JP4097274B2 - Resource search method, cluster system, computer, and cluster - Google Patents

Description

  The present invention relates to a resource search method, a cluster system, a computer, and a cluster for efficiently searching for resources when accommodating resources in a cluster system.

  The cluster system collectively manages computer resources such as computers and disks (hereinafter referred to as resources) and assigns them to applications as needed, improving reliability, processing performance, resource utilization, etc. It is a system that aims to. The cluster system is realized in a form in which a cluster manager is operated as a middleware or a module of an OS on one or a plurality of computers in the cluster to manage resources.

  A normal cluster system allocates closed resources in one cluster. However, in recent years, when resources are insufficient in one cluster due to an increase in data to be processed and cost reduction requests, resource interchange (hereinafter referred to as resource interchange) between clusters reduces resource shortage. There is a need to deal with it.

  In general, a specific CPU or a specific OS is required to operate an application. Resources such as computers can only be used if they meet this requirement. Therefore, when performing resource interchange between clusters, a mechanism for searching for available resources across the clusters is required. In the search, information about resources (hereinafter referred to as resource information) such as the CPU and OS type requested by the application is specified as a condition.

  Among these conditions, it is preferable that the condition is as large as possible after satisfying the minimum condition such as the number of clocks of the CPU, and the discrete condition that is divided into whether the condition is satisfied or not, such as the type of CPU. There are continuous conditions.

  In the conventional resource search in a single cluster, a method of specifying a desired condition in detail and searching a resource corresponding to the condition from all resources is used. The intercluster resource interchange currently under consideration is an extension of the resource search method in a single cluster as described above between the clusters. Specifically, the condition of the resource desired by the borrowing cluster is specified in detail, and the resource corresponding to the condition is searched from every resource in every cluster.

With regard to a method for storing resource information to be searched, conventionally, for example, a system registry described in Patent Document 1 has a mechanism for storing information of the entire system composed of a plurality of clusters as it is. There is a known method for maintaining the consistency of the information stored in the system with the information of the entire system for the common part.
JP 2002-49601 A

  If the resource search method within a single cluster is expanded as it is between clusters, the resource conditions are specified in detail across the clusters, and information on all resources that meet the conditions is exchanged for all clusters. There was a problem that the communication amount and the processing amount were large.

  Accordingly, the present invention has been made to solve the above-described problems. A resource search method, a cluster system, a computer, and a computer that can reduce the amount of communication and processing required for resource search in accommodating resources between clusters. The purpose is to provide a cluster.

One embodiment of the present invention is a cluster system including a plurality of clusters each including a resource that is a computer resource necessary for the operation of one or more computer programs, and a computer connected to each of the plurality of clusters. A resource search method for searching for a resource that meets a desired condition from a first cluster among the plurality of clusters from resources of another cluster, wherein each of the other clusters includes A resource to be transmitted for storing the condition information of each resource as condition information of the own cluster, and storing the condition information of the collected resources and the cluster condition information in which the cluster identification information for identifying the own cluster is associated with each other. An information aggregating step, and the computer receives the desired condition transmitted from the first cluster. From the cluster condition information their respective of said other clusters that store, to give a cluster identification information corresponding to the matching cluster condition information to the desired condition, the cluster identification information to the first cluster A candidate cluster search step for transmitting, and a resource search step for searching for a suitable resource by transmitting the desired condition to at least one of the other clusters specified by the cluster identification information by the first cluster; The condition information is composed of a set of a main part which is an information name and a predicate which is a numerical value corresponding to the information name, and the resource information aggregation step includes a predescription part of the condition information. The maximum upper limit indicated by the numerical range specification is the upper limit in the cluster condition information, and the lowest upper limit is the lower limit in the cluster condition information. Characterized by a value.

In one embodiment of the present invention , in a cluster system including a plurality of clusters each including one or more resources and a computer connected to each of the plurality of clusters, each of the plurality of clusters is A first storage means for storing condition information for each resource in the own cluster; an aggregation means for aggregating the condition information for each resource in the own cluster stored in the first storage means as the condition information for the own cluster; Transmitting the cluster condition information in which the condition information aggregated by the aggregation means is associated with the cluster identification information for identifying the own cluster; and receiving the search request with the condition information added, the first storage searching means the condition information, and a first search means for obtaining information indicating the matching resource, said condition information includes a main portion which is information name, the The aggregation means includes a combination of a predicate that is a numerical value corresponding to a report name, and the aggregation means sets the maximum one of the upper limit values indicated by the numerical range specification of the previous description part among the respective condition information as the upper limit value in the cluster condition information And the lowest one of the lower limit values is set as the lower limit value in the cluster condition information, and the computer stores, from each of the clusters, second condition storage means for storing each of the condition information of the cluster aggregated by the aggregation means, When a search request to which condition information is added is received, the second storage means is searched with the condition information, and second search means for obtaining information for identifying a suitable cluster is provided .

Further, one embodiment of the present invention is a computer connected to a plurality of clusters each including one or more resources, from each of the plurality of clusters, condition information of each resource in the own cluster, Storage means for storing each of the cluster condition information aggregated as the condition information of the own cluster, and information for identifying the matching cluster by searching the storage means with the condition information upon receiving a search request with the condition information added And a transmission unit that transmits information for identifying the cluster obtained by the search unit to a search request source. The condition information corresponds to a main part that is an information name and the information name. The upper limit value in the cluster condition information is the largest of the upper limit values indicated by the numerical range specification of the previous description part in each condition information. And than the lower limit value in the cluster condition information, wherein the at smallest of the lower limit value indicated by the numerical value range specification before description part of the condition information.

In one embodiment of the present invention , in a cluster connected to a computer, storage means for storing condition information for each resource in the own cluster, and condition information for each resource in the own cluster stored in the storage means And a means for transmitting to the computer the cluster condition information in which the condition information aggregated by the aggregation means and the cluster identification information for identifying the own cluster are associated with each other. When a search request with condition information added is received, the storage means is searched with the condition information, search means for obtaining information indicating a suitable resource, and information for identifying the resource obtained by the search means are searched. and transmitting means for transmitting the request source, wherein the condition information includes a main portion which is information name, made from a set of predicates is a value corresponding to the information name, The aggregating means uses, as the upper limit value in the cluster condition information, the upper limit value indicated by the numerical range specification of the previous description part among the above condition information, and the minimum one of the lower limit values as the lower limit in the cluster condition information It is characterized by a value.

  According to the present invention, since the resource search is performed on the clusters narrowed down by the cluster search, it is possible to reduce the communication amount and the processing amount for the resource search for inter-cluster resource accommodation.

  Embodiments of the present invention will be described below with reference to the drawings.

  In the present embodiment, the entire cluster group that performs mutual resource interchange is called a grid. FIG. 1 shows the overall configuration of the grid according to the present embodiment.

  The grid 100 includes a plurality of clusters 1, 2,..., N and a computer 50 on which one grid manager 50 a that manages the clusters operates, and these are connected by a network 40. Among the plurality of similar clusters 1, 2,..., N, the configuration of the cluster will be described using cluster 1. FIG.

  The cluster 1 includes a computer 11 on which the resource management unit 1a operates, and resources 12,..., 13 managed by the resource management unit 1a. The resources 12,..., 13 are computer resources required by services configured on the computer. In this embodiment, the resources 12,..., 13 are computers or disks, and each cluster 1, 2,. Contains one or more resources. Regardless of the present embodiment, as long as computer resources are necessary for a service using a computer, input / output devices such as printers, communication devices, and video devices, or simple data such as software license information can be used as resources of the present invention. And can be The resource management unit 1a includes resource search means 1b for searching for resources in the cluster 1, and resource condition information storage means 1c for storing condition information for each resource (hereinafter referred to as resource condition information) for the resources in the cluster 1. , Cluster condition information storage means 1d for storing condition information (hereinafter referred to as cluster condition information) in which the conditions included in each resource condition information of each resource in cluster 1 are aggregated for each type, and resource condition information storage means 1c Resource condition information aggregating means 1e for aggregating the resource condition information stored in the cluster condition information to be stored in the cluster condition information storage means 1d.

  The resource management unit 1a may be operated on a computer other than the resources 12,..., 13 as shown in FIG. 1, or on any of the resources 12,. It may be operated. Note that the configuration of the clusters 2, 3,.

  The grid 100 is provided with a computer 50 having a grid manager 50a and a plurality of cluster condition information storage means 50b. The grid manager 50a mediates resource interchange between the clusters 1, 2,. The grid manager 50a is realized as a middleware or an OS module. In this example, the grid manager 50a is operated on an independent computer 50 that does not belong to the clusters 1, 2,..., N. It may be operated with a computer.

  The plurality of cluster condition information storage means 50b sets the cluster condition information of the clusters corresponding to the pointers to the resource management units 1a, 2a,. It is possible to communicate with the resource managers 1a, 2a,..., Na using the pointer as necessary. The pointer may be described in a setting file or the like of the grid manager 50a and read by the computer 50 of the grid manager 50a at startup. Alternatively, when the clusters 1, 2,..., N join the grid 100, the resource managers 1a, 2a,. As the pointer, for example, in the case of the TCP / IP protocol, an IP address and a port number can be used.

  FIG. 2 shows an example of resource condition information stored in the resource condition information storage unit 1c, and FIG. 3 shows an example of cluster condition information stored in the cluster condition information storage unit 1d.

  In each condition information, the upper two lines are identification information, and the remaining lines are conditions. Each condition is represented by a conditional expression here. In the identification information part, the left column is the information name and the right column is the corresponding value, and in the conditional expression, the left column is the main part of the conditional expression, and the right column is the predicate. A predicate further consists of a predicate and an argument. The predicate is one of “=”, “≦”, “≧”, and “is-one-of”. “=” Indicates that the main part and the argument are equal. “≦” indicates that the upper limit of the main part is the numerical value of the argument. “≧” indicates that the lower limit of the main part is the numerical value of the argument. “Is-one-of” takes one or more arbitrary arguments and indicates that the main part is equal to one of those arguments. In the present embodiment, a plurality of conditional expressions in a single piece of condition information have their products as values.

  Next, a resource search method in the grid 100 having the above configuration will be described. FIG. 4 is a flowchart showing the operation of the resource search method of this embodiment. The leftmost column in FIG. 4 shows the operation of the resource management unit 1a of the cluster 1 that borrows resources from other clusters, and the adjacent column shows the operation of the grid manager 50a. A plurality of resource management units on the right side of the grid manager 50a indicate the operations of the resource management units 2a, 3a, ..., na of the clusters 2, 3, ..., n that lend their resources to the cluster 1.

  The cluster condition information stored in the cluster condition information storage means 1d, 2d, ..., nd of the clusters 1, 2, ..., n is also redundantly stored in the multiple cluster condition information storage means 50b of the grid manager 50a. For this reason, in order to ensure the consistency of the cluster condition information, it is necessary to transfer the cluster condition information. Therefore, the cluster condition information is transferred from the cluster condition information storage means 1d, 2d,... Nd of the clusters 1, 2,..., N to the multiple cluster condition information storage means 50b of the grid manager 50a at regular intervals, for example. And If the update date and time is given to the cluster condition information, only conditional expressions that have been updated since the previous transfer are transferred, and transfer of conditional expressions that have not been updated is omitted to reduce the transfer amount and processing amount. You can also.

  The resource search method according to the present embodiment is roughly divided into three steps, that is, an information aggregation step S100, a candidate cluster search step S200, and a corresponding resource search step S300.

  First, the information aggregation step S100 will be described. In addition, since cluster 1 ... n performs the same operation | movement of information aggregation step S100, below, cluster 1 is demonstrated as a representative. In the information aggregation step S100, prior to each search, the condition conditions (conditions) of the resource condition information stored in the resource condition information storage unit 1c are aggregated by the resource condition information aggregation unit 1e, and the cluster condition information storage unit 1d. Process to store in. Details of the information aggregation step S10 are shown in FIG.

  First, the resource condition information aggregating unit 1e acquires a list of ResourceType values of identification information stored in the resource condition information storage unit 1c, and selects one from the ResourceType values (S101). Next, all the conditional expressions included in each condition information having the selected ResourceType value are extracted from the resource condition information storage unit 1c and temporarily stored (S102). A list of main parts of the conditional expression is acquired, and one main part of the conditional expression is selected (S103). Prior to this selection, variables of an upper limit value R and a lower limit value r for this main part are prepared, and R = −∞ and r = ∞ are set as initial values. Next, one conditional expression including the main part of the selected conditional expression is selected (S104).

  The predicate of the selected conditional expression is confirmed (S105). When the predicate is “= numerical value”, “≦ numerical value”, “≧ numerical value”, or “not numeric (is-one-of)”, the following processing is performed. When the predicate is “= numerical value”, the upper value R is compared with the upper limit value R, the larger value is set as the upper limit value R, and the lower value r is compared with the lower value r. S106). When the predicate is “≦ numerical value”, it is compared with the upper limit value R, and the larger value is set as the upper limit value R (S107). When the predicate is “≧ numerical value”, it is compared with the lower limit value r, and the smaller value is set as the lower limit value r (S108). When the predicate is “not a numerical value (is-one-of)”, the value of the predicate is temporarily stored (S109).

  Next, it is determined whether there is an unconfirmed predicate in the temporarily stored conditional expression (S110). If there is an unconfirmed predicate, the process returns to step S103 to describe the temporarily stored conditional expression. Repeat the process until all parts are confirmed.

  On the other hand, if there is no unconfirmed one, then, if the upper limit value R = −∞ and the lower limit value r = ∞, it is assumed that there is no value, and (1) either the upper limit value R or the lower limit value r (The predicate means that no numerical value was taken), (2) only one, or (3) an upper limit R and a lower limit r, and an upper limit R and a lower limit r. (4) It is confirmed whether there is an upper limit value R and a lower limit value r and the upper limit value R and the lower limit value r are not equal (S111).

As a result, in the case of (1) above, a conditional expression is generated with one is-one-of as a predicate and all stored values as arguments (S112). ≦ Upper limit value ”or“ Main part ≧ Lower limit value ”is generated as a conditional expression (S113). In the case of (3) above,“ Main part = numerical value (= upper limit value or lower limit value) ”is generated as a conditional expression. (S114) In the case of (4) above, a conditional expression is generated by multiplying “main part ≦ upper limit value” and “main part ≧ lower limit value” (S115). The output conditional expression is temporarily held (S116).

  The process for one condition main part is completed as described above, and it is confirmed whether there is still another unprocessed condition main part (S117). If there is, the process returns to step S102, and there is no unprocessed condition main part. Repeat until the process.

On the other hand, if it is confirmed in step S117 that there is no unprocessed condition main part, the processing for one ResourceType value ends as described above. Based on the conditional expression temporarily held in step S116, the value of this ResourceType value is determined. Cluster condition information is stored in the cluster condition information storage means 1d (S118).
Then, it is confirmed whether there is another unprocessed ResourceType value (S119). If there is, the process returns to step S101 and the process is repeated until there is no unprocessed ResourceType value.

  When the above processing is completed for all ResourceType values (YES in S119), the resource aggregation processing is completed for all resources in the cluster 1.

  Returning to FIG. 4, each resource management unit 1a,..., Na transmits all cluster condition information stored in the cluster condition information storage means 1d to the grid manager 50a. The grid manager 50a stores / updates the received cluster condition information in the multiple cluster condition information storage unit 50b (S121).

  FIG. 3 shown above is an example showing the cluster condition information after performing the resource aggregation described above for the resource condition information of FIG. 2. Here, the conditions for Max Bandwidth are aggregated by logical sum. It is simple and shows that the conditions are aggregated by logical sum for the CPU.

  Next, candidate cluster search step S200 will be described.

  First, the resource management unit 1a of the cluster 1 issues a cluster search request in which desired condition information is set to the grid manager 50a (S201). When the grid manager 50a receives a cluster search request from the resource management unit 1a, the grid manager 50a searches the multiple cluster condition information storage unit 50b (S202), and sets the matched cluster as a cluster that may have a rentable resource. Those) indicating the cluster is transmitted to the resource management unit 1a as candidate cluster information (S203). The cluster resource management unit 1a receives candidate cluster information (S204).

  Next, the corresponding resource search step S300 will be described.

  The resource management unit 1a transmits a resource search request in which desired condition information is set to all or part of the received candidate cluster information (S301). The condition information set here is the same as the condition information at the time of the cluster search request in step S201. In FIG. 4, since two clusters 2 and 3 are candidate clusters, a resource search request is transmitted to both clusters 2 and 3, but generally only a part of the clusters that are candidate clusters. The resource search may be transmitted to

  If a resource search request is transmitted to more candidate clusters, the possibility that a more preferable resource can be found is increased. If a resource search is transmitted to only a part of clusters, transmission and processing of the resource search are performed. Cost efficiency can be improved.

  Upon receiving the resource search request (S302), the resource managers 2a and 3a read the resource search request condition information, and search the resource information storage means 2c and 3c using the resource search means 2b and 2c, respectively (S303). ), The corresponding resource in each cluster is searched, and the corresponding resource information is transmitted to the resource management unit 1a (S304). Each of the resource managers 1a receives the resource information (S305), and selects one from the resources indicated by the resource information (S306). This selection may be a method of selecting the resource with the best corresponding resource information, a method of selecting the resource of the resource information received earliest, or a method of randomly selecting an arbitrary resource.

  Thus, resource search can be easily performed. A resource interchange request may be made for the resource searched and selected in this way.

  According to the resource search method of the present embodiment as described above, the condition information of the resources in each cluster is aggregated in each cluster and registered in the grid manager 50a. (Resource) can make a cluster search request to the grid manager 50a, so that (high) cluster candidates that may have resources that meet the conditions can be obtained. Therefore, resource interchange is possible without searching all clusters. It became easier to find resources to receive.

  Note that the multiple cluster condition information registered in the multiple cluster condition information storage unit 50b is simplified by the information aggregation step S10. Therefore, even if some candidate clusters are searched in the corresponding resource search step S30, the resources are not found. It may not be found. However, on the other hand, the cluster having the requested resource is always selected as one of the candidate clusters in the candidate cluster search step S20. Therefore, if a corresponding resource search request is transmitted to all candidate clusters in the corresponding resource search step S30, if the requested resource exists, it can be surely discovered by resource search.

  In addition, the transmission of the corresponding resource search request to the corresponding resource search step S30 to some candidate clusters can be repeated until the corresponding resource can be found or the candidate clusters are exhausted. Such a search for a plurality of candidate clusters increases the amount of communication in the corresponding resource search step. However, assuming that there are a large number of resources in each cluster, the information aggregation step S10 allows each cluster to be The amount of communication transmitted from the resource manager to the grid manager can be reduced, and the processing amount of the entire resource search can also be reduced.

  In this embodiment, the resource condition information aggregating means 1e belongs to a cluster. However, the resource condition information aggregating means 1e is installed on the grid manager side and information transmitted from the cluster manager to the grid manager is used as the resource condition information. It may be performed on the side. In this case, the communication amount is not reduced, but the same effect can be obtained with respect to the reduction of the processing amount.

  Next, modifications of the above embodiment will be described below.

  FIG. 6 shows the configuration of this modification, and knowledge base storage means 1f,..., Nf are added to the resource management units 1a,. The point provided is the difference in configuration, and the difference is that step S112 in the resource aggregation step S100 is performed by a more advanced resource aggregation method shown in FIG. 10 described later.

  In this modification, a new predicate “is-subset-of” indicating that the main part is a subset of arguments is added, so that the cluster condition information is compared with the above embodiment as shown in FIG. And simple configuration.

  FIG. 8 is an example of knowledge information stored in the knowledge base storage means 1f,..., Nf. The knowledge base is expressed as a directed acyclic graph, and an arrow is drawn from a child node expressing a set included in the inclusion relation of the set to a parent node expressing the set included. In the example of FIG. 8, CPUArch-X is a set consisting of CPUModel-A, CPUModel-B, CPUModel-C, and CPUModel-D, and CPUArch-Y is a set consisting of CPUModel-E and CPUModel-F. CPUVendor-I is a set consisting of CPUModel-A, CPUModel-B and CPUModel-D, CPUVender-J is a set consisting of CPUModel-C and CPUModelE, and CPUVender-K is a set consisting of CPUModel-F. In other words, CPUType-M indicates that it is a set of CPUArch-X and CPUArch-Y by a graph structure.

  Compared with the above embodiment, the present modification is characterized in that the simplification process of FIG. 10 is performed instead of step S112 in the flow of FIG. 5, and this point will be described.

  In the simplification process, first, the knowledge base storage unit 1f is referred to obtain a node that expresses a value stored when the predicate is "not a numerical value" (hereinafter referred to as a stored value) (S61). Note that values that do not exist in the knowledge base are not subject to simplification.

  Next, the entire path of the knowledge base graph is traced from the obtained node in the parent direction. At this time, information [[numerical value, numerical value,..., Numerical value), numerical value] is recorded in each node on the route (S62). (Numerical value, numerical value,..., Numerical value) is information of a node as a starting point, and is used as a subscript of each stored value temporarily stored in step S109. The last number in [(number, number, ..., number), number] is the maximum path length, and is updated when the node is traced. In this case, the starting point is [(1), 0], [(2), 0], ..., [(n), 0]. For example, a graph generated using the knowledge base of FIG. 8 for the resource condition information CPU of FIG. 2 is as shown in FIG.

  It is confirmed whether or not there is a node with a value of [(1, 2,..., N), x] in the information of the node remaining after the route has been traced (S63). Corresponds to a set containing all of the stored values. In FIG. 9, the nodes CPUArch-X and CPUType-M correspond to this. Since there are generally a plurality of such nodes, the node having the shortest maximum path length from the starting point is selected, and the stored value is rewritten with the value (S65). In FIG. 9, CPUArch-X corresponds, and each stored value composed of CPUModel-A, CPUModel-B, and CPUModel-C is rewritten into one CPUArch-X.

  On the other hand, when there is no node having a value of [(1, 2,..., N), x], a set including all of 1 to n can be created by selecting a plurality of nodes from the joining point and the starting point. The node set having the smallest number of nodes is selected from these node sets (S64). The shortest maximum path length is selected from the selected minimum ones, and the stored value is replaced with that value (S65).

Then, a predicate “is-subset-of” indicating that the main part is a subset of arguments is used as a predicate, and a conditional expression using a stored value as an argument is generated (S66). If there are a plurality of stored values, the union of arguments is assumed to be an argument.

  The above-described processing for each condition main part is performed for all the condition main parts, and the set of conditional expressions output is stored in the cluster condition information storage means 1d as cluster condition information of the ResourceType value to which attention is paid.

  By using the modification described above, in addition to the effects obtained in the above-described embodiment, it is possible to further reduce the communication amount and processing amount for cluster search by the condition information aggregation processing.

  In this embodiment, the resource condition information aggregating unit 1e and the knowledge base storage unit 1f belong to a cluster, but this is installed on the grid manager side and information transmitted from the cluster manager to the grid manager is used as resource condition information. The aggregation process may be performed on the grid manager side. In this case, the communication amount is not reduced, but the same effect can be obtained for the reduction of the processing amount.

It is a block diagram which shows the structure of the cluster system which concerns on embodiment of this invention. It is a figure which shows the example of the resource condition information used for embodiment of this invention. It is a figure which shows the example of the cluster condition information used for embodiment of this invention. The flowchart which shows the resource search operation | movement which concerns on embodiment of this invention. The flowchart which shows the information aggregation operation | movement which concerns on embodiment of this invention. It is a block diagram which shows the structure of the cluster system which concerns on a modification. It is a figure which shows the example of the cluster condition information used for a modification. It is a figure which shows the example of the knowledge base information used for a modification. It is a figure which shows the example of the graph used by the array element simplification operation | movement used for a modification. The flowchart which shows the simplification operation | movement which concerns on a modification.

Explanation of symbols

1, n ... clusters 1a, 2a, ..., na ... resource management units 1b, 2b, ..., nb ... resource search means 1c, 2c, ... , Nc ... resource condition information storage means 1d, 2d, ..., nd ... cluster condition information storage means 1e, 2e, ..., ne ... resource condition information aggregation means 1f, 2f, ... .., nf... Knowledge base storage means 12, 13, 22, 23,..., N2, n3... Resource 40... Network 50. Cluster condition information storage means 100... Grid

Claims (8)

A cluster system comprising a plurality of clusters each having a resource that is a computer resource necessary for the operation of one or more computer programs, and a computer connected to each of the plurality of clusters, A resource search method for searching a resource that meets a desired condition from a first cluster from resources of another cluster,
Cluster conditions in which the condition information of each resource in the own cluster is aggregated as the condition information of the own cluster by each of the other clusters, and the aggregated condition information is associated with the cluster identification information for identifying the own cluster A resource information aggregation step for transmitting information to be stored in the computer;
By said computer, said receiving a desired condition transmitted from the first cluster, the cluster condition information their respective of said other clusters that store, compatible cluster condition information to the desired condition A candidate cluster search step of obtaining the cluster identification information corresponding to and transmitting the cluster identification information to the first cluster;
A resource search step of transmitting the desired condition to at least one of the other clusters specified by the cluster identification information by the first cluster and searching for a suitable resource;
The condition information consists of a set of a main part that is an information name and a predicate that is a numerical value corresponding to the information name,
In the resource information aggregation step, the maximum value among the upper limit values indicated by the numerical range specification of the previous description part is set as the upper limit value in the cluster condition information, and the minimum value among the lower limit values is set as the cluster condition information. A resource search method characterized in that the lower limit is set in A cluster system comprising a plurality of clusters each having a resource that is a computer resource necessary for the operation of one or more computer programs, and a computer connected to each of the plurality of clusters, A resource search method for searching a resource that meets a desired condition from a first cluster from resources of another cluster,
Cluster conditions in which the condition information of each resource in the own cluster is aggregated as the condition information of the own cluster by each of the other clusters, and the aggregated condition information is associated with the cluster identification information for identifying the own cluster A resource information aggregation step for transmitting information to be stored in the computer;
The computer receives the desired condition transmitted from the first cluster, and corresponds to the cluster condition information that matches the desired condition from the respective cluster condition information of the other cluster stored. A candidate cluster search step of obtaining cluster identification information and transmitting the cluster identification information to the first cluster;
A resource search step of transmitting the desired condition to at least one of the other clusters specified by the cluster identification information by the first cluster and searching for a suitable resource;
Each of the other clusters includes a knowledge base that stores a graph expressing an inclusion relation of a set of condition information values indicated by non-numeric values among the condition information, and is aggregated in the resource information aggregation step. In addition, a resource obtained by obtaining nodes on the graph stored in the knowledge base with condition information values indicated by non-numeric values and aggregating them using the obtained path lengths between the nodes. Search method. In a cluster system comprising a plurality of clusters each having one or more resources, and a computer connected to each of the plurality of clusters,
Each of the plurality of clusters is
First storage means for storing condition information of each resource in the own cluster;
Aggregation means for aggregating condition information of each resource in the own cluster stored in the first storage means as condition information of the own cluster;
Transmission means for transmitting cluster condition information in which the condition information aggregated by the aggregation means and cluster identification information for identifying the own cluster are associated with each other;
When receiving a search request to which condition information is added, the first storage means is searched with the condition information, and first search means for obtaining information indicating a suitable resource is provided.
The condition information consists of a set of a main part that is an information name and a predicate that is a numerical value corresponding to the information name,
The aggregation means uses the maximum upper limit value indicated by the numerical range designation in the previous description part among the condition information as the upper limit value in the cluster condition information, and the minimum lower limit value as the lower limit in the cluster condition information. Value and the calculator is
Second storage means for storing condition information of each cluster aggregated by the aggregation means from each of the clusters;
A cluster system comprising: a second search unit that receives the search request to which condition information is added, searches the second storage unit with the condition information, and obtains information for identifying a matching cluster. In a cluster system comprising a plurality of clusters each having one or more resources, and a computer connected to each of the plurality of clusters,
Each of the plurality of clusters is
First storage means for storing condition information of each resource in the own cluster;
Aggregation means for aggregating condition information of each resource in the own cluster stored in the first storage means as condition information of the own cluster;
Transmission means for transmitting cluster condition information in which the condition information aggregated by the aggregation means and cluster identification information for identifying the own cluster are associated with each other;
When receiving a search request with condition information added, the first search means searches the first storage means with the condition information, and obtains information indicating a suitable resource;
A knowledge base storing a graph expressing the inclusion relation of the set of the condition information values indicated by non-numeric values among the condition information,
The aggregation means obtains nodes on the graph stored in the knowledge base with values of condition information indicated by non-numeric values, and aggregates them using the obtained path lengths between the nodes,
The calculator is
Second storage means for storing condition information of each cluster aggregated by the aggregation means from each of the clusters;
A cluster system comprising: a second search unit that receives the search request to which condition information is added, searches the second storage unit with the condition information, and obtains information for identifying a matching cluster. A computer connected to a plurality of clusters each having one or more resources,
Storage means for storing condition information of each resource in the own cluster from each of the plurality of clusters, each of the cluster condition information aggregated as condition information of the own cluster;
Upon receiving a search request with condition information added, search means for searching the storage means with the condition information and obtaining information for identifying a matching cluster;
Transmission means for transmitting information identifying the cluster obtained by the search means to a search request source,
The condition information consists of a set of a main part that is an information name and a predicate that is a numerical value corresponding to the information name,
The upper limit value in the cluster condition information is the maximum of the upper limit values indicated by the numerical range specification of the previous description part among the condition information,
The computer is characterized in that the lower limit value in the cluster condition information is the smallest one of the lower limit values indicated by the numerical range designation in the preceding description portion among the condition information . A computer connected to a plurality of clusters each having one or more resources,
Storage means for storing condition information of each resource in the own cluster from each of the plurality of clusters, each of the cluster condition information aggregated as condition information of the own cluster;
Upon receiving a search request with condition information added, search means for searching the storage means with the condition information and obtaining information for identifying a matching cluster;
Transmission means for transmitting information identifying the cluster obtained by the search means to a search request source,
The cluster condition information, said in a graph representing the inclusion relation of the set of the values of the condition information shown in a non-numeric among the condition information, the node value of the condition information shown in a non-numeric is obtained, resulting A computer that is aggregated using the path length between nodes. In a cluster connected to a computer,
Storage means for storing condition information of each resource in the own cluster;
Aggregating means for aggregating the condition information of each resource in the own cluster stored in the storage means as the condition information of the own cluster;
Transmitting means for transmitting to the computer cluster condition information in which the condition information aggregated by the aggregation means and cluster identification information for identifying the own cluster are associated;
Upon receiving a search request with condition information added, search means for searching the storage means with the condition information and obtaining information indicating a suitable resource;
Transmission means for transmitting information identifying the resource obtained by the search means to a search request source,
The condition information consists of a set of a main part that is an information name and a predicate that is a numerical value corresponding to the information name,
The aggregation means uses the maximum upper limit value indicated by the numerical range designation in the previous description part among the condition information as the upper limit value in the cluster condition information, and the minimum lower limit value as the lower limit in the cluster condition information. A cluster characterized by a value. In a cluster connected to a computer,
Storage means for storing condition information of each resource in the own cluster;
Aggregating means for aggregating the condition information of each resource in the own cluster stored in the storage means as the condition information of the own cluster;
Transmitting means for transmitting to the computer cluster condition information in which the condition information aggregated by the aggregation means and cluster identification information for identifying the own cluster are associated;
Upon receiving a search request with condition information added, search means for searching the storage means with the condition information and obtaining information indicating a suitable resource;
Transmitting means for transmitting information identifying the resource obtained by the search means to a search request source;
A knowledge base storing a graph expressing the inclusion relation of the set of the condition information values indicated by non-numeric values among the condition information,
The aggregating means obtains nodes on the graph stored in the knowledge base with values of condition information represented by non-numeric values, and aggregates them using the obtained path lengths between the nodes. And a cluster.

Images