JP2003271440A - Contents delivery management system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a contents delivery management system whose operation costs are small, and flexibility concerning construction or configuration change, reliability and availability are excellent. <P>SOLUTION: In the contents delivery management system providing retrieved contents utilizing prescribed logical structure on a wide area distributed communication network, a contents storage server, the most significant identification information server, a proxy identification information server, and a normal state supervisory server are arranged on the logical structure. A node receiving notice issued by the normal state supervisory server (excluding the most significant identification information server positioned at the most significant position on the logical structure, the contents storage server or the proxy identification information server) performs connection change operation for changing a parent when the node itself is a child of the node (the proxy identification information server) determined abnormal by the notice. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a content distribution management system. For example, different contents are accumulated in a plurality of content distribution servers on the Internet or the like, and one of the content distribution servers responds to a distribution request. It is suitable to be applied to cases such as delivery of.

[0002]

2. Description of the Related Art Conventional content distribution management systems include the inventions disclosed in the following documents 1 and 2 relating to US patents.

Reference 1: System and method for se
rver-sideoptimization of datadelivery on a di
stributed computer network Patent Number: 6,112,239 Document 2: Method and system for directing a
The invention of the above-mentioned reference 1 relates to a method / system for analyzing network performance between a client and a plurality of content distribution servers and distributing Web content access from the client to an appropriate content distribution server. It is an invention. This is a method of selecting an appropriate content distribution server for each client when a content request from the client occurs based on the contents of a database that stores the analysis result of network performance.

In the invention of the above-mentioned document 2, the content type specified in the content request issued from the client / QoS request specification, load of the content distribution server group available at the time of request / network congestion / client and server The distance from the client to the We
b An invention relating to a method / system for allocating content access to an appropriate content distribution server. By managing the collected information as the performance metrics of the content distribution server, a suitable content distribution server is selected for each client when a content request from the client occurs.

The inventions of Document 1 and Document 2 all have a plurality of servers for delivering contents, but it is premised that the same contents are available in those servers.

Further, there are the following conventional file supply systems.

System 1: http: //www.oceanstore.cs.berkeley.ed
u System 2: http://www.napster.com In the above system 1, multiple servers connected to the network are connected in a network form (a sparse connection in which one server is not connected to all other servers). It is a system with a different configuration. In this system 1, it is possible to search for a target file by sequentially tracing the servers connected by the network. If the target file is found, transfer the file directly.

That is, the file can be directly transferred from the server having the file and acquired instead of being bucket-relayed according to the search procedure. In this system 1, as a search means, "bloo"
The m method is used. There is no server that centrally manages the information of files that exist in the system.

On the other hand, the system 2 has a server for centrally managing information of files existing in the system,
The server that holds the file must be connected to the server that centrally manages the file information. When a server that has a file acquires a specific file, the server that centrally manages the file information searches for the specific file,
Obtain a specific file directly from another server that has that specific file.

[0010]

However, in the content delivery systems of the above-mentioned Documents 1 and 2, all servers that deliver content (text, audio, video, etc.) hold the same content (consistent content consistency). Therefore, if this system is applied when the content changes frequently, the more servers that distribute the content in a wider area, the more the content in the server becomes the same. There is a problem that a lot of communication traffic occurs between servers, and input / output processing for storing new content and deleting unnecessary content occurs frequently in the server.
The cost of operating the system was high.

Further, since all the servers that distribute the contents require a storage medium such as a large-capacity hard disk capable of storing all the contents, it is difficult to add a small-scale server, and the system construction and configuration are difficult. There was the problem of low flexibility regarding changes.

Further, the above-mentioned system 1 and system 2
In this file sharing system, files with the same file name but different contents may be registered, and the uniqueness of the file name to the contents of the file is not guaranteed, so a file different from the target file may be However, there was a problem that the system was unreliable because it was sometimes taken out.

In order to solve the above problems, all content information uniquely indicating the location of each content in the system is stored in the index server, and at least a part of the content information stored in the index server. A system configuration is conceivable in which is stored in the index proxy and the content delivery server (the content delivery server also stores the main body of the content).

In this system configuration, for example, a tree having an index server as a root, an index proxy as a middle node, and a content distribution server as a leaf is used, and when the content distribution server receives an acquisition command for a certain content. If the content information of the content is held, the content distribution server returns the content. However, if the content information is not held, a search request for the content information corresponding to the acquisition command is passed to the higher section, and the section accumulating the content information returns the content information. Then, the content distribution server that has obtained the content information can receive the desired content from any of the content distribution servers in the system specified by the content information and can respond to the acquisition command.

However, in such a system configuration, there is a case where access to the parent using the tree may be necessary, and therefore, when the parent fails or communication with the parent cannot be performed, the contents are not stored. There are cases where various operations such as distribution are hindered and normal services cannot be provided.

In such a case, it is effective for each section to redefine the definition of its parent. However, if the definition of the parent depends on manual work by a maintenance worker or the like, the work may take time. As a result, it takes a long time for normal service to be performed, resulting in a problem that the availability of the system decreases.

Also, if the newly defined parent has a large number of hops from that node or is in a state where the processing capacity cannot be afforded due to an excessive load, sufficient performance can be obtained even if it is redefined. It may not be possible. In order to obtain sufficient performance, it is necessary to understand the connection form on the public network such as the Internet to some extent, but it is not possible to select an appropriate parent in consideration of these various conditions. It is extremely difficult for ordinary maintenance workers.

[0018]

In order to solve such a problem, according to the present invention, in a content distribution management system for providing contents searched using a predetermined logical structure on a widely distributed communication network, (1)
In the logical structure, it is the lowest node with no children,
The contents and at least contents identification information including position information indicating the position of a node on the logical structure in which the contents are accumulated are accumulated, and in response to an inquiry from the outside of the logical structure, to the upper node of the inquiry. A content storage server that relays or distributes the content, and (2) all the content storage servers that are top-level nodes that have no parent in the logical structure and that are present in the logical structure Content identification information relating to the content of, and returning the content identification information in response to an inquiry from a lower node, and (3) located in a middle position having a parent and a child in the logical structure. It is a node that stores at least a part of the content identification information stored by the highest identification information server. By returning the content identification information as long as it is an inquiry about the stored content identification information, it functions as a proxy for the highest-level identification information server to the lower node that is a child of itself, and identifies the content that is not stored. For inquiries about information, a proxy identification information server that relays to the upper node,
(4) Normality monitoring that monitors the normality of operations related to the highest-level identification information server or proxy identification information server, and if a monitoring result indicating an abnormality is output, notifies that fact to the node in the logical structure. (5) The node having the server performs the connection change operation for changing the parent when (5) the node that received the notification is a child of the node that is abnormal in the notification. .

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION (A) Embodiment Hereinafter, an embodiment of a content distribution management system according to the present invention will be described.

In this embodiment, when a failure occurs in a component in a content delivery management system using a P2P (peer-to-peer) content delivery server, the system is configured to dynamically change the logical configuration of the system. It provides a fault-tolerant system that can maintain the overall functionality.

(A-1) Configuration of Embodiment FIG. 1 shows an example of the overall configuration of a content distribution management system 10 according to this embodiment. Content distribution management system 1
The respective constituent elements 11 to 0 are distributed geographically in a wide area and can be connected online by the Internet or the like.

In FIG. 1, the content distribution management system 10 includes an index server 11
And Index Proxy 12-14
, A P2P server (P2P Server) 15 to 18, and a lookup server (Lookup Server) 19.

As is apparent from FIG. 1, the components 11 to 11 are
18 constitutes a logical tree as a whole. That is, the root (top node) corresponds to the index server 11, the middle node corresponds to the index proxies 12 to 14, and the leaf (bottom node) corresponds to the P2P servers 15 to 18.
Is equivalent to.

Of these, the P2P servers 15-18, which are the leaves
Is a server that stores and manages original content, content information, and a cache thereof, and is the closest among the components shown in FIG. 1 to a client that requests distribution of content or registers content. Exists in a position.

Since the P2P servers are connected peer-to-peer on an equal basis, and information such as contents can be freely exchanged, the P2P servers 15 to 18 complement each other. It is possible to deliver contents.

That is, content that is not stored by itself is delivered after being received from another P2P server, and conversely, whenever there is a request from another P2P server, the content that is stored by another P2P server is sent to another P2P server. Send to P2P server.

To enable such complementation, each P2
The P servers 15 to 18 basically store and manage different contents, which are described in Document 1 and Document 2 above.
This is one of the features of the present embodiment when compared with. Figure 1
In the example, the content GF2 (for example, still image data GF2) is accumulated in the P2P server 15, and the content G different from the content GF2 is stored in the P2P server 16.
F1 is accumulated, and these GF1, and GF1, are stored in the P2P server 17.
Content GF3 different from GF2 is stored.

In the present embodiment as well, depending on the operation state such as content distribution, the same content (for example, still image data GF1) may be stored in the P2P servers 15-18, but the original content GF1 is stored in the system 10. There is only one inside, and the others are contents obtained by copying GF1 as the original contents. For example, as illustrated, the P2P server 16
And 18 store and manage the content GF1,
If the content GF1 on the P2P server 16 side is the original content GF1 on the P2P server 18 side
Is a copy.

Copy content in each P2P server (for example, the content GF in the P2P server 18)
The accumulation of 1) is dynamically executed by using the cache storage device, and therefore is dynamically deleted when the capacity of the cache storage area becomes insufficient. Even if the copy content (cache content) is deleted, the deletion does not affect the content stored in another P2P server (for example, 16).

In the P2P servers 15 to 18, content (including copy content) and content information of the content are always managed as one unit. Therefore, the original content GF1 is stored in the P2P server 16.
When the content is stored, the content information IF1 of the content GF1 is also stored in the P2P server 16. Similarly, copy content GFC1
Is stored (cached) in the P2P server 18, its content information IFC1 is also stored in the P2P server 1.
Cached in 8.

The content information such as the content information IF1 and IF2 is information attached to the content such as the content name in which the content is stored, the creation date, the update date, the size, the service ID, and the type. Of particular importance are the content name and service ID. The service ID is the identifier of the P2P server, and the type is the type of content information such as original content and cache content.

When the client issues a content acquisition command requesting distribution of content, if the content having the content name of the content acquisition command exists in the P2P server, the content is returned. If there is a cache of the content information, the P
The 2P server can acquire the content by immediately communicating with another P2P server based on the content information and deliver the content to the client via the Web server described later, but the P2P server holds all the content. Not not. That is, when the content information does not exist in the own P2P server, it becomes necessary to search the content information (content information search).

In this embodiment, the original and the copy are guaranteed to be the same. Therefore, when registering (accumulating) new content in any P2P server (for example, 16), it is important to give a unique value in the system 10 as the content name in advance. In the present embodiment, since each P2P server 15-18 can obtain complete content information from the index proxy or the like, it is possible to confirm whether or not the content name to be given at the time of registration is unique, and it is not unique. In such a case, such uniqueness can be given to the content name by performing processing such as prompting re-input of the content name.

Although not shown, each of these P2P servers 15 to 18 is connected to a Web server for accepting access from a client.
The content acquisition command for content distribution is received by an external call from the Web server by rvlet, CGI, or the like. This We
The b server is a general one and does not perform any special operation for the content distribution management system 10.
Also, it is possible to provide a plurality of Web servers for one P2P server.

Further, the client is client software such as a Web browser which operates according to a user's operation.

It is desirable to interpose a DNS system having a load balancing function between the client and the Web server. The load balancing function distributes the access from the clients according to the loads on the P2P servers 15 to 18, thereby controlling the loads on the P2P servers 15 to 18 as evenly as possible.

When a content acquisition command is received via these DNS systems or the Web server, each P2P server 15-18 accumulates the content body requested by the content acquisition command by itself (original content or copy content). If it is not stored, the content is immediately returned, and if it is not stored, the content information is obtained by the content information search, the content requested by the content acquisition command is obtained, and then the content is returned.

Further, P which has received the content acquisition command
The 2P servers 15-18 have contents (for example, GF1)
Content is not cached and content information (eg I
When only F1) is cached, the content can be obtained based on the content information.

In order to ensure the consistency of the content information distributed in the system 10, each P2P server 1
5 to 18 need to exchange information such as update regarding the content information with the index proxies 12 to 14 and the index server 11 when the content body is changed, deleted, or added.

P2P server 15-having such a function
The internal structure of 18 is as shown in FIG. 5, for example. P2P
Although the internal configuration of the servers 15 to 18 may be substantially the same, here, mainly, the P2P server 15 is shown in FIG.
Will be described as follows.

(A-1-1) Internal Configuration Example of P2P Server In FIG. 5, the P2P server 15 includes a distributed environment 21 and
P2P module 22 and Admin module 23
And a database 24 and a hard disk 25.

Among these, the distributed environment 21 is the P2P server 1
5 is a basic environment in which it operates, and is usually configured by Jini.

The P2P module 22 is the P2P server 1
5 is an interface for connecting P5 to other P2P servers 16-18, index proxies 12-14, or index server 11, and an interface mounting unit for changing the setting of P2P server 15.

The Admin module 23 is a mounting unit of an interface for performing management. Since the P2P server 15 that is a leaf always has a parent that is an upper node, the Admin module 23 manages the setting of the upper site that is the parent and the termination of the P2P server 15.

The database 24 stores the content information (eg, IF2) registered in the P2P server 15.
It is a database for storing. Database 24
For this, a relational database is usually used. However, it is natural that the relational database needs a non-volatile storage device such as a hard disk in order to store a real table and the like.

The hard disk 25 is the P2P server 15
This is an area for storing the content (for example, GF2) registered in.

It is located in the middle of the tree and always has one parent (another index proxy or index server 1
1) and index proxies 12-14 having 0 or more children (other index proxies or P2P servers)
Is a server that caches content information.

The logical arrangement of the index proxies 12 to 14 is as shown in FIG. The advantage of providing the index proxies 12 to 14 is that the index server 11 can reduce the load for responding to inquiries about content information and the like, and each P2P server 15 to 18 can obtain a response time longer than an inquiry to the index server 11. Since it can be shortened, it is desirable to arrange it so as to obtain such an advantage. As an example, compared with the number of hops of the route from the P2P servers 15 to 18 to the index server 11,
Index proxies 12 to 1 at positions with few hops
Arranging 4 is preferred.

Therefore, the index server 11 is set to P
When the 2P server (for example, 15) is located outside the site to which it belongs, it is preferable to place the index proxies 12 to 14 inside the site, and both the index server 11 and the index proxies 12 to 14 are placed outside the site. In this case, it is preferable to arrange the index proxies 12 to 14 at a position where the number of hops is smaller.

However, even if the index proxies 12 to 14 are arranged at the positions contrary to such conditions, it is possible to obtain the advantage of reducing the load on the index server 11.

The index proxies 12 to 14 can be freely added, changed, or deleted from the system 10. For example, when the load of the entire system 10 increases and the throughput decreases, adding a new index proxy is also an effective measure.

Index server 11 which is the root of the tree
Are connected to one or more index proxies or P2P servers, and the P2P servers 15-1 in the system 10
8 is a part for accumulating the content information of all the contents accumulated in 8.

In the illustrated state, the index server 11 is not directly connected to the P2P server, but in principle, the index server 11 can be directly connected to any P2P server.

All of the servers (11 to 18) except the lookup server are normally operated by the lookup server 19 and the servers (11 and) 12 to 18 depending on whether or not an update request signal described later is received. Since it has a program that listens for (receives) a server communication failure notification that is sent to indicate whether or not it is present, the server that is not operating normally at that time is a P2P server (for example, 1
In the case of 5), the deletion function is provided to delete the content information regarding the original content and the copy content registered in the P2P server from the storage area in each server (11-18).

By this delete function, the contents of the down P2P server are excluded from the search target,
It is possible to return the search result according to the state of each P2P server at that time.

The internal structures of the index server 11 and the index proxies 12 to 14 having such a function are as shown in FIG. 6, for example. Although the roles played in the system 10 are largely different, the index server 11 and the index proxy 12 ...
There is no difference between 14 and it can be said that they are all substantially the same.

(A-1-2) Example of Internal Configuration of Index Proxy In FIG. 6, the index proxy 12 to 14 or the index server 11 includes a distributed environment 31, a P2P module 32, an Admin module 33, and a database 34. I have it.

Here, the function of each component having the same name is
It is the same as the P2P server shown in FIG.

Since the index proxies 12 to 14 and the index server 11 store only the content information and not the content body, they correspond to the hard disk 25 which is an area for storing the content. The components do not exist on FIG.

In FIG. 1, the index proxies 12 to 1
The number of four is three and the number of P2P servers 15 to 18 is four, but these numbers are not limited to those illustrated. If the content distribution management system 10 is large, these numbers are likely to be much higher.

The lookup server 19 existing outside the above-mentioned tree is a part for registering and managing all the nodes of the tree, that is, the P2P servers 15-18, the index proxies 12-14, and the index server 11. . In FIG. 1, it is shown as a single server.

More specifically, the lookup server 1
Reference numeral 9 has a function of processing a registration update request from each of the servers 11 to 18.

In the present embodiment, each of the registered servers 11 to 18 normally transmits an update request signal for updating the registration of the lookup server 19 on a regular basis, but the lookup server 19 receiving this signal. Updates the registration of each server in response to the update request signal, and deletes the registration of the server (for example, 12) that cannot receive the update request signal, and notifies the server communication failure notification regarding the server. It is necessary to have a function of transmitting to each server in the system 10.

A server that cannot receive the update request signal is highly likely to have a failure in the server itself or to have a problem such as congestion in a transmission line around the server.
In any case, there is a high possibility that effective communication cannot be performed with other servers (communication unavailable state).

In such a communication disabled state, the servers 11 to 1
8 can occur in any server, but the influence and the corresponding operation differ depending on which server occurs. Generally, the more incommunicable state occurs in the upper node of the tree,
The influence on the entire system 10 is large, and the influence is smaller in the incommunicable state that occurs in a lower node.

For example, the lowest server (eg, P2P)
When a communication incompatibility state occurs in the server 16), it is not possible to perform all the processing that causes access to the P2P server 16, but it is possible to perform all the processing that does not require access to the P2P server 16. it can.

Even when the index proxies 12 to 14 and the index server 11 located at a higher level are in a communication-disabled state, similarly to this, whether or not the process can be executed is determined according to the necessity of accessing the section. .

For example, the operation of changing or deleting the content of any content stored in the P2P servers 15 to 18 or registering a new content is performed by using the content information stored in the index server 11. Since access is required to change or delete, or to add new content information to the index server 11, when the index server 11 becomes incommunicable, all P2P servers in the system 10 perform these operations. Can not be.

The operation of the present embodiment having the above-mentioned structure will be described below with reference to FIGS.

The operation of this embodiment is performed by any one of the servers 1.
The operation can be divided into an abnormal operation in which the communication disabled state has occurred in 1 to 18 and a normal operation in which the communication disabled state has not occurred in any of the servers 11 to 18.

The normal operation will be described with reference to FIG. 1, and the abnormal operation will be described with reference to FIGS.

The operation sequence of FIG. 2 shows a connection changing procedure (procedure for changing the parent) by the P2P server, and S10 to
It is composed of each step of S24. Further, the operation sequence of FIG. 3 shows a connection change procedure by the index proxy, and includes steps S30 to S41.

(A-2) Operation of the Embodiment (A-2-1) Normal Operation During normal operation, all the servers 11 to 18 shown in FIG. 1 should periodically transmit the above update request signal. The operation of updating the registration of itself in the lookup server 19 is repeated, and along with this operation, a content delivery operation (obtaining content from the client side) operation, a content update operation, a content operation in response to a request from the client Add (register new content) operation, delete content operation, etc. In any of these operations, the Web server described above and the DNS system equipped with the load balancing function are interposed between the client and the content distribution management system 10.

Of these, the outline of the content acquisition operation has already been described. For example, the content acquisition command uses the content name as an argument.
5 is supplied, the operation in that case is as follows.

Content information I whose content name is already stored in the P2P server 15 in the state shown in FIG.
If it corresponds to F2 and the content GF2, the P2P server 15 sends the content GF2 to the Web.
Content can be delivered to clients simply by returning it to the server. At this time, the content information I
Since it is not necessary to change F2, other servers 11-18
Access is not generated.

However, the relevant content is not stored in the own P2P server 15 and other P2P servers (for example,
16 etc.) (eg GF1)
If it refers to, a different operation is performed.

Even if the main body of the content GF1 is not stored, if the content information IF1 is cached in the P2P server 15, the P2P server 16 is immediately accessed based on the content information and the content GF1. However, since the content information IF3 is not currently stored in the P2P server 15, it is necessary to execute the content information search for the content information IF3. Occurs.

The content information search is executed using the tree.

The P2P server 15 first accesses (inquires) the index proxy 12, which is its own parent, for content information retrieval. If the index proxy 12 caches the content information IF1, the content information I from the index proxy 12
By receiving the supply of F1, the content information search is completed.

However, since the index proxy 12 does not cache the content information IF1 in the state of FIG. 1, a secondary inquiry is made from the index proxy 12 to its parent. In FIG. 1, since the parent of the index proxy 12 is the index server 11 that stores all content information, the content information IF1 is supplied to the index proxy 12 by this secondary inquiry, and the index proxy 12 continues to send P2P. The content information IF1 is supplied to the server 15. By this supply, the content information search by the P2P server 15 is completed.

The processing after the completion of the content information search is the same as when the content information IF1 is cached in the P2P server 15 from the beginning.

As a side effect of executing the content information search, the index proxy 12 and the P2P
The server 15 receives the content information IF1 supplied here.
The cached content information IF1 can be used when a content acquisition command for the same content information GF1 is supplied immediately thereafter.

It should be noted that the P2P server 15 is the other server 11
In order to execute the above-mentioned access to 14 to 16 and 16 to 18, assuming that each server 11 to 18 is arranged on the Internet, it is necessary to know its IP address and the like.

I of the index proxy 12 corresponding to the parent
For the P address, etc., the index proxy 1
Although it may be possible to register it in the P2P server 15 itself in advance when it is determined as 2, the other P2P servers 16 to
Since 18 is highly likely to be dynamically added or deleted, it is generally difficult to statically acquire the IP address in advance in a wide area and when the scale becomes large.

The service ID in which the content is stored is directly stored in the content information. The lookup server 19 has a function as a kind of name server that returns a service ID corresponding to content information and a communication means (proxy) for the service.

The basic processing for the normal operation other than the content distribution operation is the same as that for the content distribution operation.

However, in the case of the content update operation, the update date and size of the content information are usually changed, so that the same content information stored in each of the servers 11 to 18 in the system 10 (for example, IF
It is necessary to change the update date and size of 1) to ensure the consistency of content information.

In addition, in the content adding operation, only the content that did not exist in the system 10 up to that point is registered. For example, if the content GF2 is newly registered in the P2P server 15, the index server Registration of the content information IF1 in 11 is performed via the index proxy 12. A cache of the content information IF1 is left in the index proxy 12. As a result, the P2P server 15 does not need to be aware of the index server 11, and only needs to communicate with the index proxy 12, which is its parent, among the servers 11 to 14. The index proxy 12 completes the process by registering the content information IF1 in the index server 11.

Further, in the content deleting operation, basically the operation opposite to the content adding operation is performed.

However, in FIG. 1, for example, when the original content GF1 stored in the P2P server 16 is deleted, the content information IF3 is deleted.
Is deleted not only in the P2P server 16 itself, the index proxy 13 that is the parent of the P2P server 16 and the index server 11 (that is, each server corresponding to the content name), but also a server having a cache of the content ( For example, it is necessary to delete even 14 and 18).

Further, the deletion of the content body GF1 needs to be executed not only by the P2P server 16 which stores the original, but also by the other cached P2P server (for example, 18). The method of knowing the P2P server at the cache destination is the same as in the case of the content update operation.

By the above operation, the copy content for the original can be easily copied to another P2P server as necessary, and the original content and the copy content can be easily deleted. Further, it is possible to easily and flexibly add and delete the index proxy, the P2P server, etc. at an arbitrary position in the system as needed. As a result, the client can easily and surely access the target content regardless of which P2P server is distributed.

Since the target content can be easily and surely accessed, even if a part of the P2P servers are in the incommunicable state, the entire system 10 does not go down and continuous service is provided. Will be possible.

Further, since the P2P server in the incommunicable state can be automatically deleted from the search target, it is possible to search only the information of the P2P server which is currently activated, and the efficient access is possible. become.

On the other hand, the operation at the time of abnormality is as follows.
The operation at the time of abnormality is that the lookup server 1 does not receive the update request signal from any of the servers 11 to 18.
It is executed when the registration is deleted in 9.

(A-2-2) Operation at Abnormality If the update request signal is a signal requesting the continuation of the contract concluded between each of the servers 11 to 18 and the lookup server 19, the deletion is performed. The contract will expire.

It is the above-mentioned server communication inability notification (contract expiration notification) that informs the server in the other system 10 of this contract expiration, and the server communication inability notification is transmitted to the processing of steps S11 to S14 in FIG. This corresponds to the processing of steps S31 to S33 in FIG.

In both FIGS. 2 and 3, the index proxy 13 shown in FIG. 1 is in the incommunicable state, and the server incommunicability notification notifying the fact is the lookup server 1.
9 shows a connection change operation when the connection is transmitted from the communication terminal 9.

Generally, each of the servers 11 to 18 is a server which is a child of itself or the index server 11 in the tree.
It is not necessary to perform the connection changing operation even when the server that is in the inaccessible state, and the connection changing operation is executed only when the server that is the parent of itself is in the incommunicable state.

Therefore, it is necessary to perform the connection changing operation in the case where the index proxy 13 is in the communication disabled state, because the P2P servers 16 and 17 having the index proxy 13 as a parent and the index proxy 14
And other servers 11, 12, 15, 18
Does not need to perform any operation upon receiving the server communication failure notification.

Note that even if the server is a higher-level server on the path, it does not need to perform any operation unless it is the parent. For example, the index proxy 13 is higher than the P2P server 18 on the path from the root to the leaf, but is not the parent (the parent of the parent), so the P2P server 18
The server itself does not need to take any action in response to this server communication failure notification.

In FIG. 3, the P2P server 16 (or 1) that performs a connection change operation in response to the server communication failure notification is sent.
7) issues a search request to the lookup server 19 and receives communication means to the index proxy and index server existing in the system 10 (S34, S).
35). In the example of FIG. 1, the index proxy 12 and
It is a communication means to the index server 11.

The example of FIG. 3 corresponds to the inclusion case described here, and the P2P server 16 is the index server 1
1, the reply request signal is transmitted to the index proxies 12 and 14 (S17, S19, S21).
The operation of the P2P server 17 is basically the same as that of the P2P server 16.

The reply request signal is sent to each destination server.
This is a signal for inquiring the current load status of each server.

Therefore, the reply signal transmitted by each server in response to the reply request signal contains the load information indicating the load status of each server (S18, S19, S19).
22). Note that in FIG. 2, each server (for example, 12)
Although the reply request signal S19 to the next server (for example, 14) is shown to be transmitted after the reply signal S18 from the server is received, in reality, all the servers 12, 14, 11 that are the destinations are shown. On the other hand, it is efficient that the reply request signal is transmitted simultaneously or continuously and then the reply signal from each server is sequentially returned.

Since the current load status of each server 12, 14, 11 can be known from the load information contained in the reply signal, a server having a large load and a small processing capacity can be connected to a new connection destination (new You can choose not to choose as a parent. Therefore, each server needs to have a function of measuring its own load.

Also, the time from the transmission of the reply request signal to the reception of the corresponding reply signal (reply response time)
Is the congestion state of the transmission path interposed between the server that is the source of the reply request signal and the server that is the destination, the number of hops,
Physical propagation distance or each server 12, 14, 11
Since it is useful information for estimating the objective processing capacity of the above, it is used together with the load information when selecting a new connection destination. As a matter of course, the shorter the reply response time, the better as a new connection destination.

In the example of FIG. 2, the P2P server 16 is
The index proxy 12 is selected as a new connection destination based on the load information and the reply response time (S2).
3, S24).

Since the reply response time differs depending on the relative positional relationship between the source server and the destination server of the reply request signal on the network, the P2
The P2P server 17, which searches for a new parent in a similar procedure at substantially the same time as the P server 16, does not always select the same index proxy 12 as the P2P server 16 as a new parent. For example, the P2P server 17 may use the index server 11 as a new parent.

The connection changing operation executed by the index proxy 14 according to FIG. 3 is also performed by the P2P server 16,
This is the same as in the case of 17.

In FIG. 3, S34 is the above step S1.
5, S35 corresponds to step S16, and S35
36 to S39 correspond to the steps S17 to S22,
S40 corresponds to the step S23, and S41 corresponds to the step S24.

As a result of the server 16, 17, 14 executing the above connection changing operation of FIGS. 2 and 3, the tree structure of FIG. 1 changes as shown in FIG.

In FIG. 4, the P2P server 16 has the index proxy 12 as a parent, the P2P server 17 has the index server 11 as a parent, and the index proxy 1
4 has the index server 11 as a parent.

(A-3) Effects of the Embodiment As described above, according to this embodiment, the operating cost of the system is low, the flexibility regarding the system construction and the configuration change is high, and the reliability of the system is high. high.

Moreover, in this embodiment, the availability can be improved by the fault-tolerant configuration in which the connection is changed to change the parent when the parent is abnormal.

(B) Other Embodiments The tree in the above embodiment is a so-called binary tree, A
It is possible to use a VL tree, a B tree, or the like.

The shortest time for content information retrieval in the above embodiment is P when a content acquisition command or the like is received.
It corresponds to the case where the 2P server itself stores and caches the content information, but the worst time (maximum time)
Has a high possibility of dealing with the case where the above path is searched for and the index server 11 which is the highest level is reached. Therefore, the worst time can be shortened by reducing the height of the tree (for example, the height of the tree is 3 in the state of FIG. 1) as a whole, and the height differences of the partial trees are equalized. Therefore, the worst time can be equalized.

Further, although the logical structure in the above embodiment is a tree structure in which each node (each server) has only one parent, a network structure in which each node can have one or more parents is used. It is also possible. In the case of a network, if any one of the plurality of parents is normal, it is not necessary to execute the connection changing operation, and thus more stable system operation becomes possible.

Although the present invention has been mainly implemented by software in the above description, the present invention can also be implemented by hardware.

[0120]

As described above, according to the content distribution management system of the present invention, the operating cost of the system is low, the flexibility regarding the system construction and the configuration change is high, and the system reliability is high. .

In addition, according to the present invention, availability can be improved by the fault-tolerant configuration in which the connection is changed to change the parent when the parent is abnormal.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing an example of the overall configuration of a content distribution management system according to an embodiment.

FIG. 2 is an operation explanatory diagram when a connection is changed in the embodiment.

FIG. 3 is an operation explanatory diagram when a connection is changed in the embodiment.

FIG. 4 is a schematic diagram showing an example of the overall configuration of the content distribution management system according to the embodiment after connection change.

FIG. 5 is a schematic diagram showing a configuration example of a main part of a P2P server used in the embodiment.

FIG. 6 is a schematic diagram showing a configuration example of main parts of an index proxy and an index server used in the embodiment.

[Explanation of symbols]

10 ... Content distribution management system, 11 ... Index server, 12-14 ... Index proxy, 15-
18 ... P2P server, 19 ... Lookup server, GF
1, GF2 ... content.

Claims (3)

[Claims] 1. A content distribution management system for providing content searched using a predetermined logical structure on a widely distributed communication network, wherein the logical structure is the lowest node having no children,
The contents and at least contents identification information including position information indicating the position of a node on the logical structure in which the contents are accumulated are accumulated, and in response to an inquiry from the outside of the logical structure, to the upper node of the inquiry. A content storage server that relays or distributes the content, and is a top-level node having no parent in the logical structure,
A highest-level identification information server that stores content identification information regarding all the content stored by the content storage server existing on the logical structure and returns the content identification information in response to an inquiry from a lower node; This is a middle-level node having a parent and a child in the logical structure, stores at least a part of the content identification information accumulated by the highest identification information server, and is an inquiry about the accumulated content identification information. As long as the content identification information is returned, it functions as a proxy for the highest-level identification information server with respect to lower nodes that are its own children, and relays inquiries about the content identification information that has not been accumulated to the upper node. A proxy identification information server, and the highest-level identification information server or proxy identification information server A normality monitoring server that monitors the normality of the operation related to the server and notifies the node in the logical structure of that when a monitoring result indicating an abnormality is output, and the node that receives the notification is A content distribution management system characterized by performing a connection changing operation for changing a parent when the child itself is a child of an abnormal node in the notification. 2. The content distribution management system according to claim 1, wherein the proxy identification information server and the content storage server include one of the highest-level identification information server and the proxy identification information server when executing the connection change operation. First response request means for transmitting a response request signal for requesting a response of the margin information indicating the margin of the processing capacity of each server to the plurality of servers, and the margin information arrived in response to the response request signal as a selection criterion. And a first connection changing means for selecting a new parent of the content distribution management system. 3. The content distribution management system according to claim 2, wherein the proxy identification information server and the content storage server include one of the highest-level identification information server and the proxy identification information server when performing the connection change operation. Second response request means for transmitting a response request signal requesting a response of predetermined response information to a plurality of servers, and detecting an elapsed time until the response information arrives from each server according to the response request signal. A content distribution management system, comprising: an elapsed time detecting means for performing a change and a second connection changing means for selecting a new parent of itself based on the elapsed time as a selection criterion.