JP5399276B2 - Content distribution system and method and program - Google Patents

Description

  The present invention relates to a technique suitable for improving IP network utilization efficiency and user communication quality when performing content distribution such as VoD (Video on Demand) using an IP (Internet Protocol) network. is there.

  The Internet, which represents an IP network, has been developed and spread so as to be able to accommodate various applications. On the other hand, there is an increasing demand for QoS (Quality of Service).

  For example, services for distributing moving images over an IP network are also being developed. There are two main types of video distribution. One is a form in which a predetermined program or live broadcast is distributed at a predetermined time, and the other is a form of VoD in which a moving image is distributed in response to a request from a user.

  As for the former, by using IP multicast distribution to simultaneously distribute the same content, the same content need not be distributed over the same link.

  The latter VoD is normally distributed by IP unicast communication because the timing of content distribution requests varies from user to user.

  However, as described in Non-Patent Document 1, there is a bias in popularity of VoD content, and when content is biased in popularity, content with high popularity is distributed from the original server many times. As a result, the IP link is used inefficiently.

  In addition, when the network is congested due to such redundant traffic transfer, it takes time for the user to receive the content, and the quality may deteriorate.

H. Yu et al., “Understanding User Behavior in Large-Scale Video-on-Demand Systems,” ACM EuroSys 06. [Search January 8, 2010], Internet <URL: http: //www.cs. kuleuven.ac.be/conference/EuroSys2006/papers/p333-yu.pdf>.

  The problem to be solved is that in the conventional technology, highly popular content is distributed many times from the original server, and the IP link is used inefficiently.

  The object of the present invention is to solve these problems of the prior art and to avoid both congestion and quality degradation of the IP network associated with the distribution of popular content.

  In order to achieve the above object, in the present invention, even in content distribution such as VoD, a request is made to a cache server installed in a geographically dispersed place on an IP network using IP multicast or overlay network technology. A large amount of content is distributed in advance, and a request for content already distributed to the cache server is distributed from the cache server.

  According to the present invention, it is possible to improve both utilization efficiency and user quality in an IP network.

It is a block diagram which shows the network structural example which provided the content delivery system which concerns on this invention. It is a block diagram which shows the structural example of the overlay network used with the content delivery system which concerns on this invention. It is a block diagram which shows the structural example of the original server apparatus in FIG. It is a block diagram which shows the structural example of the cache server apparatus in FIG. It is a sequence diagram which shows the processing operation example of the content delivery system which concerns on this invention.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. In FIG. 1, 1-4 are IP routers, 10 is an original server device (in the figure and hereinafter referred to as "original server"), 11, and 12 are cache server devices (in the figure and hereinafter referred to as "cache server"), 13a ˜13d is a user terminal device (in the figure and hereinafter referred to as “user terminal”), and 14-16 are links. It is assumed that the user terminals 13a to 13d access the nearest cache servers 11 and 12.

  Each of the original server 10 and the cache servers 11 and 12 and the user terminals 13a to 13d in FIG. 1 has a computer configuration including a CPU (Central Processing Unit), a main memory, a display device, an input device, an external storage device, and the like. Each processing unit is installed by installing a program or data recorded in a storage medium such as a CD-ROM via an optical disk drive or the like in the external storage device, then reading the program or data from the external storage device into the main memory and processing it by the CPU. Perform the function.

  For example, as shown as the original server 31 in FIG. 3, the original server 10 includes a content distribution unit 31a and a content management unit 31b as means for executing programmed computer processing. 4, a distribution request transmission / reception unit 41a, a content distribution unit 41b, a content list holding unit 41c, and a content cache unit 41d are provided as means for executing programmed computer processing.

  The original server 31 having the configuration shown in FIG. 3 holds and manages all the content provided by the content distribution system of this example in the content management unit 31b.

  For example, if there is a content distribution request with the content name X from any cache server existing on the network, the content X is extracted from the content management unit 31b and passed to the content distribution unit 31a. The content distribution unit 31a distributes the corresponding content X by IP multicast. Each cache server on the network is joined to the IP multicast distribution.

  When the distribution request for content X is received from the user terminal at the distribution request transmission / reception unit 41a, the cache server 41 having the configuration shown in FIG. 4 inquires of the content list holding unit 41c and checks whether the content X is held.

  If held, the content cache unit 41d is instructed to transmit the content X to the content distribution unit 41b. The content cache unit 41d reads the content X and transfers it to the content distribution unit 41b. The content distribution unit 41b distributes the content X to the requesting user terminal.

  If it is confirmed that the content holding list unit 41c does not hold the content X, the distribution request transmitting / receiving unit 41a is notified of this fact. The distribution request transmission / reception unit 41a requests the original server to distribute the content X.

  When the content X is distributed from the original server in response to the request, the content X is received by the content distribution unit 41b, and at the same time, the content X is distributed to the user terminal. Also, the received content X is held (cached) in the storage device in the content cache unit 41d.

  The content cache unit 41d manages the time T_k at which the content request was last made for each content k existing in the cache.

  If the content X distributed by IP multicast distribution from the original server is not cached in the cache server, the content X is cached. If the value obtained by adding the content X exceeds the cache capacity, the content k having the oldest T_k is deleted and the content X is cached.

  When the cached content is exchanged as described above, the fact is notified to the content list holding unit 41c.

  Hereinafter, <first embodiment> of the content distribution system of this example having such a configuration will be described using the configuration of FIG. 1 as an example.

  In the content distribution system shown in FIG. 1, the content requested by the user terminals 13a to 13d via the IP network is sent to the original server 10 holding the content or the cache server closest to the requesting user terminals 13a to 13d. 11 and 12, when the original server 10 receives a request for content from the user terminals 13a to 13d, the original server 10 sends the content to the requesting user terminals 13a to 13d and the cache servers 11 and 12, respectively. And distributed by IP multicast.

  When each of the cache servers 11 and 12 receives a request for content from the user terminals 13a to 13d, the cache server 11 or 12 checks whether or not the content is held, and if so, the user terminal device 13a that is the request source. To the original server 10, the request for the content from the user terminal devices 13a to 13d is transferred to the original server 10, and the content that has been IP multicast distributed from the original server device is transferred to the original server 10. Hold in storage.

  For example, it is assumed that there is a request for content X for the first time from the user terminal 13a under the IP router 3. At that time, the content X is distributed from the original server 10 to the cache servers A11 and B12 by IP multicast and also to the user terminal 13a.

  Next, when there is a request for content X from another user 13b under the control of the IP router 3, the content is distributed from the cache server A11. Next, when there is a request for content X from the user terminals 13c and 13d under the IP router 4, the content X is distributed from the cache server B12 to each user terminal 13c and 13d.

  At this point, in the content distribution technology of this example, the content X only needs to pass through each of the links 14, 15, and 16 once, whereas when IP unicast is used, the content X is The link 14 passes four times, the link 15 passes twice, and the link 16 passes twice. According to the content distribution technique of this example, such redundant transfer can be reduced.

  Furthermore, if there is a request for the content X from another user terminal thereafter, it becomes more efficient than unicast. In addition, since traffic can be reduced, even when congestion occurs in unicast, congestion can be avoided and the quality of the user can be improved.

  Furthermore, since the user receives content from a nearby cache, the propagation delay can be shortened in the first place.

  Next, a <second embodiment> of the content distribution system of this example will be described.

  In the content distribution system in FIG. 1, a cache server that is the closest to the original server 10 holding the content or the requesting user terminals 13a to 13d for the content requested from the user terminals 13a to 13d via the IP network. For example, when the original server 10 receives a request for content from the user terminal 13a, the original server 10 distributes the content to the requesting user terminal 13a by IP unicast.

  When the router 3 directly connected to the cache server 11 by the routers 1, 2 and 3 on the IP unicast distribution route of the content receives the request for the content from the user terminals 13a and 13b, the directly connected cache server 11 confirms whether or not the content is held, and if held, the content must be distributed and held by IP unicast from the directly connected cache server 11 to the requesting user terminals 13a and 13b. For example, the request for the content from the user terminals 13a and 13b is transferred to the original server 10, and the content distributed from the original server 10 in response to the transferred request is transferred to the directly connected cache server 11. Hold.

  Also, when the router having a cache function in the IP routers 1, 2 and 3 on the content IP unicast distribution path receives the request for the content from the user terminals 13a and 13b, whether or not the content is held. If the content is held, the content is distributed to the requesting user terminals 13a, 13b by IP unicast. If not, the content is requested from the user terminals 13a, 13b to the original server 10. In addition to the transfer, the content distributed from the original server 10 is held in the storage device in response to the transferred request.

  Thus, in the second embodiment, instead of distributing content to the cache server by multicast distribution as in the first embodiment, when a request for content X occurs from a user, the cache closest to the user is cached. Whether or not the content X exists in the server is confirmed. If there is no content X, the content X is received from the original server.

  At that time, the data is transferred to the user using IP unicast distribution. At the same time, if there is a router directly connected to the cache server or a router having a cache function among the routers on the distribution route, the cache server or cache Content X is cached in the server / router.

  Next, when there is a request for distribution of the content X from another user, the content X is distributed by IP unicast from the cache server nearest to the user.

  This processing includes a function of analyzing whether a packet is a packet constituting a content to be cached and a function of caching or a function of transferring the corresponding packet to a cache server in some IP routers. It is feasible. This does not require that all IP routers be replaced, and may be a partial arrangement.

  Next, a <third embodiment> of the content distribution system of this example will be described.

  In the third embodiment, for example, the first known document (“L. Zhi and P. Mohapatra,“ QRON: QoS-aware routing in overlay net-works, ”IEEE J. Select. Areas is used. Commun., Vol.22, pp.29-40, January 2004. <http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1258113&isnumber=28139> and second public literature ("WIDE Project, "Integrated Distributed Environment with Overlay Network" WIDE Project Research Report, Part 17, 2002. <http://203.178.136.57/project/document/reports/pdf2002/part17.pdf> ")) Is used.

  An overlay network is a network in which a logical (virtual) link is formed according to the purpose in an upper layer using an existing link, and is configured by using such an overlay network. Explain the basic concept of QoS management.

  In FIG. 2, 21 a to 21 c are overlay nodes constituting the overlay network (described as “Overlay-NW”) 22, and 23 a to 23 e configure the IP network (described as “IP-NW” in the figure) 24. IP router.

  It is assumed that traffic is flowing from x to y along a path indicated by a broken-line arrow. Further, it is assumed that there is an IP router that is congested 25 on this route, and as a result, the QoS between x and y is lowered.

  At this time, by using the overlay network 22 formed by the overlay nodes 21a, 21b, and 21c, the traffic is detoured along a route (x → overlay node 21a → overlay node 21b → overlay node 21c → y) represented by a practical arrow. If possible, the above congestion 25 can be avoided.

  In the third embodiment using such an overlay network 22, the content requested to be distributed from the user terminals 13 a to 13 d via the IP network is held in the network having the content distribution system configuration in FIG. 1. When the original server 10 receives a request for content from the user terminals 13a to 13d in order to distribute from the nearest cache server 11 or 12 to the original server 10 or the requesting user terminals 13a to 13d, Then, it is delivered to the requesting user terminals 13a to 13d and the cache servers 11 and 12 by IP unicast via an overlay network which is a logical network constructed by overlay nodes.

  Each of the cache servers 11 and 12 receives a request for content from the user terminals 13a to 13d, and checks whether or not the content is held. If not, the original server 10 sends the user terminals 13a to 13d. 2 and the content distributed from the original server 10 in accordance with the transferred request is held in the storage device, and if held, the content is transferred to the overlay node shown in FIG. The data is distributed by IP unicast to the requesting user terminals 13a to 13d via the overlay network 22 which is a logical network constructed by 21a to 21c.

  Further, the cache servers 11 and 12 that construct the overlay network 22 as the overlay nodes 21a to 21c hold the content distributed via the overlay network in the storage device and receive the content request from the user terminals 13a to 13d. The content is confirmed whether it is held, and if held, the content is distributed to the requesting user terminals 13a to 13d via the overlay network 22 by IP unicast, and if not held, the original The content request from the user terminals 13a to 13d is transferred to the server 10, and the content distributed from the original server 10 in response to the transferred request is held in the storage device.

  By doing so, even if congestion occurs on the IP path between the cache servers 11 and 12 and the user terminals 13a to 13d, the content can be delivered while avoiding the congestion. At the same time, the content X can be distributed to the entire network by caching the content in the overlay nodes 21a to 21c that relay the content.

  As described above, in the third embodiment, instead of distributing content to the cache server by multicast distribution as in the first embodiment, when a request for content X is generated from the user, the cache closest to the user is cached. Whether or not the content X exists in the server is confirmed. If there is no content X, the content X is received from the original server.

  If there is content X in the closest cache server, the content X is received from the cache server. At that time, the content X is transferred from the cache server to the user by using an overlay network which is a logical network constructed by several overlay nodes connected to the IP network.

  Specifically, instead of delivering directly from the cache server to the user by IP unicast, one or a plurality of overlay nodes are selected from the overlay nodes constituting the overlay network, and the cache is passed through the overlay nodes. The content X is transferred from the server to the user.

  At this time, if there is an overlay node having a cache function among the overlay nodes on the distribution route, the content X is cached in the cache server / overlay node. Next, when there is a request for distribution of the content X from another user, the nearest one is selected from the cache servers, and the content X is distributed by IP unicast.

  Next, the <fourth embodiment> of the content distribution system of this example will be described.

  In the fourth embodiment, the content to be deleted from the cache server is determined based on the last request time, and the cache servers 11 and 12 request each time the content held in the storage device is requested. When the registration time, which is the received time, is updated and new content is stored in the storage device exceeding the predetermined cache capacity, the content with the oldest registration time is deleted.

  That is, each cache server manages the time T_k at which the content request was last made for each content k existing in the cache. If the distributed content X is not cached in the cache server, the content X is to be cached, but the content X is added to the number of already cached content or the total content size. If the calculated value exceeds the cache capacity, the content k having the oldest T_k is deleted and the content X is cached.

  Next, the <fifth embodiment> of the content distribution system of this example will be described.

  In the fifth embodiment, instead of managing the time when the content request was last, as in the fourth embodiment, the new time of either the content request or the multicast distribution is managed.

  That is, each time the content is held (cached) in the storage device, each of the cache servers 11 and 12 updates the registration time, which is the time of receiving the delivery, every time a request is received or delivered, and new content is created. Is stored (cached) in a storage device exceeding a predetermined cache capacity, the content with the oldest registration time is deleted.

  In some cache server A, there is a case where the content X is distributed by multicast even if the content X is not requested. This means that the content requested by another cache server has been distributed.

  In the fifth embodiment, such a case is also used as time information used at the time of cache replacement. This is equivalent to making the content X remain in the cache even if the cache server A itself does not request it, and is effective when there is a high probability that there will be a request from a user under the cache server A in the future. is there.

  Next, <Sixth embodiment> of the content distribution system of this example will be described.

  In the sixth embodiment, the registration times of all cache server devices are updated all at once, and the cache servers 11 and 12 notify the updated registration times and the identification information of the contents to other cache servers. Then, the registration time of the content notified of the identification information from the other cache server device is updated with the notified registration time.

  In this way, in the sixth embodiment, instead of each cache server independently caching content as in the fourth embodiment, if there is a request for content X in any cache server, The process of updating the time when the request for the content was last is executed not only in the cache server but also in all the cache servers.

  This corresponds to synchronizing the caches of all the cache servers. By doing so, for example, the cache status of the entire system can be grasped without managing each cache status individually. .

  Next, <seventh embodiment> of the content distribution system of this example will be described.

  In the seventh embodiment, instead of performing cache replacement using the time at which the content request was last made, as in the fourth embodiment, the past content access frequency is counted and a new When the content is cached, the old new access frequency is deleted and the corresponding new content is cached.

  That is, the cache servers 11 and 12 count the past access frequency for each content held (cached) in the storage device, and hold new content in the storage device exceeding a predetermined cache capacity. Delete past infrequently accessed content.

  Next, <Eighth embodiment> of the content distribution system of this example will be described.

  In the eighth embodiment, as in the first to seventh embodiments, what content is cached in which cache server instead of accessing the nearest cache server at the time of a request from the user terminal If there is a central control station (management server) that knows the contents and a request for content X is received from the user terminal, the central control station (management server) distributes it from the cache server or original server that holds the content X Select the original server.

  As a selection criterion, a server having the closest distance to the user terminal (the number of hops in the IP network) is selected. Then, the content is distributed from there. At that time, the content X is cached in the nearest cache server to the user and distributed to the user.

  That is, in the management server device that accepts content distribution requests from all user terminals, content identification information held by the original server 10 and each of the cache servers 11 and 12 is stored, and the contents requested from the user terminals 13a to 13d Cache servers 11 and 12 that hold the original server 10 and the original server 10 are selected, and the servers 10 to 12 having the smallest number of hops to the requesting user terminals 13a to 13d are selected and selected from the specified servers 10 to 12 The nearest cache server that forwards the content distribution request from the user terminals 13a to 13d to the servers 10 to 12 and does not hold the content with a smaller number of hops to the requesting user terminals 13a to 13d than the selected server If there is, the nearest cache from the selected server To over server, to deliver content that is delivered to the requesting user terminal 13a~13d and is cached.

  Next, <Ninth embodiment> of the content distribution system of this example will be described.

  In the ninth embodiment, as in the eighth embodiment, when the distribution source server is selected from the cache server or the original server that holds the content X, instead of the number of hops in the IP network, the server The server that minimizes the maximum value of the link usage rate of each IP link existing between the user and the user, or the quality (delay time and packet loss rate) between the user and the server is measured and minimized. Select a server.

  That is, in the management server device, for each of the original server 10 and the cache servers 11 and 12, identification information of contents held by the servers 10 to 12 is stored, and contents requested from the user terminals 13a to 13d are held. The cache servers 11 and 12 and the original server 10 are specified, and the communication efficiency with the requesting user terminals 13a to 13d, for example, the link usage rate, the delay time, and the packet loss rate is the highest among the specified servers 10 to 12. A good server is selected, the content distribution request from the user terminals 13a to 13d is transferred to the selected server, the content is distributed from the selected server to the requesting user terminals 13a to 13d by IP unicast, and the route at the same time If there is a router with a cache function above, the content being distributed To cache.

  Next, <Tenth Embodiment> of the content distribution system of this example will be described.

  In the tenth embodiment, when using the overlay network in the third embodiment, for example, an overlay node that minimizes the maximum value of each link usage rate of an IP link, or quality (delay time and packet). Loss rate) is measured in advance, and an overlay node that minimizes it is selected to select an efficient overlay network.

  That is, in the management server that accepts content distribution requests from all the user terminals 13a to 13d, content identification information held by the servers 10 to 12 is stored for each of the original server 10 and the cache servers 11 and 12. The cache servers 11 and 12 holding the content requested from the user terminals 13a to 13d and the original server 10 are specified, and the communication efficiency with the requesting user terminals 13a to 13d from the specified servers 10 to 12 ( A server having the best link usage rate, delay time, and packet loss rate is selected, the content distribution request from the user terminals 13a to 13d is transferred to the selected server, and communication between the selected server and the requesting user terminal is performed. Overlay network consisting of overlay nodes with the highest efficiency When the content is delivered via the overlay network, the cache server device serving as the overlay node existing on the route is notified that the selected server delivers the content via the network. If there is a cache server device that does not hold the content, the distributed content is cached.

  Next, the content distribution processing operation according to the present invention will be described with reference to FIG. In this example, a management server is used.

  As shown in FIG. 5, the user terminal 13 inquires of the management server 51 about the nearest cache server 41 (step S501). The management server 51 identifies the cache server 41 that minimizes the number of hops with the user terminal 13, for example, and notifies the user terminal 13 (step S502).

  Upon receiving the notification, the user terminal 13 accesses the notified cache server 41 and requests distribution of the content (step S503). Upon receiving the request, the cache server 41 checks whether or not the content is held (cached). If the content is held, the content is read and returned to the user terminal 13 (step S504). Then, the content distribution request from the user terminal 13 is transferred to the original server 31 (step S505).

  When the original server 31 receives the content distribution request from the user terminal 13 transferred from the cache server 41, the original server 31 distributes the content to the cache server 41 and the user terminal 13 by IP multicast (steps S506 and S507). In 41, the content distributed from the original server 31 is transferred to the user terminal 13 and held (cached) in the storage device.

  As described above with reference to FIGS. 1 to 5, in the content distribution system of this example, the requested content is IP multicast distributed from the original server in response to the request from the cache server.

  That is, it is assumed that an original server holding content is connected to the IP network, and M cache servers are connected. In a content distribution system constituted by these servers, a request for content X is received from a user terminal. When it occurs, it is confirmed whether or not the content X is in the cache server closest to this user. At that time, using IP multicast distribution, not only the user but also the M cache servers are simultaneously distributed, and the content X is cached. Next, when there is a request to distribute the content X from another user, the content X is distributed from the cache server nearest to the user by IP unicast.

  In addition, when unicast distribution is performed from the original server in response to a request from the user terminal, the content is cached in an arbitrary router having a cache control function on the route. That is, as described above, instead of distributing content from the original server to the cache server by IP multicast distribution, if a request for content X is generated from the user, whether or not there is content X in the cache server closest to this user If it is not confirmed, the content X is received from the original server. At that time, the data is transferred to the user using IP unicast distribution. At the same time, if there is a router directly connected to the cache server or a router having a cache function among the routers on the distribution route, this cache server or cache Content X is cached in the server / router. Next, when there is a request to distribute the content X from another user, the content X is distributed from the cache server nearest to the user by IP unicast.

  In addition, an overlay network is used when caching. That is, as described above, instead of distributing content to the cache server by multicast distribution, if a request for content X is generated from the user terminal, it is confirmed whether there is content X in the cache server closest to this user. Receives content X from the original server. If there is content X in the closest cache server, the content X is received from the cache server. At that time, the content X is transferred from the cache server to the user by using an overlay network which is a logical network constructed by several overlay nodes connected to the IP network. Specifically, instead of delivering directly from the cache server to the user by IP unicast, one or more overlay nodes are selected from the overlay nodes constituting the overlay network, and the cache is passed through the overlay nodes. The content X is transferred from the server to the user. At this time, if there is an overlay node having a cache function among the overlay nodes on the distribution route, the content X is cached in this cache server / overlay node. Next, when there is a request for distribution of the content X from another user, the closest one is selected from the cache servers, and the content X is distributed by IP unicast.

  The content to be deleted from the cache is determined as follows. That is, each cache server manages the time T_k at which the content request was last made for each content k existing in the cache. If the content X distributed by multicast distribution is not cached in the cache server, the content X is to be cached. However, the content X or the total size of the content X is already cached. If the value obtained by adding the minutes exceeds the cache capacity, the content k having the oldest T_k is deleted and the content X is cached.

  Alternatively, instead of managing the last time when the content request was made, the new time of either the content request or the multicast distribution is managed.

  Also, instead of caching each content independently, if there is a request for content X in any of the cache servers, a process of updating the time when the last request for that content was made This is performed not only on the cache server but also on all cache servers.

  In addition, as described above, instead of performing cache replacement using the time at which the last content request was made, past content access frequency is counted and new content is cached. Delete infrequent content and then cache the corresponding new content.

  In addition, instead of the request from the user terminal accessing the nearest cache server, there is a centralized control station (management server) that knows which content is cached in which cache server. If there is a request for content X, a distribution source server is selected from a cache server or an original server holding content X. The selection criterion at this time is to select a server that is closest to the user terminal (the number of hops in the IP network). Then, the content is distributed from there. At that time, the content X is cached in the nearest cache server to the user terminal and distributed to the user terminal.

When selecting the distribution source server from the cache server or the original server holding the content X, instead of the number of hops in the IP network, the link usage rate of each IP link existing between the server and the user terminal A server with the smallest maximum value may be selected.
Alternatively, the quality (delay time or packet loss rate) between the user and the server may be measured and the server that minimizes the quality may be selected.

  Also, when distributing content via the overlay node, select the overlay node that minimizes the maximum value of the link usage of each IP link that exists between the cache server and the overlay node and between the overlay node and the user terminal. May be. Alternatively, quality (delay time and packet loss rate) may be measured and an overlay node that minimizes the quality may be selected.

  As described above, according to the content distribution system of the present example, even in content distribution such as VoD, using IP multicast or overlay network technology, a cache server installed in geographically dispersed locations on the IP network can be used. By distributing the contents, it is possible to improve both the use efficiency of the IP network and the quality of the user.

  In addition, this invention is not limited to the example demonstrated using FIGS. 1-5, In the range which does not deviate from the summary, various changes are possible. For example, in FIG. 5, the management server is used, and the user terminal is configured to query the management server for the nearest cache server, but the user terminal is configured to recognize the nearest cache server in advance. It may be possible to request content distribution directly from the user terminal to the nearest cache server.

  In this example, only one original server 10 is provided in FIG. 1, but a plurality of original servers may be provided. Further, the cache servers 11 and 12 may be configured to have more cache servers.

  Also, the computer configuration example of this example may be a computer configuration without a keyboard or optical disk drive. In this example, an optical disk is used as a recording medium. However, an FD (Flexible Disk) or the like may be used as a recording medium. As for the program installation, the program may be downloaded and installed via a network via a communication device.

  1 to 4: IP router, 10, 31: original server device, 11, 12, 41: cache server device, 13, 13a to 13d: user terminal device, 14 to 16: link, 21a to 21c: overlay node, 22: Overlay network, 23a to 23e: IP router, 24: IP network, 25: congestion, 31a: content distribution unit, 31b: content management unit, 41a: distribution request transmission / reception unit, 41b: content distribution unit, 41c: content list holding unit , 41d: content cache unit, 51: management server.

Claims (15)

A content distribution system for distributing content requested from a user terminal device via an IP network from an original server device holding the content or a cache server device nearest to the requesting user terminal device,
The original server device is
Upon receiving a content request from the user terminal device, the content has means for IP unicast delivery to the requesting user terminal device;
Of the routers on the IP unicast delivery route of the content, the router directly connected to the cache server device is:
Upon receiving a request for content from the user terminal device, means for confirming whether or not the directly connected cache server device holds the content;
Means for delivering the content by IP unicast from the directly connected cache server device to the requesting user terminal device if held,
Means for transferring the content request from the user terminal device to the original server device if not held;
The content which is distributed from the original server device in response to request the transfer and means for holding is transferred to the cache server apparatus that is connected directly above,
Of the routers on the IP unicast distribution route of the above content, the router having a cache function is:
Upon receiving a request for content from the user terminal device, means for confirming whether or not the content is held;
Means for delivering the content to the requesting user terminal device by IP unicast if held;
Means for transferring the content request from the user terminal device to the original server device if not held;
Content distribution system characterized by having a means for holding the content which is distributed from the original server device in response to request the transfer to the storage device. The content distribution system according to claim 1,
  The router directly connected to the cache server device has means for analyzing whether or not the content distributed from the original server device should be cached by analyzing the packet of the content, and determined that the content should be cached. Transfer and hold the content to the directly connected cache server device,
  The router having the cache function includes means for analyzing whether or not the content distributed from the original server device should be cached by analyzing the packet of the content, and storing the content determined to be cached as a storage device Hold on
  A content distribution system characterized by that. A content distribution system for distributing content requested from a user terminal device via an IP network from an original server device holding the content or a cache server device nearest to the requesting user terminal device,
The original server device is
Upon receiving a content request from the user terminal device, the content is delivered by IP unicast to the requesting user terminal device via an overlay network, which is a logical network constructed by overlay nodes. Have
Each of the above cache server devices
Upon receiving a request for content from the user terminal device, means for confirming whether or not the content is held;
Means for transferring the content request from the user terminal device to the original server device if not held;
Means for holding the content distributed from the original server device in response to the transferred request in a storage device;
Means for delivering the content by IP unicast to the requesting user terminal device via an overlay network, which is a logical network constructed by overlay nodes, if held,
Among the overlay nodes on the content delivery path from the cache server device, the overlay node having a cache function is:
Means for holding in the storage device the content distributed via the overlay network;
Upon receiving a request for content from the user terminal device, means for confirming whether or not the content is held;
Means for delivering the content by IP unicast to the requesting user terminal device via the overlay network to which the content belongs, if held,
Means for transferring the content request from the user terminal device to the original server device if not held;
Content distribution system characterized by having a means for holding the content which is distributed from the original server device in response to request the transfer to the storage device. A content delivery system according to any one of claims 1 to 3,
The cache server device is
For each content held in the storage device, means for updating the registration time, which is the time when the request is received,
A content distribution system comprising: means for deleting a content having the oldest registration time when a new content is retained in a storage device exceeding a predetermined cache capacity. A content delivery system according to any one of claims 1 to 3,
The cache server device is
For each content held in the storage device, each time it is requested or distributed, a means for updating the registration time, which is the time of receiving the distribution,
A content distribution system comprising: means for deleting a content having the oldest registration time when a new content is retained in a storage device exceeding a predetermined cache capacity. A content distribution system according to any one of claims 4 and 5,
The cache server device is
Means for notifying the updated registration time and the identification information of the content to other cache server devices;
Means for updating the registration time of the content notified of the identification information from another cache server device with the notified registration time;
A content distribution system comprising: A content delivery system according to any one of claims 1 to 3,
The cache server device is
Means for counting the past access frequency for each content held in the storage device;
A content distribution system comprising: means for deleting the previously inaccessible content when the new content exceeds a predetermined cache capacity and is held in the storage device. The computer program for causing to function as each unit in a content distribution system as claimed in any one of claims 7. A content distribution method for distributing content requested from a user terminal device via an IP network from an original server device holding the content or a cache server device nearest to the requesting user terminal device,
The original server device is
Upon receiving the content request from the user terminal device, the content is IP unicasted to the requesting user terminal device,
Of the routers on the IP unicast delivery route of the content, the router directly connected to the cache server device is:
Upon receiving a content request from the user terminal device, it is confirmed whether or not the directly connected cache server device holds the content,
If held, the content is distributed from the directly connected cache server device to the requesting user terminal device by IP unicast,
If not, transfer the content request from the user terminal device to the original server device,
In response to the transferred request, the content distributed from the original server device is transferred to the directly connected cache server device and held,
Of the routers on the IP unicast distribution route of the above content, the router having a cache function is:
Upon receiving the content request from the user terminal device, confirm whether or not the content is held,
If so, the content is distributed to the requesting user terminal device by IP unicast,
If not, transfer the content request from the user terminal device to the original server device,
A content distribution method characterized by holding content distributed from the original server device in response to a transferred request in a storage device. The content distribution method according to claim 9,
  The router directly connected to the cache server device determines whether or not the content distributed from the original server device should be cached by analyzing the packet of the content, and determines the content determined to be cached as the direct cache. Transfer and hold to the connected cache server device,
  The router having the cache function determines whether or not the content distributed from the original server device should be cached by analyzing the packet of the content, and holds the content determined to be cached in the storage device.
  A content distribution method characterized by the above. A content distribution method for distributing content requested from a user terminal device via an IP network from an original server device holding the content or a cache server device nearest to the requesting user terminal device,
The original server device is
Upon receiving a content request from the user terminal device, the content is delivered to the requesting user terminal device via an overlay network, which is a logical network constructed by overlay nodes, by IP unicast,
Each of the above cache server devices
Upon receiving the content request from the user terminal device, confirm whether or not the content is held,
If not, transfer the content request from the user terminal device to the original server device,
The content distributed from the original server device in response to the transferred request is held in the storage device,
If held, the content is delivered by IP unicast to the requesting user terminal device via the overlay network, which is a logical network constructed by the overlay node,
Among the overlay nodes on the content delivery path from the cache server device, the overlay node having a cache function is:
Upon receiving a request for content from the top SL user terminal confirms whether holds the content,
If so, the content is delivered to the requesting user terminal device via the overlay network to which it belongs by IP unicast,
If not, transfer the content request from the user terminal device to the original server device,
Content distribution method characterized by holding the content which is distributed from the original server device in response to request the transfer to the storage device. A content distribution method according to claims 9 to any one of claims 11,
The cache server device is
For each content held in the storage device, each time it is requested, the registration time that is the time when the request is received is updated,
A content distribution method, wherein when new content is retained in a storage device exceeding a predetermined cache capacity, the content with the oldest registration time is deleted. A content distribution method according to claims 9 to any one of claims 11,
The cache server device is
For each content held in the storage device, every time it is requested or distributed, the registration time that is the time of receiving the distribution is updated,
A content distribution method, wherein when new content is retained in a storage device exceeding a predetermined cache capacity, the content with the oldest registration time is deleted. A content distribution method according to any one of claims 12 or claim 13,
The cache server device is
Notify the updated registration time and the identification information of the content to other cache server devices,
A content distribution method, comprising: updating a registration time of content notified of the identification information from another cache server device with the notified registration time. A content distribution method according to claims 9 to any one of claims 11,
The cache server device is
For each content held in the storage device, the past access frequency is counted,
A content distribution method, wherein when new content is stored in a storage device exceeding a predetermined cache capacity, the past content with low access frequency is deleted.

Images