WO2013047207A1 - Cache system, cache method, and cache server - Google Patents

Abstract

A cache system is provided with an origin server and a plurality of cache servers which delivers content requested from a terminal instead of the origin server. When the cache server does not hold the content requested from the terminal, and delivers the content to the terminal by acquiring the content from another cache server holding the content, the cache server controls duplication of the content on the basis of the index indicating a load caused by transmitting/receiving the content to/from the other cache server.

Description

Cache system, cache method, and cache server

The present invention relates to a cache system, a cache method, and a cache server. In particular, the present invention relates to a cache system, a cache method, and a cache server that deliver content requested from a terminal to a terminal from a cache server instead of an origin server.

FIG. 14 shows an example of a cache system. The cache system in this example includes a network 1, a network 2, a terminal 100, an origin server 200, and a cache server 300.

The cache server 300 is communicatively connected to the origin server 200 via the network 1. Further, the cache server 300 is communicatively connected to the terminal 100 via the network 2.

The origin server 200 distributes the content requested from the terminal 100 to the terminal 100. The cache server 300 mediates content distribution from the origin server 200 to the terminal 100. More specifically, the cache server 300 delivers the requested content instead of the origin server 200 when the content is requested by the terminal 100 by caching the content. Thereby, the load of the origin server 200 and the network 1 can be reduced. As a result, the cache server 300 is generally disposed at a location closer to the terminal 100 than the origin server 200, so that the delay in content distribution can also be shortened.

Further, the cache server 300 may hold the content to be distributed in advance in accordance with the business agreement with the owner of the origin server 200 or the content owner held in the origin server 200.

Fig. 15 shows a sequence diagram of the content distribution system. First, in step S <b> 100, the terminal 100 requests content from the origin server 200. Here, various protocols can be considered as a protocol for making a content request. For example, in the case of HTTP, the content request corresponds to HTTP GET.

There are several methods for causing the cache server 300 to mediate content requests made by the terminal 100 to the origin server 200. One is a method of setting the information of the cache server 300 as a WEB Proxy server in the terminal 100 and accessing the cache server 300 by the operation of the terminal 100.

Next, it is a method using a DNS (Domain Name System) server. The method using the DNS server is a method in which when the terminal 100 acquires the address of the origin server 200 using the DNS server, the DNS server responds with the address of the cache server 300 instead of the origin server 200. There are other methods, such as a method using HTTP redirection and a method in which the cache server 300 is arranged on a path through which the terminal 100 communicates with the origin server 200, but the details are omitted because they are not directly related to the present invention. .

When the cache server 300 receives the content request, the cache server 300 identifies the requested content based on information such as a URL (Uniform Resource Locator) included in the content request. Then, the cache server 300 checks whether or not the requested content is held, and if not, transfers the requested content request to the origin server 200 (this step holds the targeted content in this step). Not). Here, the content request from the terminal 100 may be transferred without being changed, or may be transferred after changing the address or the like.

Upon receiving the content request transferred from the cache server 300, the origin server 200 starts a process of distributing the requested content to the cache server 300 (S101).

When the cache server 300 receives the content started to be distributed from the origin server 200, in step S102, the cache server 300 distributes the content to the terminal 100 and stores the cache of the content (hereinafter also referred to as caching).

Thereafter, the terminal 100 receives the content distributed via the cache server 300.

Next, consider a case where the terminal 100 requests the same content. The terminal 100 here does not have to be the same terminal as the terminal 100 that requested the content for the first time. In this case, when receiving a content request from the terminal 100, the cache server 300 holds a cache of the content. Therefore, the content requested from the cache server 300 is distributed to the terminal 100 without transferring the content request to the origin server 200 (S103).

Next, FIG. 16 shows an example in which a plurality of cache servers shown in FIG. 14 are deployed in a distributed manner. That is, FIG. 16 is an example of a non-cooperative cache system in which the cache does not exchange caches between cache servers.

The cache servers 300a, 300b, and 300c, terminals 100a, 100b, and 100c, and networks 2a, 2b, and 2c in FIG. 16 indicate a situation where there are a plurality of cache servers 300, terminals 100, and networks 2 described in FIG. The networks 2a, 2b and 2c may be networks of different ISPs or network operators, or may be networks of the same ISP and network operators.

In FIG. 16, when the terminal 100a requests content, the cache is held in the cache server 300a. For the terminals 100b and 100c, caches are held in the cache server 300b and the cache server 300c, respectively. At this time, the cache operation is the same as in FIG.

In the cache system of this example, if the cache server 300a does not cache the content requested from the terminal 100a, the content is acquired from the origin server 200 even if there is a cache in another cache server.

In order to improve this, a cache system that can be linked between each cache server is considered. Specifically, if the content requested from the terminal is not cached in a certain cache server, does not immediately acquire the content from the origin server, but does the other cache server hold the requested content? Confirm whether or not. As a result, if there is a cache server holding the content, the cache is acquired from the cache server and distributed to the terminal. Hereinafter, this cooperation is referred to as inter-cache server cooperation.

FIG. 17 shows a cache system that can be linked between cache servers. In FIG. 17, the cache servers 300'a, 300'b, and 300'c are cache servers each having a cooperation function. The management server 400 provides content information held by other cache servers to each cache server 300 ′, and controls content to be held by each cache server.

Fig. 18 shows a graph showing the effect of cooperation between cache servers. FIG. 18 shows the difference between the hit rate of a single cache server and the hit rate when cooperation between cache servers is applied, assuming that the content request trend of the terminal 100 follows the Zipf distribution. The Zipf distribution is a distribution in which access concentrates on content with high popularity, and the access probability to the kth most popular content can be expressed by the following formula (1).

　

Here, Σ {N, n = 1} (1 / n ^ s) indicates the sum of (1 / n ^ s) when n is changed from 1 to N. In addition, s is a parameter that determines the degree of concentration of access to popular content, and there is a research report that s = 0.55 in the video distribution service, so here s = 0.55. .

In the graph of FIG. 18, the horizontal axis represents the number of cache servers that can be linked, and the vertical axis represents the hit rate of the entire system, that is, the hit rate obtained as a result of linkage. When the number of cache servers that can be linked is 0, the hit rate when a cache server without linkage is applied is shown.

Each line assumes a different cache server size. The size of the cache server is expressed as a percentage of the origin server.

FIG. 18 shows that the cache hit rate can be improved by inter-cache server cooperation regardless of the size of the cache server. However, in the case of FIG. 18, the content allocation to each cache server is not controlled in consideration of cooperation, and the requested content is only cached at the judgment of each cache server.

Regarding the content allocation to each cache server, the cache hit rate improvement effect by the cooperation is more than the value shown in FIG. For example, if some cache servers cache content that is popular to some extent, control can be performed so that other cache servers do not cache the same content. On the other hand, for very popular contents, it may be better to place them on a plurality of cache servers in terms of load distribution.

Also, Non-Patent Document 1 describes a technology related to CRISP (Caching and Replication for Internet Service Performance) in which a plurality of cache servers cooperate.

When trying to realize a content arrangement that allows the cache servers 300 ′ shown in FIG. 17 to be efficiently linked, the management server 400 stores information on the contents held by each cache server 300 ′, Access information must be collected from each cache server 300 'and each cache server 300' must be controlled to achieve an optimal placement.

However, in the case of the method in which the management server 400 controls the arrangement of contents in an integrated manner, the amount of information to be exchanged with the cache server and the optimum arrangement are calculated when the type of contents and the number of cache servers increase. The amount of calculation for this becomes large. For this reason, there is a problem that the management server 400 cannot complete the processing. When such a state is reached, there is a concern that each cache server has more processing depending on the management server 400 and cannot continue normal operation and fails as a system.

An object of the present invention is to provide a cache system, a cache method, and a cache server that solve the above-described problems.

In order to solve the above-described problem, according to the cache system according to the first aspect of the present invention, the cache system includes an origin server and a plurality of cache servers that distribute the content requested from the terminal instead of the origin server. When the server does not hold the content requested from the terminal, it indicates the load caused by transmitting / receiving the content to / from the other cache server when acquiring the content from the other cache server holding the content and distributing it to the terminal Based on the index, content duplication is controlled.

According to the caching method according to the second aspect of the present invention, there is provided a caching method in a cache system comprising an origin server and a plurality of cache servers that deliver content requested from a terminal instead of the origin server, When the server does not hold the content requested from the terminal, it indicates the load caused by transmitting / receiving the content to / from the other cache server when acquiring the content from the other cache server holding the content and distributing it to the terminal Based on the index, content duplication is controlled.

According to the cache server according to the third aspect of the present invention, when the content requested from the terminal is distributed instead of the origin server, the content requested from the terminal is not held. When the content is acquired from another cache server holding the content and distributed to the terminal, content replication is controlled based on an index indicating the load caused by transmitting / receiving the content to / from another cache server.

Note that the above summary of the invention does not enumerate all the necessary features of the present invention, and sub-combinations of these feature groups can also be the invention.

As is apparent from the above description, according to the cache system according to the present invention, it is possible to provide a distributed cache system capable of obtaining a good cache hit rate.

1 is a diagram illustrating an example of a distributed cache system according to a first embodiment of the present invention. It is a figure which shows an example of the block configuration of the cache server which concerns on 1st Embodiment of this invention. It is a flowchart which shows the flow of the process which the cache server which concerns on 1st Embodiment of this invention performs replication determination and replication instruction | indication. It is a flowchart which shows the flow of a process when the cache server which concerns on 1st Embodiment of this invention receives the content request from a terminal. It is a figure which shows an example of the operation | movement sequence of the terminal which concerns on 1st Embodiment of this invention, and a cache server. It is a figure which shows an example of the block configuration of the cache server which concerns on 2nd Embodiment of this invention. It is a figure which shows an example of the operation | movement sequence of the terminal which concerns on 2nd Embodiment of this invention, and a cache server. It is a figure which shows an example of the block configuration of a cache server. It is a flowchart which shows the flow of a process when the cache server concerning 3rd Embodiment of this invention receives the content request from a terminal. It is a figure which shows an example of the operation | movement sequence of the terminal concerning 3rd Embodiment of this invention, and a cache server. It is a figure which shows an example of the distributed cache system which concerns on 4th Embodiment of this invention. It is a figure which shows an example of the block configuration of the cache server which concerns on 4th Embodiment of this invention. It is a figure which shows an example of the block configuration of the cache server which concerns on 5th Embodiment of this invention. It is a figure which shows an example of a cache system. It is a figure which shows an example of the operation | movement sequence of a terminal, a cache server, and an origin server. It is a figure which shows an example of the distributed cache system which does not cooperate between cache servers. It is a figure which shows an example of the distributed cache system which can be cooperated between cache servers. It is a graph which shows the effect by cooperation between cache servers.

Hereinafter, the present invention will be described through embodiments of the invention. However, the following embodiments do not limit the invention according to the scope of claims, and all combinations of features described in the embodiments are included. It is not necessarily essential for the solution of the invention.

<First Embodiment>
FIG. 1 shows an example of a distributed cache system according to the first embodiment. The distributed cache system of the first embodiment includes a network 1, a plurality of networks 2a, b, c,... (Hereinafter collectively referred to as network 2), and a plurality of terminals 100a, b, c,. , Terminal server 100), origin server 200, and a plurality of cache servers 301a, b, c,... (Hereinafter collectively referred to as cache server 301).

The cache server 301 is communicatively connected to the origin server 200 and other cache servers 301 via the network 1. Further, the cache server 301 is communicatively connected to the terminal 100 via the network 2.

FIG. 2 shows an example of a block configuration of the cache server 301. The cache server 301 includes a distribution unit 310, a cooperation unit 311, a replication determination unit 312, and a storage device 320.
Hereinafter, the function and operation of each component will be described.

When receiving the content request from the terminal 100, the distribution unit 310 checks whether the requested content is recorded as a cache in the storage device 320. If the requested content is recorded in the storage device 320, the distribution unit 310 distributes the content to the terminal 100. On the other hand, when the requested content is not recorded in the storage device 320, the cooperation unit 311 is requested to acquire the content from the other cache server 301 by cooperation between the cache servers. As a result, when the target content is obtained from another cache server 301, it is delivered to the terminal 100. When the target content cannot be obtained from the other cache server 301, the target content is requested to the origin server 200 and the responded content is distributed to the terminal 100.
The distribution unit 310 further holds the content obtained from the origin server as a cache in the storage device 320 on the condition that the origin server 200 has requested the same content a predetermined number of times.

The distribution unit 310 also records content information including information such as the number of accesses to the content from the terminal 100 in the storage device 320. The content information also includes a content identifier (URL or an identifier assigned by the system), and the content information can be referred to using the content identifier as key information. The content information may include other information such as size and media type.

In response to a request from the distribution unit 310, the cooperation unit 311 searches for another cache server 301 that holds the content requested from the terminal 100.

Here, when searching for the cache server 301 that holds the content requested from the terminal 100, it is assumed that information necessary for access, such as the IP address of the cache server 301 group to be a cooperation candidate, is set in advance. To do.

The cache server 301 requests the target content from the cache server group that is a candidate for cooperation, in close order or randomly. As a result, if there is a cache server 301 that holds the target content, the target content is acquired from the cache server 301 and distributed to the terminal 100 via the distribution unit 310.

In the above description, it is assumed that the cooperation candidate cache servers are sequentially inquired, but other methods may be adopted. For example, there may be a case where a cache server as a cooperation candidate is arranged in the same station. In that case, the target content is obtained from the cache server that has returned the response indicating that the target content is held by inquiring about the target content by means of broadcasting or multicasting means capable of broadcasting to a plurality of cache servers. You may do it. If the number of cache servers that are cooperation candidates is large, the sequential inquiry method requires a long time to find the target content. Therefore, the cache servers that are candidates for cooperation may be limited to some cache servers in the system.

When an arbitrary content is requested from another cache server 301 and the requested content is held in the storage device 320, the cooperation unit 311 responds to the request-source cache server 301 with the requested content. . At this time, the cooperation unit 311 records the number of accesses from other cache servers in the storage device 320 as one piece of content information.

In addition, the cooperation unit 311 is notified of the copy of the content together with the content identifier from the copy determination unit 312. After that, when there is a content request for the content from another cache server, the information instructing the copy of the content in the response of the content Is stored. In this way, the content requesting cache server is instructed to copy the content. Here, the content response is stored with information for instructing content duplication, but content duplication may be instructed by another signal.

The cooperation unit 311 also saves the received content in the storage device 320 as a cache if the content response stores information instructing content replication. Even when the content response is instructed to copy the content by another signal, when the content to be copied is received, the content is stored in the storage device 320 as a cache.

When the cooperation unit 311 requests content from another cache server, the free space of the storage device 320 of the cache server 301 and information indicating the load may be stored in the request. Here, the information indicating the load is, for example, information such as a CPU usage rate, a communication interface usage rate, a traffic amount related to content transmission / reception with another cache server, and the number of simultaneous connections. Further, the linkage unit 311 of the cache server 301 on the content instructing cache server 301 selects the cache to be instructed to copy the content based on information indicating the free capacity and load of the storage device 320 of the cache server on the content requesting side. A server may be selected.

The replication determination unit 312 monitors information related to the communication interface of the cache server. Then, if any of the number of contents transmitted and received at the same time by the inter-cache server cooperation or the total bandwidth by the inter-cache server cooperation exceeds a threshold value, it is considered that the inter-cache server cooperation is being used too much. Then, the content information recorded in the storage device 320 is referred to, and one or a plurality of contents contributing to the load between the cache servers is determined.

Threshold value is a value that depends on the CPU performance of the cache server and the performance of the communication interface, and can be set in advance. Furthermore, the threshold value may be corrected based on dynamically changing information such as the number of simultaneous contents transmitted / received in the entire cache server, the total distribution bandwidth, the CPU load factor, and the like. *

Also, the number of accesses from other cache servers can be used as an index for determining that it contributes to the load of cooperation between cache servers. Further, the content size may be an index that takes into account the size of the content, such as a value obtained by multiplying the number of accesses. Furthermore, it is good also as an parameter | index which determines that it contributes to the load of cooperation between cache servers in consideration of another factor.

When the content that contributes to the load of cooperation between the cache servers is determined, the copy determination unit 312 means that the other cache server is determined to perform the copy together with the corresponding one or more content identifiers. Information to be notified to the cooperation unit 311.

In the storage device 320, cached content and content information are recorded.

Content information includes the content identifier, the number of accesses from the terminal, and the number of accesses from other cache servers. In addition, information regarding content such as content size and media type may be included.

The distribution unit 310, the linkage unit 311, and the replication determination unit 312 described above can be implemented by installing all of them in software and operating on the CPU. Alternatively, some or all of them can be configured by hardware. On the other hand, the storage device 320 can be configured by a device capable of recording and reading information, such as a semiconductor memory and a hard disk drive.

Next, a procedure in which the cache server 301 according to the first embodiment determines the copy of the cached content and instructs the other cache server to copy the content will be described using the flowchart of FIG.

First, in step S200, the duplication determination process in the duplication determination unit 312 starts at a regular opportunity or when the cooperation unit 311 receives a content request from another cache server.

Next, in step S201, as described above with respect to the function of the replication determination unit 312, the content replication is performed by referring to the number of contents simultaneously transmitted / received by the inter-cache server cooperation, the bandwidth by the inter-cache server cooperation, Decide whether or not to do.

Next, in step S202, the copy determination result is evaluated, and if the copy is made, step S203 is processed (Y in S202), and if not copied, the process is ended (N in S202).

In step S203, the replication determination unit 312 determines one or more contents to be replicated using the number of accesses from other cache servers as an evaluation index.

Finally, the duplication determination unit 312 notifies the cooperating unit 311 of the duplication along with the identifier of the content to be duplicated (S204). When the linked unit 311 requests the content notified here from another cache server, the linkage unit 311 stores information instructing to copy the content in the response of the content, and the cache server that becomes the response destination of the content Is instructed to copy the content, and the process ends.

Next, the flow of processing when the cache server 301 of the first embodiment receives a content request from the terminal 100 will be described using the flowchart of FIG.

First, when receiving a content request from the terminal 100, the distribution unit 310 searches the storage device 320 and determines whether the requested content is already held as a cache (S301).

Here, when there is a cache (Y in S301), the distribution unit 310 reads the target content from the storage device 320, distributes it to the terminal 100 (S310), and ends the processing.

On the other hand, when there is no cache (N in S301), the function of the linkage unit 311 performs a process of finding another cache server holding the target content. This method has already been described.

If there is a cache server that holds the target content (Y in S303), the cooperation unit 311 requests the content from the cache server 301 that holds the content, and sends the responded content via the distribution unit 310. Distribute to the terminal 100 (S304). At this time, if the content response stores information instructing to copy the content, the process proceeds to Y in step S305, and the content is held in the storage device 320 as a cache in step S306. Thereafter, in step S307, information such as the number of accesses in the content information of the storage device 320 is updated, and the process is terminated.

On the other hand, if there is no content duplication instruction (N in S305), the content is not retained, the content information is updated in step S307, and the process ends.

Here, returning to step S303, the case where there is no cache server holding the target content (N in S303) will be described. In this case, in step S308, the distribution unit 310 requests the target content from the origin server, and distributes the content returned from the origin server to the terminal 100. At this time, based on information such as the number of accesses to the content, it is determined whether to store the content in the storage device 320 as a cache. When the data is held as a cache (Y in S309), the processing after step S306 already described is processed. On the other hand, when not holding as a cache (N of S309), the content information update of step S307 is implemented and a process is complete | finished.

FIG. 5 shows an example of an operation sequence of the terminal 100a and the cache servers 301a to 301c. Next, the cache server 301c instructs the cache server 301a to copy the content, and the flow of processing until the cache server 301a holds the content as a cache according to the instruction, using a more specific example. explain.

In this description, it is assumed that the terminal 100a is a terminal accommodated under the cache server 301a, and that the terminal requests content. The requested content is assumed to be held in the cache server 301c.

First, in step S400, the cache server 301a receives a content request from the terminal 100a. Since the cache server 301a holds the requested content as a cache, the cache server 301a requests the content from the cache server 301b among the cooperation candidate cache servers. However, since the cache server 301b does not hold the requested content, it responds “no content” to the cache server 301a (S401).

The cache server 301a requests the content from the next candidate cache server 301c again (S402).

At this time, it is assumed that the cache server 301c has been determined to copy the content by the function of the copy determination unit 312. In this state, when the cache server 301c is requested for content in step S402, when the linkage unit 311 of the cache server 301c responds to the requested content, the instruction to copy the content is stored in the response. (S404).

When the cache server 301a receives content storing information instructing content replication, the cache server 301a distributes the content to the terminal 100a (S405) and holds it as a cache in the storage device 320 (S406).

As described above, in the first embodiment, it is determined using the information obtained internally by each cache server that the load due to cooperation between the cache servers has become excessive. In addition, it operates to identify content that is the main cause of excessive load caused by cooperation between cache servers and increase the number of copies of the content. As described above, the distributed cache system according to the first embodiment can use inter-cache server cooperation without excess or deficiency even when there is no management server. In addition, a good cache hit rate by cooperation between cache servers can be obtained without communication load and processing load for calculating an appropriate content arrangement.

<Second Embodiment>
Next, a second embodiment will be described in detail with reference to the drawings. In the second embodiment, the configuration other than the cache server 302 is the same as that in the first embodiment, and therefore, the entire cache system is omitted here.

As shown in FIG. 6, the cache server 302 of the second embodiment is substantially the same as the cache server 301 of the first embodiment, but the functions of the linkage unit 313 and the replication determination unit 314 are the same as those of the linkage unit 311 of the first embodiment. This is different from the replication determination unit 312. Therefore, here, an explanation will be given focusing on the functions of the cooperation unit 313 and the replication determination unit 314.

The duplication determination unit 314 has substantially the same function as the duplication determination unit 312 of the first embodiment, but when the duplication of content is decided, the duplication target content and the number of duplications are also decided. More specifically, when notifying the cooperation unit 313 to instruct the copying of content, information on the number of copies is included in addition to the content identifier.

The replication determining unit 314 refers to the number of accesses and the number of simultaneous connections by cooperation between cache servers when determining the number of replicas. When it becomes necessary to increase the number of copies of content in the system immediately, such as when these values greatly exceed the threshold, the number of copies can be increased according to the degree of urgency.

The cooperation unit 313 has substantially the same function as the cooperation unit 311 of the first embodiment. Further, the linkage unit 313 receives either the number of copies notified from the copy determination unit 314 or the number of copies stored together with information indicating a content copy instruction as a content response from another cache server. Then, based on the received number of copies, when another cache server requests content to be copied, it instructs to copy the content. Further, by storing the number-of-replications information as required, it requests the other cache server 302 to instruct to copy the contents.

The cooperation unit 313 and the replication determination unit 314 described above can be implemented by installing all of them in software and operating on the CPU. Alternatively, some or all of them can be configured by hardware.

The processing flow of the cache server 302 of the second embodiment is almost the same as the processing flow of the cache server 301 of the first embodiment shown in FIGS. However, the cache server 302 is different from the process shown in FIG. 3 in that the copy number is determined at the same time when determining the copy instruction. In addition, when the content requested from another cache server is a copy target according to the notified copy number, the cache server 302 returns the copy number information together with information for responding the content and instructing the copy of the content. It is also different in that it is stored as necessary.

In addition, the processing of the cache server 302 also confirms the existence of the number of copies information when determining whether or not the content responded from the other cache servers by the cooperation between the cache servers includes information instructing the copy. Is different from the process shown in FIG. If the copy number information is stored, it is acquired and recorded together with the content identifier. As described above, the copy number information recorded here is referred to when instructing another cache server to copy the content based on the copy number information.

In order to clarify the processing flow of the second embodiment, the processing flow of the distributed cache system of the second embodiment will be described using the sequence diagram of FIG.

First, it is assumed that the communication amount due to the cooperation between cache servers is excessive in step S500. As described above, the determination as to whether or not the number is excessive may be, for example, the case where the number of simultaneous content distributions and the communication bandwidth by cooperation between cache servers greatly exceed a predetermined threshold. However, it may be determined that it is excessive or likely to be excessive based on other indicators such as the number of simultaneous distributions and the speed of increase in bandwidth.

In this state, the terminal 100 makes a content request to the cache server 302a (S501).

When the cache server 302a does not hold the requested content, the cache server holds the content (in this case, the cache server 302b. The procedure for sequentially searching for the cache server holding the content is irrelevant here. Therefore, the requested cache server 302b responds to the cache server 302a with the requested content. At this time, the cache server 302b stores the number of copies in the content response in order to request another cache server together with the content duplication instruction because the inter-cache server cooperation is excessively used. . Then, this is notified to the cache server 302a (S502). Here, it is assumed that the calculated number of replicas is 8, and that the cache server 302a is requested to request four of the eight replicas. After requesting four content copies, the number that the cache server 302b should instruct to copy the content is 3 (= 8-4-1) considering that one copy is made in the cache server 302a. Become one.

When the cache server 302a receives the content response, it delivers it to the terminal 100 (S503). At the same time, according to the copy instruction, the cache server 302a holds the received content as a cache (S504).

The cache server 302a also waits for an opportunity to instruct another cache server to copy the content for four more contents when the copy instruction is requested as the number of cache copies (= 4). In other words, it waits for a request for the content to be copied from another cache server.

Here, as shown in step S505, when content to be copied is requested from the cache server 302c, information indicating a copy instruction and a content response storing the number of copies are transmitted to the cache server 302c. In the example of step S505, it is assumed that the cache server 302a has set 2 as the number of copies. In this case, with respect to the content to be copied by the cache server 302a, the remaining number to be instructed to copy is 1 (= 4—number of copies requested 2—copy 1 to the cache server 302c).

As described above, in the second embodiment, in addition to the decision to execute replication, the number of replicas is determined, and according to the determined number of replicas, another cache server is also requested for a replication instruction.
In this way, the number of content copies can be increased at a high speed even when the cache server cooperation is excessively used or when excessive use is predicted, reducing the risk of system failure. .

<Third Embodiment>
Next, a third embodiment will be described in detail with reference to the drawings. In the third embodiment, the configuration other than the cache server 303 is the same as that of the first embodiment, and therefore, the illustration of the entire cache system is omitted here.

As shown in FIG. 8, the cache server 303 of the third embodiment is almost the same as the cache server 301 of the first embodiment, but the function of the linkage unit 315 is different from the linkage unit 311 of the first embodiment. Therefore, here, an explanation will be given focusing on the function of the cooperation unit 315.

The cooperation unit 315 requests content from other cache servers by cooperation between the cache servers. When the response is made, if the copy of the content is required by the function of the copy determination unit 312 and the responded content is a copy target, the content is held in the storage device 320 as a cache. This is different from the cooperation unit 311 of the first embodiment. Other than that, it has the same function as the cooperation part 311 of 1st Embodiment.

That is, in the case where the content requiring replication is requested, the linkage unit 311 of the first embodiment instructs the requested cache server to copy the content to the requested cache server. On the other hand, the cooperation unit 315 of the third embodiment is different in that the content requesting side determines the necessity of copying the content and the content to be copied.

The cooperation unit 315 described above can be implemented by installing all of them in software and operating on the CPU. Alternatively, some or all of them can be configured by hardware.

Next, when the cache server 303 according to the third embodiment receives the content distributed from another cache server by inter-cache server cooperation using the flowchart of FIG. 9, the content is cached as necessary. Explain the procedure.

First, in step S600, when the cooperation unit 315 receives content from another cache server, the copy determination unit 312 starts the copy determination process.

Next, in step S601, as described in the description of the function of the replication determination unit 312, the number of contents simultaneously transmitted / received by inter-cache server cooperation, the bandwidth by inter-cache server cooperation, and the like are referred to. Decide whether or not to replicate.

Next, in step S602, the copy determination result is evaluated, and if copying is performed, step S603 is processed (Y in S602), and if not copied, the process is terminated (N in S602).

In step S603, the copy determination unit 312 determines one or more contents to be copied using the number of accesses from other cache servers as an evaluation index.

Next, the cooperation unit 315 determines whether or not the content received from another cache server is included in the content to be copied determined in step S603 (S604). If it is included, the process proceeds to Y, and the content is held as a cache in the storage device 320 in step S605. On the other hand, if it is determined in step S604 that the received content is not included in the content to be copied, the process proceeds to N and is not held as a cache, and the process ends.

The processing flow when the cache server 303 receives a content request from the terminal 100 is obtained by replacing the processing from step S305 to step S306 shown in FIG. 4 with the processing from step S600 to step S605 shown in FIG. Become.

Next, referring to the sequence diagram of FIG. 10, the processing flow until the cache server 303a acquires the content requested from the terminal 100 from the cache server 303c and holds the content as a cache based on the copy determination result. explain.

In this description, it is assumed that the terminal 100a is a terminal accommodated under the cache server 303a, and the terminal requests content. Assume that the requested content is not held by the cache server 303a but is held by the cache server 301c.

First, in step S700, the cache server 303a receives a content request from the terminal 100a. Since the cache server 303a holds the requested content as a cache, the cache server 303a requests the content from the cache server 303b selected from the cooperation candidate cache servers. However, since the cache server 303b does not hold the requested content, “no content” is returned to the cache server 303a (S701).

Therefore, the cache server 303a requests content from the next candidate cache server 303c (S702).

When the cache server 303c receives the content request in step S702, the cache server 303c holds the requested content, and therefore responds to the cache server 303a with the content (S703).

Upon receiving the content response, the cache server 303a responds to the terminal 100a (S704), and determines whether the received content should be cached according to the flowchart shown in FIG. If it is determined that the content should be cached, the content is held as a cache (S704).

As described above, in the third embodiment, when a cache server receives content from another cache server through cooperation between cache servers, the cache server that receives the content determines whether or not to cache. . In this way, the cache server on the content response side does not need to store information instructing copying in the content response. For this reason, when the existing protocol such as HTTP is reused as a protocol for exchanging contents between the cache servers, it is not necessary to change the existing protocol, so that the cache system according to the present invention can be easily applied. Become.

<Fourth embodiment>
Next, a fourth embodiment will be described in detail with reference to the drawings. An overall view of the distributed cache system of the fourth embodiment is shown in FIG. Here, the devices and networks having the same names and numbers are the same as in the overall view (FIG. 1) of the distributed cache system of the first embodiment, and thus the description thereof is omitted here.

The cache server 304 and the management server 401 are different from the distributed cache system according to the first embodiment of the present invention.

As shown in FIG. 12, the cache server 304 according to the fourth embodiment further includes an information exchange unit 316 in addition to the cache server 301 according to the first embodiment. Moreover, the function of the cooperation part 317 is a little different from the cooperation part 311 of 1st Embodiment. The other functional blocks are the same as the configuration of the cache server 301 of the first embodiment, and a description thereof will be omitted.

The information exchanging unit 316 has a function of notifying the management server 401 of a list of contents held in the storage device 320, acquires a list of contents held by another cache server 304 from the management server 401, and stores this list Record in device 320. At this time, the storage device 320 records the identifier of the cache server 304 in association with the list of contents held by the cache server.

As for the timing at which the information exchange unit 316 notifies the management server of the content list and the opportunity to acquire the content list of another cache server, for example, a part or all of the content list is periodically notified and acquired. It doesn't matter. Further, it may be held at the timing when the cache is deleted because the content is held in the storage device 320 as a cache, or because the capacity of the storage device 320 is low. Alternatively, the management server 401 may be notified of deleted content information. Furthermore, when the content is requested from the terminal 100, the holding status of another requested cache server of the requested content may be acquired from the management server 401.

The cooperation unit 317 has substantially the same function as the cooperation unit 311 according to the first embodiment of the present invention. Further, when determining the content requesting cache server, the linkage unit 317 refers to the content list of another cache server recorded in the storage device 320. Then, after specifying a cache server that holds the requested content, one of the specified cache servers is determined as a content request destination. This eliminates the need to find a cache server that holds the requested content in a sequential manner, thus enabling more efficient cooperation between cache servers.

The management server 401 holds the content list notified from the cache server 304. In response to a request from the cache server 304, a content list of the cache server 304 in the cache system is returned. In order to reduce the amount of information exchange, the responding content list may be limited to a part of the information to be responded, such as limiting the content list to the content list for the cache servers around the cache server 304 that has requested the content list. .

In the cache system according to the fourth embodiment of the present invention, the content list is shared via the management server 401, thereby enabling smooth cooperation between the cache servers. On the other hand, the management server 401 is necessary. Communication between the management server 401 and the cache server 304 occurs. However, even in this case, the calculation load and communication load of the management server, which are listed as problems in the background art, are not a big problem here. This is because the functions of the management server 401 of the fourth embodiment are limited to functions that mediate limited information. More specifically, the information to be exchanged is limited to the content list, and the management server 401 calculates the content allocation to each cache server and uses the cache server for the calculated content allocation. This is because there is no need to control.

The information exchange unit 316 and linkage unit 317 described above can be implemented by installing all of them in software and operating on the CPU. Alternatively, some or all of them can be configured by hardware.

The management server 401 can be implemented on a server equipped with a CPU or storage device with the above-described functions implemented by software.

The operation flow of the cache server 304 according to the fourth embodiment of the present invention is almost the same as the processing when the cache server 301 of the first embodiment receives a content request from the terminal 100, as already shown in FIG. Become. The only difference is that the content list acquired from the management server 401 is referred to when the cache server as the content request destination is found in step S302.

The distributed cache system according to the fourth embodiment further includes a management server 401 compared to the distributed cache system of the first embodiment. The cache server 304 includes an information exchange unit 316 for exchanging content lists with the management server 401. The cooperation unit 317 uses the content list of another cache server obtained from the management server 401 by the information exchange unit 316 to determine a cache server as a content request destination. The elements that are the differences between the first embodiment and the fourth embodiment can be similarly applied to the configurations of the second embodiment and the third embodiment in the same manner as in the first embodiment.

As described above, in the fourth embodiment, when acquiring content from another cache server, the cache server uses the content list obtained by the management server to determine the content request destination. Efficient linkage between cache servers can be realized.

<Fifth Embodiment>
Next, a fifth embodiment will be described in detail with reference to the drawings. In the fifth embodiment, the cache server 305 shares the functions of the management server 401 of the fourth embodiment. Therefore, the management server 401 is not necessary in the distributed cache system of the fifth embodiment. For this reason, the configuration of the distributed cache system of the fifth embodiment is the same as that of the first embodiment, and thus the figure is omitted.

As shown in FIG. 13, the cache server 305 of the fifth embodiment further includes a management server function 318 in addition to the cache server 304 of the fourth embodiment.

The management server function 318 exchanges content list information with the information exchange unit 316 of the other cache server 305 in the same manner as the function of the management server 401 in the fourth embodiment. However, it is different from the management server 401 in the fourth embodiment in that a part of content is managed instead of managing a list of all content. A method for determining content to be managed by the management server function 318 of each cache server 305 will be described later.

The information exchange unit 316 ′ basically has the same function as the information exchange unit 316 of the fourth embodiment. The information exchanging unit 316 ′ is different in that the notification destination and acquisition destination of the content list are not the management server 401 but the management server function 318 of another cache server 305. Another difference is that a cache server as a content notification destination / acquisition destination is determined for each content.

An example of a method in which the information exchanging unit 316 ′ determines a cache server as a content list notification destination and acquisition destination for each content is shown below.

Here, a cache server index (CacheServerIndex) that is a continuous value from 0 is assigned to each cache server. Assume that the cache server index and the number of cache servers (NumOfCacheServer) in the distributed cache system are shared by the cache servers.

At this time, the information exchanging unit 316 ′ determines a cache server that notifies the holding state of the content when the content identifier whose content identifier is ContentID is held as a cache, when the content is deleted, or when the content is periodically deleted. At that time, the cache server index is obtained by the calculation shown in Expression (2). Then, the cache server to which the obtained cache server index is assigned is determined as the content list notification destination of the content.

　　　　　　　　　　　　　　　　　　

Here, Hash (x) indicates a process of calculating a hash value of x by a hash function such as MD5 or SHA1. X% y means a remainder when x is divided by y.

Formula 2 is also used when determining the cache server that manages the content list of content in order to acquire information on the cache server that holds a certain content. That is, the cache server to which the cache server index obtained by Expression 2 is assigned is set as the content list acquisition destination.

The cache list sharing method shown above is merely an example, and the content list may be shared and managed by other methods.

The information exchange unit 316 'and the management server function 318 described above can be implemented by installing all of them in software and operating on the CPU. Alternatively, some or all of them can be configured by hardware.

As described above, in the fifth embodiment, efficient management between cache servers can be realized even if there is no management server by managing the functions of the management server by each cache server.

The cache system according to the first embodiment uses information obtained internally by each cache server to determine that the load due to cooperation between cache servers has become excessive, and the load due to cooperation between cache servers has become excessive. Identify content that is the main factor that you have, and increase duplication of that content. Thereby, even when there is no management server, inter-cache server cooperation can be used without excess or deficiency. In addition, it is possible to provide a distributed cache system that can obtain a good cache hit rate by cooperation between cache servers without communication load and processing load for calculating an appropriate content arrangement.

The cache system according to the second embodiment is configured to determine the number of replicas in addition to the determination of the execution of replication, and to request replication instructions from other cache servers according to the determined number of replicas. In this way, the number of content copies can be increased at a high speed even when the cache server cooperation is excessively used or when excessive use is predicted, reducing the risk of system failure. .

According to the cache system according to the third embodiment, when a cache server receives content from another cache server by cooperation between cache servers, the cache server side that receives the content determines whether or not to cache. . In this way, the cache server on the content response side does not need to store information instructing copying in the content response.
For this reason, when an existing protocol such as HTTP is reused as a protocol for exchanging contents between cache servers, it is not necessary to change the existing protocol, so that the present invention can be easily applied.

According to the cache system according to the fourth embodiment, when the cache server acquires content from another cache server, the content server obtained by the management server is used to determine the content request destination. More efficient linkage between cache servers can be realized.

According to the cache system according to the fifth embodiment, by efficiently managing the functions of the management server by each cache server, even when there is no management server, efficient cooperation between cache servers can be realized.

By applying the distributed cache system according to the present invention to a fixed ISP network or a mobile network, it is possible to obtain a high cache hit rate utilizing the cooperation between cache servers in the distributed cache system. Thereby, the traffic on an interconnection point (POI: Point Of Interface) can be reduced effectively, and the operator of the network can reduce OPEX (Operation Expense). In addition, high scalability and high availability can be realized by enabling distributed processing such as content copying.

As described above, the cache system according to the present invention has been described using the embodiment. However, the technical scope of the present invention is not limited to the scope described in the above embodiment. It is apparent from the scope of the claims that embodiments in which various modifications or improvements are added to the above embodiments can also be included in the technical scope of the present invention.

This application claims priority based on Japanese Patent Application No. 2011-215850 filed in Japan on September 30, 2011, the contents of which are incorporated herein by reference.

According to the cache system according to the present invention, there is provided a cache system, a cache method, and a cache server that do not require a management server and can realize an effective cooperation between cache servers based on information that each cache server can have. be able to.

DESCRIPTION OF SYMBOLS 1 Network 2 Network 100 Terminal 200 Origin server 301 Cache server 302 Cache server 303 Cache server 304 Cache server 304 Cache server 305 Cache server 310 Distribution part 311 Cooperation part 312 Replication determination part 313 Cooperation part 314 Replication determination part 315 Cooperation part 316 Information exchange part 316 ' Information exchange unit 317 Cooperation unit 318 Management server function 320 Storage device 401 Management server

Claims (30)

1. An origin server,
   A plurality of cache servers that deliver the content requested from the terminal instead of the origin server;
   The cache server is
   An index indicating a load caused by transmitting / receiving content to / from another cache server when the content requested from the terminal is not held and acquired from another cache server holding the content and distributed to the terminal -Based cache system that controls content replication.
2. The index is an index obtained from at least one of a traffic amount related to content transmission / reception between cache servers, the number of simultaneous sessions between cache servers, a communication interface usage amount of a cache server, and a CPU usage rate. The described cache system.
3. When determining the copy of content, the cache server determines one or more contents to be copied based on at least the number of accesses to the content held by the cache server and the size of the content The cache system according to claim 1 or 2.
4. The cache server decides to copy the content, determines one or more contents to be copied, and then determines the cache server from the other cache servers that constitute the cache system. 4. The cache system according to claim 3, wherein when any one of the contents is requested, a cache server as a content request source is instructed to respond with the contents and hold the contents as a cache. .
5. The cache server determines one or more contents to be copied when determining to copy the contents, and determines the number of copies of the contents,
   When any one of the determined contents is requested to the cache server from another cache server that constitutes the cache system, the content is requested to the cache server that is a content request source. In addition to the information instructing to hold the content as a cache, the number of copies set to the number of requests to request a copy from the requesting cache server among the determined number of content copies is notified. ,
   When the content requesting cache server receives the number of copies, it is instructed to copy the content to the other cache server as many copies as the number of copies requested, Alternatively, the cache system according to claim 4, wherein a content duplication instruction is requested by notifying another cache server of the number of copies.
6. When the cache server requests content from another cache server, the request relates to free information on the storage capacity of the cache server, CPU usage status, communication interface usage, and content transmission / reception with another cache server. Store at least one of the traffic volume and the number of simultaneous connections as cache server load information,
   5. The cache system according to claim 4, wherein the cache server that has received the content request of the cache server determines a cache server to be a content duplication instruction based on the cache server load information acquired from the content request.
7. 4. The cache system according to claim 3, wherein when the content is acquired from another cache server and the acquired content is included in the content determined as a copy target, the cache server holds the content as a cache. .
8. The cache server holds the requested content by a sequential method or a broadcast method when content is requested from a terminal and the cache server does not hold the content. The cache system according to claim 1, wherein a cache server is found.
9. The cache server notifies the management server of content information held by the cache server;
   When the cache server requests content from the terminal and the self-cache server does not hold the content, the cache server holds the requested content based on information obtained from the management server. The cache system according to claim 1, wherein a cache server is found.
10. The function corresponding to the management server is shared by each cache server, and each cache server determines a cache server in charge of managing content holding information for each content based on a content identifier. Cache system.
11. A cache method in a cache system comprising: an origin server; and a plurality of cache servers that deliver content requested from a terminal instead of the origin server,
    The cache server is
    An index indicating a load caused by transmitting / receiving content to / from another cache server when the content requested from the terminal is not held and acquired from another cache server holding the content and distributed to the terminal Based on the above, the content duplication is controlled.
12. 12. The index is obtained by at least one of a traffic amount related to content transmission / reception between cache servers, the number of simultaneous sessions between cache servers, a communication interface usage amount of a cache server, and a CPU usage rate. The described caching method.
13. When determining the copy of content, the cache server determines one or more contents to be copied based on at least the number of accesses to the content held by the cache server and the size of the content The cache method according to claim 11 or 12.
14. The cache server decides to copy the content, and after determining one or more contents to be copied, the cache server determines the determined content from another cache server constituting the cache method to the cache server. The cache method according to claim 13, wherein when any one of the contents is requested, a cache server as a content request source is instructed to respond with the contents and hold the contents as a cache. .
15. The cache server determines one or more contents to be copied when determining to copy the contents, and determines the number of copies of the contents,
    When any one of the determined contents is requested from the other cache server constituting the cache method to the cache server, the content is requested to the cache server that is a content request source. In addition to the information instructing to hold the content as a cache, the number of copies set to the number of requests to request a copy from the requesting cache server among the determined number of content copies is notified. ,
    When the content requesting cache server receives the number of copies, it is instructed to copy the content to the other cache server as many copies as the number of copies requested, The cache method according to claim 14, wherein the content copy instruction is requested by notifying another cache server of the number of copies.
16. When the cache server requests content from another cache server, the request relates to free information on the storage capacity of the cache server, CPU usage status, communication interface usage, and content transmission / reception with another cache server. Store at least one of the traffic volume and the number of simultaneous connections as cache server load information,
    The cache method according to claim 14, wherein the cache server that has received the content request of the cache server determines a cache server that is a target of a content duplication instruction based on the cache server load information acquired from the content request.
17. The cache method according to claim 13, wherein when the content is acquired from another cache server and the acquired content is included in the content determined as a copy target, the cache server holds the content as a cache. .
18. The cache server holds the requested content by a sequential method or a broadcast method when content is requested from a terminal and the cache server does not hold the content. The cache method according to claim 11, wherein a cache server is found.
19. The cache server notifies the management server of content information held by the cache server;
    When the cache server requests content from the terminal and the self-cache server does not hold the content, the cache server holds the requested content based on information obtained from the management server. The cache method according to claim 11 or 12, wherein a cache server is found.
20. 20. The function corresponding to the management server is shared by each cache server, and each cache server determines a cache server in charge of managing content holding information for each content based on a content identifier. Caching method.
21. A cache server that delivers content requested from the terminal instead of the origin server,
    An index indicating a load caused by transmitting / receiving content to / from another cache server when the content requested from the terminal is not held and acquired from another cache server holding the content and distributed to the terminal A cache server that controls content duplication based on.
22. The index is obtained by at least one of a traffic amount related to content transmission / reception between cache servers, the number of simultaneous sessions between cache servers, a communication interface usage amount of a cache server, and a CPU usage rate. The listed cache server.
23. 23. The content to be copied is determined based on at least one of the number of accesses to the content held by the cache server and the size of the content when determining the copy of the content. The cache server described in.
24. After deciding to duplicate the content and determining one or more contents to be duplicated, the other cache servers that constitute the cache server send the cache server to the cache server. 24. The cache server according to claim 23, wherein when any content is requested, the cache server that is a content request source is instructed to respond to the content and hold the content as a cache.
25. 25. The cache server according to claim 24, wherein when the response of the content is received, the content is distributed to the content terminal obtained by the content response according to the duplication instruction and held as a cache.
26. When deciding to duplicate content, determine one or more content to be duplicated, determine the number of content duplicates,
    When any one of the determined contents is requested to the cache server from another cache server constituting the cache server, the content is requested to the cache server that is a content request source. In addition to information instructing to hold the content as a cache, the number of copies set to the number of requested copy requests to the cache server of the request among the determined number of content copies is notified. The cache server according to claim 24.
27. When the content request is received, it is assumed that the copy instruction for the number of copies is requested, the other cache server is instructed to copy the content, or the other cache server is set with the copy number. 27. The cache server according to claim 26, wherein the cache server requests an instruction to copy the content by notifying.
28. When requesting content from another cache server, the request includes free information on the storage capacity of the cache server, CPU usage, communication interface usage, traffic related to content transmission / reception with other cache servers, and simultaneous connection. The cache server according to claim 24, wherein at least one of the numbers is stored as cache server load information.
29. 25. The cache server according to claim 24, wherein when the content request is received, a cache server that is a target of content duplication instruction is determined based on an index indicating the load acquired from the content request.
30. 24. The cache server according to claim 23, wherein when content is acquired from another cache server, if the acquired content is included in content determined as a copy target, the content is retained as a cache.
    

Images