery instruction (step 255), and when the replacement delivery server 2-Lnew has successfully received the content, The upstream delivery server 2-U is notified that the reception has been normally completed (step 256), and the upstream delivery server 2-U returns a notification that the delivery according to the delivery instruction has been executed to the content management server 3. (Step 257), the content management server 3 gives information indicating that the content has been delivered to the replacement delivery server 2-Lnew to the delivery information database 4, and the delivery status of the delivery server 2-Lnew in the delivery information database 4 is determined. The content management server 3 is updated to the delivery completed (step 258), and receives the storage response from the delivery information database 4. The administrator 6, delivery of a given delivery instruction is notified of the completion (step 259). However, in this end notification, since a failure has occurred in the delivery node 2-L, the administrator is notified that the content has been delivered to the delivery server 2-Lnew that replaces it.

  For example, in the case of a delivery instruction from the delivery server 2-0 to the delivery server 2-2 in FIG. Content cannot be delivered to the distribution servers 2-5, 2-6, and 2-10. For this reason, a distribution server that replaces the distribution server 2-2 is determined so that the content can be distributed to the distribution servers 2-5, 2-6, and 2-10 downstream of the distribution server 2-2. In the case of the first embodiment, the distribution server replaced with the failed distribution server also performs distribution to the user terminals assigned to the failed distribution server.

  Further, for example, if a failure has occurred in the delivery server 2-9 at the time of delivery from the delivery server 2-4 to the delivery server 2-9 in FIG. 2, the downstream of the delivery server 2-9 Has no distribution server. However, since it is necessary to perform distribution to the user terminal assigned to the faulty distribution server, even when a fault occurs in such an end node, the distribution of the faulty distribution server 2-9 is taken over. It is necessary to define a distribution server.

  Although FIG. 8 shows that the content is always delivered to the replacement delivery server 2-Lnew, the content management server 3 delivers the delivery status of the replacement delivery server 2-Lnew when the replacement delivery server 2-Lnew is determined. When the delivery to the replacement delivery server 2-Lnew has been completed, the notification is sent to the administrator immediately without executing the delivery of the content, and the delivery to the replacement delivery server 2-Lnew is performed. If not completed, delivery of content to the replacement delivery server 2-Lnew may be executed to notify the administrator of the end.

  FIG. 8 shows a case where a failure has occurred in the delivery server of the delivery destination related to the delivery instruction, but also when a failure has occurred in the delivery server of the delivery source (upstream) related to the delivery instruction, The content management server 3 may determine the distribution server 2-Unew to be replaced with the failed node 2-U and instruct the replacement distribution server 2-Unew to deliver the content to the distribution server 2-L. For example, even if the session establishment operation performed in order to give a delivery instruction cannot establish a session with the delivery server (upstream) delivery server 2-U, the content management server 3 causes the delivery server (upstream) delivery server to The occurrence of failure can be detected.

  FIG. 9 is a flowchart showing a tree reconfiguration process (reconfiguration program) executed by the content management server 3. For example, a tree reconstruction process is executed as the process of step 253 in FIG. 8 described above. When the reconfiguration process is activated, the situation is not limited to the situation shown in FIG. For example, when a failure of one of the nodes is detected in the periodic failure monitoring process, a tree reconfiguration process may be activated so as to be able to respond to subsequent content delivery instructions.

  When the content management server 3 starts the process shown in FIG. 9, first, a replacement candidate is selected (step 300). The method for selecting replacement candidates is not limited. For example, the minimum number of candidates to be selected may be determined. Further, for example, only end nodes in a basic tree configuration may be set as replacement candidates. Further, for example, in a basic tree configuration, a node that is not located downstream of the failed node may be used as a replacement candidate. Further, for example, a node having a larger number of hops from the root node may be set as a replacement candidate.

  After selecting the replacement candidates, the content management server 3 next narrows down the replacement candidates based on the specification information stored in the delivery information database 4 for each replacement candidate delivery server (step 301). For example, the number of downstream nodes that can be accommodated by the replacement candidate is equal to or greater than the number of downstream nodes accommodated (connected) by the failed node. The replacement candidates that can secure the communication band in each path are narrowed down to replacement candidates that satisfy all three requirements that all user terminals assigned to the failed node can be accepted. In this step 301, based on the specification information, only the satisfaction or non-satisfaction of the requirements is marked, the replacement candidates are not narrowed down, and the replacement requirements are determined in the replacement node determination processing in step 303 described later. You may make it utilize the sufficiency of. In the above description, the failure node specification or actual value is used as a reference (threshold value) to determine the sufficiency of the replacement candidate specification. However, the failure node specification or actual value multiplied by a predetermined number is used as the threshold value. The sufficiency of the specification of the replacement candidate may be determined. When the number of user terminals accommodated by the failed node is M, M × a (a is 1.2) may be used as a threshold value to evaluate the number of surplus delivery candidates for replacement.

  After that, the content management server 3 sequentially measures the distance from each of the narrowed replacement candidates to the upstream node directly connected to the failed node, and stores the obtained distance information in the delivery information database 4 (step 302).

  Then, the content management server 3 determines the replacement candidate with the shortest measured distance as the node to be replaced with the failed node (step 303), and stores the internal storage information so that the determined node is replaced with the failed node and applied. Is changed (step 304). For example, the content management server 3 stores information on a node that replaces the failed node in the built-in failed node storage unit 23 in association with the failed node information.

  FIG. 10 is a sequence diagram showing processing (step 302) for measuring the distance between the replacement candidate and the upstream node directly connected to the failed node, which is executed under the initiative of the content management server 3.

  The content management server 3 requests the upstream node 2-X that is directly connected to the failed node to obtain a distance information specifying the replacement candidate node 2-Y (for example, the replacement candidate node 2-Y with a network address). (Specify) is transmitted (step 400). At this time, the upstream node 2-X transmits a packet including information instructed to return to the replacement candidate node 2-Y, and acquires and stores the transmission time (steps 401 and 402). The replacement candidate node 2-Y returns the packet when such a packet arrives (step 403). When such a return packet returns, the upstream node 2-X acquires the reception time (step 404), and obtains the average communication speed in the network 5 obtained in advance as the difference between the transmission time and the reception time. The network distance is calculated by multiplication (step 405), and the calculated network distance is returned to the content management server 3 (step 406). In determining the replacement node, the size of the distance becomes a problem. Therefore, there is no problem even if the round trip distance as described above is calculated and used.

  When a node that replaces the failed node is determined and the tree is reconfigured, content delivery from the root node to each node is executed according to the tree configuration configured at this time until the failed node is restored.

  Next, the replacement node determination method described above will be described with a specific example. Hereinafter, it is assumed that a failure occurs in the distribution server 2-2 in the basic tree configuration illustrated in FIG. 2 and it is necessary to determine a replacement node to replace the distribution server 2-2. As shown in FIG. 2, the distribution server 2-2 is a server having two downstream paths. In addition, the communication bandwidth of each of these paths is 5 [GB / s] and 5 [GB / s], respectively, and the distribution server 2-2 has 90 distributions allocated. .

  Here, the replacement candidate is at least three candidates, and the end node that is not located downstream of the failed node and that has the larger number of hops from the root node is given priority as the replacement candidate. Suppose that Since the end node has not performed the content delivery to the downstream, it is estimated that the capacity for delivering the content downstream is higher than the intermediate node. Further, since it is estimated that the number of hops from the root node to the replacement node is improved by the replacement, it is preferable to give priority to the end node having the larger number of hops from the root node. In addition, if a node located downstream of the faulty node is used as a replacement candidate, an end that is not positioned downstream of the faulty node is generated because a part of the tree is inverted and the degree of change of the tree is large. It is preferable to make a node a candidate for replacement.

  Of course, the selection method is not limited to the above. For example, a selection method that makes an intermediate node a replacement candidate may be used.

  When the selection method described above is applied, the replacement candidates for the failed node (distributed distribution server 2-2) are the nodes 2-7 to 2-9. If at least four candidates are used, the replacement candidates are nodes 2-7 to 2-9 and 2-11.

  It is assumed that the replacement candidates 2-7 to 2-9 have the specifications shown in FIG. Since the failure node 2-2 has two downstream paths, any of the replacement candidates 2-7 to 2-9 satisfies the requirement for the number of downstream paths. Since the communication bandwidth of each path of the failed node 2-2 is 5α and 5β [B / s], the replacement candidates 2-7 and 2-9 that can secure a communication bandwidth higher than this satisfy the requirements for the communication bandwidth. . Since the number of distributions assigned to the faulty node 2-2 is 90, the replacement candidates 2-7 and 2-9 having a surplus number of distributions satisfy the requirement of the number of distributions that can be assumed.

  In the selection of replacement candidates based on the above specifications, replacement candidates 2-7 and 2-9 remain. Therefore, the content management server 3 measures the network distance between each of the replacement candidates 2-7 and 2-9 remaining in this way and the upper node 2-0 of the failed node 2-2. When the measurement result is as shown in FIG. 5, the content management server 3 determines that the node (distribution server) 2-9, which is the replacement candidate with the shorter network distance, is replaced with the failed node 2-2. .

  FIG. 11A shows a reconfigured tree configuration in which the node 2-9 is replaced with the failed node 2-2. Since the node 2-9 is replaced with the failed node 2-2, the content is delivered from the upstream node 2-0 of the failed node 2-2, unlike before the reconfiguration.

  FIG. 11B shows the configuration of a tree reconfigured by applying the reconfiguration method described in Patent Document 1 when a failure occurs in the node 2-2. That is, the reconfiguration is executed from the viewpoint of accommodating the nodes 2-5 and 2-6 downstream of the failed node 2-2, and the node 2-5 is accommodated in the node 2-7. In this case, it is determined that −6 is accommodated in the node 2-9.

  As is clear from the comparison between FIG. 11A and FIG. 11B, the one to which the reconfiguration method of the first embodiment is applied approximates the original basic configuration shown in FIG. .

  The above shows the case where a failure occurs in the intermediate node 2-2 (the same method can be applied when a failure occurs in the root node), but when the failure occurs in the end node, for example, the following Process as follows. Here, it is assumed that a failure has occurred in the end node 2-9. Since this node 2-9 is an end node, it does not have a function of delivering content downstream. Therefore, when a failure occurs in the node 2-9, it is only necessary to determine a node to take over the number of distributions (user terminals) assigned to the node 2-9.

  Although illustration of a flowchart etc. is omitted, the content management server 3 first selects a takeover candidate. The selection method of the takeover candidate here is also arbitrary. For example, since the content distribution load does not increase, an intermediate node may be set as a takeover candidate. The content management server 3 narrows down to takeover candidates having a surplus number of distributions equal to or greater than the number of distributions allocated to the failure node 2-9. If a plurality of takeover candidates remain even after narrowing down, the content management server 3 finally determines a node to take over based on the distance of the failed node 2-9 to the upstream node 2-4. Since the fault node 2-9 is not replaced, the distance between the fault node 2-9 and the upstream node 2-4 is not directly affected, but the distance between the fault node and the takeover node is preferably short. Since the distance between the failed node and the succeeding node cannot be measured, the distance between the upstream node 2-4 of the failed node 2-9 and the succeeding node is used instead of the distance between the failed node and the succeeding node.

  Although illustration is omitted, for example, the content management server 3 periodically monitors whether or not the failed node has been recovered, and when the failed node is recovered, the content is delivered to each node according to a basic tree configuration. Return to. Also, for example, when the administrator inputs that the failed node has been recovered, the content management server 3 returns to a state in which the content is delivered to each node according to the basic tree configuration.

(A-3) Effect of First Embodiment According to the first embodiment, when a server failure occurs in a delivered node during content delivery, the basic configuration is not applied without applying the reconfigured tree configuration. The tree structure is applied as it is, and the content delivery to all nodes on the tree is realized, and the tree structure reconstructed from a new type of content delivery is applied. By understanding the tree structure to be cut, it is possible to reduce the possibility of a node not being delivered.

  Further, according to the first embodiment, when a failure node occurs, a node to be replaced with the failure node is determined and replaced on the tree. Therefore, even if the tree configuration is reconfigured, the tree configuration Change can be reduced. That is, even if the tree is reconfigured, it is possible to suppress the appearance of a node that has a large number of hops from the root node and takes a long time to deliver the content.

  Furthermore, according to the first embodiment, specifications (processing capability) information is stored for nodes that are candidates for replacement, and it is confirmed that the function that the faulty node has been able to be secured is determined. As a result, it is possible to prevent the content delivery and distribution from being adversely affected by the lack of capacity of the replaced node.

  Furthermore, according to the first embodiment, when there are a plurality of replacement candidates that can satisfy the function of the failed node in terms of specifications, the replacement node is determined based on the network distance. In a tree configuration in which a node is replaced, the content delivery time to the replaced node can be shortened.

  Further, according to the first embodiment, when a failure occurs during content delivery, a node that replaces the failed node is determined as part of the delivery process, and the content is delivered to the decision node. Therefore (see FIG. 8), it is possible to prevent the time required for delivering and deploying the contents to all the nodes from increasing due to the occurrence of the faulty node.

(B) Second Embodiment Next, a second embodiment of the tree-shaped network reconfiguration device and program and the tree-shaped network reconfiguration method according to the present invention will be described with reference to the drawings. The tree-like network system of the second embodiment is also a content distribution system.

  The overall configuration of the content distribution system 1A of the second embodiment can also be represented in FIG. 1 used in the description of the first embodiment. The internal configuration of the distribution server 2-n of the second embodiment can also be represented in FIG. 3 used in the description of the first embodiment. Furthermore, the content management server 3 of the second embodiment can also be represented by FIG. 4 used in the description of the first embodiment.

  The second embodiment is different from the first embodiment in a tree reconfiguration process executed by the content management server 3, that is, a determination process of a node that replaces a faulty node. Other processes are the same as those in the first embodiment. It is the same as the form. FIG. 12 is a flowchart illustrating a tree reconfiguration process (reconfiguration program) executed by the content management server 3 according to the second embodiment. For example, a tree reconstruction process is executed as the process of step 253 in FIG. 8 described above.

  When the content management server 3 starts the process shown in FIG. 12, first, a replacement candidate is selected (step 500). Also in the case of the second embodiment, the method for selecting replacement candidates is not limited, and examples include the method exemplified in the description of the first embodiment.

  After selecting the replacement candidate, the content management server 3 next calculates the performance value of the spec based on the spec information stored in the delivery information database 4 for each replacement candidate delivery server, The performance value is stored in the delivery information database 4 (step 501).

  FIG. 13 is an explanatory diagram illustrating a configuration example of candidate evaluation information in the delivery information database 4 according to the second embodiment, and illustrates information of some distribution servers that can be replacement candidates.

  In FIG. 13, the candidate evaluation information includes a row (record) for each replacement candidate identification information, and each row includes spec raw value information, spec stage value information, spec specs, and the like as a server of the candidate node. Performance value, distance information, and priority (total evaluation value). Before the processing shown in FIG. 12 is started, only spec raw value information is described, and spec stage value information, spec performance value, distance information, and priority (total evaluation value) are blank. These are described gradually as the process shown in FIG. 12 proceeds.

  The spec raw value information is information called “spec information” in the first embodiment. For example, (a) the number of downstream nodes that can be accommodated (the number of paths that can be connected to the downstream nodes), (b) This corresponds to the communication bandwidth in each path to the downstream node, and (c) the number of contents that can be distributed (the number of distributions that can be increased from the current state).

  The spec stage value information is obtained by dividing the possible range of the spec raw value information into a plurality of ranges, and replacing them with values indicating to which evaluation stage the divided range to which the raw value belongs (all items). The scale of the stage value is based on the evaluation of the item). The raw value division range corresponding to the evaluation stage varies depending on the current state of the faulty node.

  In the case of FIG. 13, the numbers that can accommodate the downstream nodes that are the raw value information (a) of the specifications in the replacement candidates 2-7 to 2-9 correspond to the downstream path of 5, 3, and 4, respectively. To get. Assume that the number of paths downstream from the failed node is two. When the number of downstream paths of the faulty node is two, the spec stage value information in the number that can accommodate the downstream node, which is the spec stage value information (a) of FIG. In this case, the step value is “0”, the step value is “1” when the number of paths that can be secured is 2 to 4, and the stage value is “2” when the number of paths that can be secured is 5 or more. If determined, the step values for the number of paths downstream of the replacement candidates 2-7 to 2-9 are “2”, “1”, and “1”, respectively. If the number of paths downstream from the failed node is 5 and the number of paths downstream from the failed node is 5, if the number of paths that can be secured is 0 to 4, the step value is set to “0”. If it is determined that the stage value is “1” when the number of possible paths is 5 or 6, and the stage value is “2” when the number of paths that can be secured is 7 or more, the replacement candidate 2-7 The step values for the number of downstream paths of ˜2-9 are “1”, “0”, and “0”, respectively.

  In the above, the case where the range of raw values that each stage value can take is switched according to the number of paths downstream of the failed node. However, each stage value is taken regardless of the number of paths downstream of the failed node. The range of raw values to be obtained may be fixedly determined.

  With regard to the communication band and the number of contents that can be distributed, the specification step value information is formed from the specification raw value information by the same method. However, since the communication band is determined for each path, after converting it to a step value for each path, the average value of the step value information is taken as step value information for the communication band (note that Decimal numbers may be accepted as average values, and rounding or rounding processing is performed when limiting to integers).

  In the case of FIG. 13, in the replacement candidates 2-7 to 2-9, the communication bandwidth in each path to the downstream node that is the raw value information (b) of the specification is 10 [GB / s] ( 5), 5 [GB / s] (3), and 10 [GB / s] (4). Assume that the communication bandwidth downstream of the failed node is 5 [GB / s]. When the communication band downstream of the faulty node is 5 [GB / s], the specification step value information in the communication band in each path to the downstream node, which is the specification step value information (b) of FIG. The step value is “0” when the communication band selected from the two having a higher communication band among the various communication bands is less than 5 [GB / s], and the communication band having the higher communication band among the reservable communication bands is 2 When the communication band selected from the book is 5 [GB / s] or more and less than 10 [GB / s], the step value is “1”, and the communication band selected from the two that have a higher communication band among the available communication bands If the determined communication band is 10 [GB / s] or more and the stage value is determined to be “2”, the stage regarding the number of paths downstream from the replacement candidates 2-7 to 2-9 The values are “2” and “1”, respectively. "2".

  Next, in the replacement candidates 2-7 to 2-9, the number of content deliverables that are the raw value information (c) of the specifications can correspond to 120, 0, and 110, respectively. Assume that the number of distributable contents of the faulty node is 100. When the number of content deliverables downstream of the faulty node is 50, the spec stage value information in the content deliverable number, which is the spec stage value information (c) in FIG. Sometimes the stage value is “0”, the stage value is “1” when the number of content deliverables is 50 or more and less than 100, and the stage value is “2” when the number of content deliverables is 100 or more. If so, the step values for the number of downstream paths of the replacement candidates 2-7 to 2-9 are “2”, “0”, and “2”, respectively.

  In FIG. 13, all items (the number of downstream paths, the communication bandwidth, and the number of contents that can be distributed) are shown as having three stages, but the number of stages may be changed depending on the item.

  The performance value of the specification is obtained by integrating the step value information of each item. In FIG. 13, the product obtained by multiplying the step value information of each item is used as the performance value of the specification. However, other integration methods may be applied. For example, the performance value of the specification may be obtained by weighted addition of the step value information of each item.

  In the case of FIG. 13, the performance value of the spec stage value information is “the number of downstream nodes that are the spec stage value information (a)” and “downstream that is the spec stage value information (b)”. From the product of the “spec stage value information in the communication band in each path to the node” and the “spec stage value information in the number of deliverable contents as the spec stage value information (c)”, replacement candidates 2-7 to The performance values that are the stage value information of the specifications 2-9 are “8”, “0”, and “4”, respectively.

  After calculating the performance value of the spec, the content management server 3 then sequentially measures the distance from the upstream node directly connected to the failed node for each replacement candidate, and uses the obtained distance information as the delivery information database. 4 (step 502). In the case of the second embodiment, the replacement candidates are not narrowed down based on the specification information, and therefore distance measurement is executed for all selected replacement candidates.

  The content management server 3 then gives priority to a total evaluation value obtained by merging the spec performance value and the distance information for each replacement candidate based on the spec performance value and the distance information stored in the delivery information database 4. The degree of replacement is calculated (step 503), the replacement candidate with the highest priority is determined as the node to be replaced with the failed node (step 504), and the internal storage information is applied so that the determined node is replaced with the failed node. Is changed (step 505).

  Any method can be used as long as it can calculate the priority reflecting the performance value and distance information of the specification. In FIG. 13, the value obtained by dividing the performance value of the spec by the distance information and multiplying by 10 (correction value) is set as the priority value. In the example of FIG. 13, the priority P is the step value information of the specifications of the replacement candidates 2-7 to 2-9 based on the quotient of “performance value as the step value information of the specification of FIG. 13” and “distance”. The certain performance values are “9.87”, “0”, and “8.88” (rounded down to the second decimal place), and the replacement candidate 2-7 having the highest priority P is the failure node. It is determined as a node to replace.

  According to the second embodiment, substantially the same effect as that of the first embodiment can be obtained. According to the second embodiment, the specification of the replacement candidate is evaluated stepwise and reflected in the priority, and the node that replaces the failed node is determined based on the priority. Can be determined as a node to be replaced, and an effect that the load is distributed from the first embodiment can be expected.

  Incidentally, in the case of the first embodiment, only the sufficiency of the spec is determined, so if the spec is satisfied, the load may be determined as a replacement node even if the load is close to the capacity (spec) and there is little margin. There was quite a lot.

(C) Other Embodiments In the description of each of the above-described embodiments, various modified embodiments have been referred to. However, modified embodiments as exemplified below can be given.

  In each of the above embodiments, the specifications of the distribution server used for selecting the replacement node are the number of paths to the downstream node, the communication bandwidth, and the number of surplus distributions. However, the specifications used for selecting the replacement node are It is not limited to. For example, the operating clock frequency of the CPU mounted on the distribution server, the total capacity or free capacity of the memory mounted on the distribution server, and the like may be used as specifications used for selecting the replacement node. The combination of specifications to be applied is not limited to those in the above embodiments. For example, the number of paths to the downstream node, the communication band, and the operation clock frequency of the CPU may be used.

  In each of the above embodiments, the parameter used to determine the node that replaces the failed node other than the specification is the network distance between the upstream node of the failed node and the replacement candidate, but instead of this, or in addition to this Other parameters may be used.

  For example, when each path is considered to be capable of bidirectional communication, the number of paths (the number of hops) that connects the upstream node of the failed node and the replacement candidate in the shortest on the basic tree configuration may be used. A path between the nodes 2-i and 2-j is represented by P (2-i, 2-j). For example, in FIG. 2, when a failure occurs in the node 2-5 and the upstream node of the failed node is the node 2-2, if the replacement candidate is the node 2-11, the path P (2-0, 2 -2), P (2-0, 2-1), and P (2-1, 11) can be used to connect both nodes. When a failure occurs in the node 2-5 and the upstream node of the failed node is the node 2-2, if the replacement candidate is the node 2-7, the paths P (2-0, 2-2), P ( Both nodes can be connected by four paths 2-0, 2-1), P (2-1, 2-3) and P (2-3, 2-7). Such parameters can be treated as approximating the network distance.

  Further, for example, the distance on the network between the content management server and the replacement candidate may be used for determining the replacement node. Here, when the content management server is shared with the root node, or when the content management server is located upstream of the root node in the basic tree configuration, the content management server is used instead of the network distance. Then, the number of paths (hops) connecting the replacement candidates in the shortest on the basic tree configuration may be applied. If there is a large amount of control information exchange between the replacement node and the content management node, such control information exchange can be quickly executed.

  In each of the above-described embodiments, the content management server 3 stores the specification information of the replacement candidate and uses the specification information stored in the replacement node determination process. However, each node that can be a replacement candidate However, the content management server 3 may collect the specification information of each candidate from the replacement candidates each time and use it in the replacement node determination process.

  Further, in each of the above embodiments, the case where the event for reconfiguring the tree configuration is a failure of one of the nodes, but other events may be used. For example, the present invention can be applied to the case of reconfiguration of a tree when it is separated from a tree-like network in order to periodically check nodes.

  Further, in each of the above embodiments, the present invention is applied to a content distribution system in which a plurality of distribution servers that distribute content to user terminals are logically distributed in a tree shape. The application of the invention is not limited to this. A system in which a plurality of servers are logically arranged in a tree shape, and delivery data (not limited to content) is distributed while branching from the root node to the end node of the tree, thereby causing each server to deploy the same data. If it is good. Therefore, a system without a function for distributing data from a server to a user terminal or the like may be used.

  DESCRIPTION OF SYMBOLS 1 ... Content delivery system, 2-0-2-N ... Distribution server, 3 ... Content management server, 4 ... Delivery information database (delivery information DB), 10 ... Communication part, 11 ... Delivery content receiving part, 12 ... Content storage , 13 ... delivery content transmission unit, 14 ... delivery control unit, 20 ... communication unit, 21 ... administrator interface unit, 22 ... tree basic configuration storage unit, 23 ... fault node storage unit, 24 ... tree reconfiguration unit, 25 ... a delivery instruction control unit.

Claims (8)

In a tree-like network reconfiguration device that reconfigures a tree configuration to a configuration in which one server is excluded based on the appearance of a predetermined event from a network in which a plurality of servers arranged in a distributed manner are logically connected in a tree shape ,
The occurrence of the predetermined event is any of the cases in which the first data is distributed from the server of the root node of the tree-like network to all the servers of the tree-like network along the tree structure. The server has failed,
Candidate selection means for selecting a replacement candidate server that is a candidate for replacing the server to be excluded;
For each of the replacement candidate servers, candidate spec information capturing means for capturing the spec information of the server,
A replacement server determination means for determining a server that replaces the server to be excluded, on the condition that the replacement candidate server can secure more than the performance of the server to be excluded as viewed from the specification information of the replacement candidate server ;
A delivered node confirmation means for determining whether the first data has already been delivered and deployed to a server downstream of the failed server;
If the first data is not delivered or deployed to a server downstream of the failed server, the reconstructed tree replaced by the server determined by the replacement server determining means is transferred to the undelivered server. If the first data is already delivered and deployed to a server downstream of the failed server, the tree before reconfiguration is determined as a tree to be applied to the delivery of the first data. In the distribution of the new second data that will be determined in the future for the delivery of the first data to the undelivered server, and the new second data that will be generated in the future, the replacement that has been replaced by the server determined by the replacement server determination means is performed. An application tree determining means for determining a tree to which the constructed tree is applied;
A tree-like network reconfiguring device characterized by comprising: The replacement server determining means individually confirms that each item of the specification information satisfies the performance required for the server to be excluded, and determines a server to replace the server to be excluded. The tree-shaped network reconfiguration apparatus according to claim 1 . 2. The tree-like network reconfiguration according to claim 1 , wherein the replacement server determining means calculates evaluation information obtained by integrating information of each item of the specification information, and determines a server that replaces the server to be excluded. Configuration equipment. The replacement server determining means prioritizes the replacement candidate server having a shorter network distance to the predetermined server or the external device, or the replacement candidate server having a smaller number of hops when communicating with the predetermined server or the external device. 4. The tree-shaped network reconfiguring apparatus according to claim 2, wherein the server is replaced with the server to be excluded. The tree-shaped network reconfiguring apparatus according to any one of claims 2 to 4 , wherein the specification information is described as a different value for each minimum unit. The tree-like network reconfiguration according to any one of claims 2 to 4 , wherein the spec information is described as a step evaluation value obtained by dividing a possible range into a plurality of ranges. apparatus. A tree-shaped network reconfiguring device that reconfigures a tree configuration from a network in which a plurality of distributed servers are logically connected in a tree configuration to a configuration in which one server is excluded based on the appearance of a predetermined event A tree-shaped network reconfiguration program for operating a computer installed therein ,
The occurrence of the predetermined event is any of the cases in which the first data is distributed from the server of the root node of the tree-like network to all the servers of the tree-like network along the tree structure. The server has failed,
The above computer
Candidate selection means for selecting a replacement candidate server that is a candidate for replacing the server to be excluded;
For each of the replacement candidate servers, candidate spec information capturing means for capturing the spec information of the server,
A replacement server determination means for determining a server that replaces the server to be excluded, on the condition that the replacement candidate server can secure more than the performance of the server to be excluded as viewed from the specification information of the replacement candidate server ;
A delivered node confirmation means for determining whether the first data has already been delivered and deployed to a server downstream of the failed server;
If the first data is not delivered or deployed to a server downstream of the failed server, the reconstructed tree replaced by the server determined by the replacement server determining means is transferred to the undelivered server. If the first data is already delivered and deployed to a server downstream of the failed server, the tree before reconfiguration is determined as a tree to be applied to the delivery of the first data. In the distribution of the new second data that will be determined in the future for the delivery of the first data to the undelivered server, and the new second data that will be generated in the future, the replacement that has been replaced by the server determined by the replacement server determination means is performed. An application tree determining means for determining a tree to which the constructed tree is applied;
A tree-shaped network reconfiguration program characterized by being made to function. In a tree-like network reconfiguration device that reconfigures a tree configuration to a configuration in which one server is excluded based on the appearance of a predetermined event from a network in which a plurality of servers arranged in a distributed manner are logically connected in a tree shape In the tree-shaped network reconfiguration method,
  The occurrence of the predetermined event is any of the cases in which the first data is distributed from the server of the root node of the tree-like network to all the servers of the tree-like network along the tree structure. The server has failed,
The tree-shaped network reconfiguring apparatus includes candidate selection means, candidate specification information fetching means, replacement server determination means, delivered node confirmation means, and application tree determination means,
A candidate selection step executed by the candidate selection means for selecting a replacement candidate server that is a candidate for replacing the server to be excluded;
For each replacement candidate server, a candidate spec information capturing step executed by the candidate spec information capturing means for capturing the spec information of the server,
Executed by the replacement server determining means for determining a server that replaces the server to be excluded, on the condition that the replacement candidate server can secure the performance capability of the server to be excluded or higher, as viewed from the specification information of the replacement candidate server A replacement server determination step;
A delivered node confirmation step executed by the delivered node confirmation means for determining whether the first data has already been delivered and deployed to a server downstream of the failed server;
If the first data is not delivered or deployed to a server downstream of the failed server, the reconstructed tree replaced by the server determined by the replacement server determining means is transferred to the undelivered server. If the first data is already delivered and deployed to a server downstream of the failed server, the tree before reconfiguration is determined as a tree to be applied to the delivery of the first data. In the distribution of the new second data that will be determined in the future for the delivery of the first data to the undelivered server, and the new second data that will be generated in the future, the replacement that has been replaced by the server determined by the replacement server determination means is performed. An application tree determining step executed by the application tree determining means for determining a tree to which the constructed tree is applied;
A tree-shaped network reconfiguration method characterized by comprising:

Images