JP2012048330A - Cluster system and software deployment method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a cluster system with a simple structure suppressing management cost even when servers with different hardware configurations exist in a resource pool.SOLUTION: A resource manager 2 has a management interface 21, a server database 22 for storing information related to all the servers 30, a template image repository 23 for storing template disk images necessary for providing services, a driver repository 24 for storing drivers corresponding to hardware specific to each of the servers 30, a server activation processing function part 25 for constructing an environment which can provide services for the servers 30 by outputting the template disk image and driver in combination to the servers 30, and a server state monitoring function part 26.

Description

  The present invention relates to a cluster system that provides services to users, and more particularly, to a server resource expansion technique in a distributed / parallel system employing a scale-out type architecture.

  In recent years, the IP (Internet Protocol) of communication networks has been advanced for the purpose of reducing equipment maintenance costs, and communication networks and services have been provided using general-purpose servers instead of dedicated hardware such as conventional exchanges. It is common to provide. At that time, in order to obtain high availability and high reliability with only a general-purpose server, a server cluster that is an aggregate of a plurality of servers may be configured. Also, with the diversification of communication services, it is difficult to predict the number of service users in advance. Therefore, from the viewpoint of reducing equipment maintenance costs, a scale-out type architecture that starts small with a minimum of server resources and adds new server resources as the number of users expands has attracted attention (for example, Non-patent document 1).

  In the scale-out type architecture described in Non-Patent Document 1, a technique has been proposed in which server resources are aggregated as a resource pool, and a plurality of services share server resources to further increase resource utilization efficiency. Hereinafter, with reference to FIG. 8, a conventional cluster system 9 adopting a technology for improving resource utilization efficiency will be described. As shown in FIG. 8, the cluster system 9 comprises a resource pool 93 with a plurality of servers 95. In the example of FIG. 8, the unique hardware type is attached in round letters for all the servers 95 in the resource pool 93. That is, the resource pool 93 includes two servers 95 of hardware type A, four servers 95 of hardware type B, and two servers 95 of hardware type C.

  When there is a request for adding a certain service, the cluster system 9 selects one server 95 from the resource pool 93. Then, the cluster system 9 acquires a template disk image Tdi suitable for the service to be added from the repository 91 and deploys it to the server 95.

  The deployment outputs software such as an operating system (OS), middleware, and application (APL) for providing a service, and a template disk image Tdi including a driver of the server 95 to the server 95, and also an IP address. , Setting of the host name, setting of parameters, and the like.

  The template disk image Tdi can be created by deleting a setting unique to each server 95 after constructing a software environment manually or using an automatic installer in any server 95. The created template disk image Tdi is centrally managed by the repository 91 and is output to the server 95 when a service expansion request is made as described above.

Nishimura Gosei, Irie Michio, Kaneko Masashi, Iio Masami, "A Study for Improving Network Server Resource Utilization Efficiency", 10th Network Software Study Group, Ishigaki City Civic Hall, June 3-4, 2010

  However, the conventional cluster system 9 has the following problems. In the cluster system 9, when adding a service, the template disk image Tdi transferred to the blank server 95 includes a driver corresponding to the server 95 that created the template disk image Tdi. That is, if the blank server 95 to be deployed and the server 95 that created the template disk image Tdi are of different hardware types, the driver required for the blank server 95 is not included in the template disk image Tdi. 95 falls into the situation where it cannot start. A server on which no template disk image Tdi is deployed is referred to as a “blank server”.

  At first glance, it seems that such a problem can be solved by unifying the hardware types of all the servers 95 in the resource pool 93. However, such unification of hardware is not always acceptable in operation. Furthermore, this kind of hardware unification means that if the server 95 of the same model as before cannot be procured due to the end of life (EOL), all servers 95 in the resource pool will be replaced with new models. It needs to be changed and is not realistic.

  Here, it is also conceivable to individually generate the template disk image Tdi according to the hardware type. In the example of FIG. 8, only one type of service is provided such that the hardware type is a template disk image for A, the hardware type is a template disk image for B, and the hardware type is a template disk image for C. A plurality of template disk images Tdi are required. Furthermore, since the template disk image Tdi includes an operating system, an application, and the like, it is general that the template disk image Tdi has a large capacity of several gigabytes or more. Therefore, since the conventional cluster system 9 needs to store a large number of large-capacity template disk images Tdi, a large storage capacity is required, and there is a problem that the configuration of the cluster system 9 is enlarged. Furthermore, in the conventional cluster system 9, when software such as an operating system and an application is updated, it is necessary to recreate a large number of template disk images Tdi, resulting in an increase in management cost.

  In addition, it is said that the hardware difference of each server 95 can be concealed by the hypervisor by applying the server virtualization technology. However, the hypervisor has a large overhead related to network communication, and when applied to the cluster system 9 that frequently uses communication, the performance is significantly reduced.

  SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a cluster system and a software deployment method that solves the above-described problems and reduces management costs with a simple configuration even when servers with different hardware configurations exist in a resource pool. And

  In view of the problems described above, the cluster system according to the first invention of the present application manages a resource pool composed of a plurality of servers that provide services to users and the services provided by all the servers in the resource pool. A cluster system including a resource manager, wherein the resource manager uniquely identifies the server, a service type indicating a service provided by the server, or the server is waiting A server database that stores a server state in association with a hardware type indicating hardware specific to the server, and a template disk image that includes the service type and software that provides the service indicated by the service type Template disc image including location A correspondence table is stored, and the storage location of the template disk image is searched from the template disk image correspondence table using the service type included in the template disk image acquisition request indicating the request for the template disk image as a search condition. A driver including a template image repository that outputs a template disk image stored in a template disk image storage location, the service type, the hardware type, and a driver storage location corresponding to hardware unique to the server A correspondence table is stored, and the storage location of the driver is searched from the driver correspondence table using the service type and hardware type included in the driver acquisition request indicating the driver request as search conditions. A driver repository that outputs drivers stored in a driver storage location, a server status monitoring function unit that monitors all servers in the resource pool and updates the server status of the server database, and the service type that is to be added Is output, the template disk image acquisition request including the service type of the service extension request is output to the template image repository, the template disk image is acquired from the template image repository, and The driver acquisition request including the service type of the service expansion request and the hardware type of the one server that is in the standby state in the server database is output to the driver repository, and the driver A server activation processing function unit that obtains the driver from the repository and outputs the driver to the one server, and the server receives a template disk image and a driver from the resource manager and inputs the template disk A start agent that deploys the image and the driver to the server is provided.

  According to this configuration, the cluster system stores a template image that does not include a driver. Then, when adding a service, the cluster system acquires a driver corresponding to the server, and outputs the driver together with the template image to the server. The cluster system deploys the template image and the driver in the server, and dynamically constructs an environment necessary for providing the service.

  As a result, the cluster system can reduce the number of template images as compared to the case where template images are stored according to the hardware types of all the servers in the resource pool. Therefore, the cluster system does not require a large storage capacity for storing the template image, and allows a simple configuration. Furthermore, even when the software is updated, the cluster system can reduce the number of template images that need to be recreated, thereby reducing the management cost.

  Further, in the cluster system according to the second invention of the present application, the server activation processing function unit includes a service addition request input unit to which the service addition request is input, and the server state of the server database is in the standby state. A standby server selection unit that selects one of the servers from the server, and generates the template disk image acquisition request and outputs the request to the template image repository, and from the template image repository according to the template disk image acquisition request A template image acquisition unit for acquiring the template disk image; a driver acquisition request for generating and outputting the driver acquisition request to the driver repository; and a driver for acquiring the driver from the driver repository in response to the driver acquisition request. A template image driver output unit that outputs the acquisition unit, the template disk image acquired by the template image acquisition unit, and the driver acquired by the driver acquisition unit to one server selected by the standby server selection unit; Are preferably provided.

  Further, in the cluster system according to the third invention of the present application, the standby server selection unit selects the one server having the minimum communication delay from the servers in which the server state of the server database is the standby. It is preferable to do.

  Further, in the software deployment method according to the fourth invention of the present application, the server activation processing function unit includes a service expansion request input step in which the service expansion request is input, and the server status is the standby with respect to the server database. A standby server inquiry step for inquiring about the server that is in standby, and a standby for selecting one of the servers from the server in which the server state of the server database is in the standby state based on a result of an inquiry to the server database A server selection step, a template image acquisition request step for generating the template disk image acquisition request and outputting it to the template image repository, and the template image repository in response to the template disk image acquisition request. A template image acquisition step for acquiring the template disk image from the source, a driver acquisition request step for generating the driver acquisition request and outputting it to the driver repository, and in response to the driver acquisition request, the driver from the driver repository. A template image driver that outputs the acquired driver acquisition step, the template disk image acquired by the template image acquisition unit, and the driver acquired by the driver acquisition unit to one server selected by the standby server selection unit And an output step.

  According to the present invention, since the number of template images can be reduced, a simple configuration is possible without requiring a large storage capacity, and the number of template images that need to be recreated even when software is updated. Management costs can be reduced.

1 is a schematic diagram showing an outline of a cluster system according to an embodiment of the present invention. 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 data item of the server database of FIG. It is a figure which shows the data item of the template disc image corresponding table memorize | stored in the template image repository of FIG. It is a figure which shows the data item of the driver corresponding table memorize | stored in the driver repository of FIG. It is a block diagram which shows the structure of the server starting process function part of FIG. It is a flowchart which shows operation | movement of the cluster system of FIG. It is the schematic which shows the outline of the conventional cluster system.

(Outline of this embodiment)
The outline of the cluster system 1 according to the embodiment of the present invention will be described below with reference to FIG. The cluster system 1 is a system that employs a scale-out type architecture in which a small start is performed by a minimum number of servers 30 and a new service is added according to the expansion of users. Here, examples of services provided by the cluster system 1 include asynchronous network services such as HTTP (HyperText Transfer Protocol), SMTP (Simple Mail Transfer Protocol), POP (Post Office Protocol), and FTP (File Transfer Protocol). It is done. Further, as a service provided by the cluster system 1, for example, a network service having a strong demand for real-time performance (throughput, delay) such as SIP (Session Initiation Protocol) can be cited. In FIG. 1, a part of the cluster system 1 is omitted for the sake of simplicity.

  As shown in FIG. 1, the resource pool 3 is a collection of server resources, and includes, for example, six servers 30. Here, the hardware type unique to the server 30 is shown in circles. That is, the resource pool 3 includes one server 30 of hardware type 1, two servers 30 of hardware type 2, and three servers 30 of hardware type 3.

  The template image repository 23 stores a template disk image Tdi including software necessary for providing a service, such as an operating system or an application, except for the driver Dr.

  The driver repository 24 stores a driver Dr according to the hardware type of the server 30. In the example of FIG. 1, the driver repository 24 stores three types of drivers Dr, “driver 1”, “driver 2”, and “driver 3”. Here, “driver 1” corresponds to the hardware type 1 server 30, “driver 2” corresponds to the hardware type 2 server 30, and “driver 3” corresponds to the hardware type 3 server 30. I will do it.

  Hereinafter, an example of adding a new service using the server 30 of the hardware type 2 will be described. In this case, the cluster system 1 acquires the template disk image Tdi necessary for providing the service to be added from the template image repository 23. Further, the cluster system 1 obtains “driver 2” corresponding to the hardware type 2 server 30 from the driver repository 24. Then, the cluster system 1 outputs the acquired template disk image Tdi and “driver 2” to the server 30. Then, the cluster system 1 deploys the template disk image Tdi and “driver 2” in the server 30.

  In addition, deployment refers to outputting a template disk image Tdi and a driver Dr to the server 30 and performing settings specific to the server 30 such as IP address setting, host name setting, parameter setting, etc. This is to construct an environment capable of providing

(Embodiment of the present invention)
Hereinafter, the service providing system 100 including the cluster system 1 will be described with reference to FIG. As shown in FIG. 2, the service providing system 100 includes a cluster system 1, a maintenance person terminal 5, and a user terminal 7. Here, the service providing system 100 connects the cluster system 1 and the maintenance person terminal 5 via a maintenance network NW1 such as the Internet or an intranet. In addition, the service providing system 100 connects the cluster system 1 and the user terminal 7 via the service network NW3. Hereinafter, the configurations of the maintenance person terminal 5 and the user terminal 7 will be described first, and then the configuration of the cluster system 1 will be described.

[Configuration of maintenance personnel terminal]
The maintenance person terminal 5 is a terminal device such as a computer or a mobile phone that performs maintenance of the cluster system 1. For example, the maintenance person terminal 5 outputs to the cluster system 1 a service addition request including the type of service for which the addition is requested according to the instruction of the maintenance person.

  Further, the maintenance person terminal 5 constructs a server cluster (not shown) for providing a service and adds the server 30 to the server cluster, for example, according to an instruction from the maintenance person. This server cluster is a server group composed of servers 30 currently used to provide individual services. For example, when three servers 30 are used to provide “service A”, it can be said that “service A” is provided by a server cluster including three servers 30. Therefore, adding the server 30 to the server cluster means adding the server 30 in the resource pool 3 to the server cluster and expanding the scale of the server cluster.

  In addition, the maintenance person terminal 5 stops a service provided by a certain server 30 according to an instruction from the maintenance person, for example. Further, the maintenance person terminal 5 registers, for example, a server database 22 (FIG. 3), a template disk image correspondence table 230 (FIG. 4), and a driver correspondence table 240 (FIG. 5), which will be described later, according to instructions from the maintenance person. Update or delete. In addition, since these processes are the same as that of a prior art, description is abbreviate | omitted.

[User terminal configuration]
The user terminal 7 is a terminal device such as a computer or a mobile phone used by the user. That is, the user uses the service provided by the cluster system 1 using the user terminal 7.

[Cluster system configuration]
The cluster system 1 includes a resource pool 3 and a resource manager 2. Here, the cluster system 1 connects the resource manager 2 and the six servers 30 by the management network NW2.

Of the six servers 30, the first server 30 and server 30 _1, the five th server 30 and server 30 _5, the six-th server 30 and server 30 _6. In FIG. 2, the _second to _fourth servers 30 ₂ to 304 are omitted in order to make the drawing easier to see.

<Resource manager>
Hereinafter, the resource manager 2 will be described. As shown in FIG. 2, the resource manager 2 manages services provided by all the servers 30 in the resource pool 3, and includes a management interface 21, a server database 22, a template image repository 23, a driver repository. 24, a server activation processing function unit 25, and a server state monitoring function unit 26.

  The management interface 21 is an interface with the maintenance person terminal 5. That is, the management interface 21 inputs an instruction from the maintenance person terminal 5 and executes an appropriate process according to this command in conjunction with other function units. For example, when a service addition request is input from the maintenance person terminal 5, the management interface 21 outputs this service addition request to the server activation processing function unit 25.

  Hereinafter, the server database 22 will be described with reference to FIG. 3 (see FIG. 2 as appropriate). The server database 22 is a database that stores information regarding all the servers 30 in the resource pool 3, and stores a server ID (server identifier) 22A, a server state 22B, and a hardware type 22C in association with each other.

The server ID 22A is an ID (IDentification) that uniquely identifies all the servers 30 in the resource pool 3. In the example of FIG. 3, it is shown that unique IDs “server 1” to “server 6” are registered as the server ID 22A. The servers 30 ₁ to 30 ₆ correspond to “server 1” to “server 6”, respectively.

  The server state 22B is information indicating a service type indicating a service provided by the server 30 or that the server 30 is waiting. In the example of FIG. 3, “Server 1” provides “Service A”, and “Server 4” and “Server 5” provide “Service B”. The server state 22B indicates that “server 2”, “server 3”, and “server 6” do not provide any service and are “standby”.

  The hardware type 22 </ b> C is information indicating a hardware type indicating hardware unique to the server 30. In the example of FIG. 3, “Server 1” is hardware type 1, “Server 2” and “Server 3” are hardware type 2, and “Server 4” to “Server 6” are hardware type 3. Indicates that there is. In FIG. 3, hardware type 1 to hardware type 3 are abbreviated as “hard 1” to “hard 3”.

  The server database 22 receives an inquiry about the “standby” server 30 from a server activation processing function unit 25 described later. Then, as a result of the inquiry, the server database 22 acquires the server ID and hardware type of the server 30 whose server state 22B is “standby” and outputs the server ID to the server activation processing function unit 25.

  Of the data items in the server database 22, the server ID 22A and the hardware type 22C are manually set by a maintenance person. On the other hand, the server state 22B is updated as needed by the server state monitoring function unit 26 described later.

  Hereinafter, the template image repository 23 will be described with reference to FIG. 4 (see FIG. 2 as appropriate). As shown in FIG. 4, the template image repository 23 stores a template disk image correspondence table 230 including a service type 230A and a template disk image storage location 230B.

  The service type 230A is information indicating a service provided to the user. The example in FIG. 4 indicates that “service A” and “service B” are registered as the service type 230A.

  The template disk image storage location 230B is information indicating the storage location of the template disk image Tdi required when providing the service indicated by the service type 230A. In the example of FIG. 4, the template disk image Tdi stored in “file: //path/to/serviceA.img” corresponds to “service A”. In addition, the template disk image Tdi stored in “file: //path/to/serviceB.img” corresponds to “service B”.

  Here, the template disk image storage location 230 </ b> B is not particularly limited as long as it is accessible to the resource manager 2 and the server 30 in the resource pool 3. In the present embodiment, the template image repository 23 includes a storage (not shown), and the template disk image Tdi is stored in this storage.

  The template image repository 23 searches the template disk image storage table 230 for the template disk image storage location 230 using the service included in the template disk image acquisition request as a search condition (search key). Then, the template image repository 23 acquires the template disk image Tdi stored at the position indicated by the searched template disk image storage location 230 and outputs it to the server activation processing function unit 25.

  The template disk image acquisition request is information indicating that a template disk image Tdi is requested from the template image repository 23, and is input from the server activation processing function unit 25.

  The template disk image Tdi is generated in advance so as to include software (operating system, application) necessary for providing each service. Therefore, the template disk image Tdi can be generated by a method similar to the conventional method except that it does not include any driver Dr. The template disk image correspondence table 230 is manually set by a maintenance person.

  Hereinafter, the driver repository 24 will be described with reference to FIG. 5 (see FIG. 2 as appropriate). As shown in FIG. 5, the driver repository 24 stores a driver correspondence table 240 including a service type 240A, a hardware type 240B, and a driver storage location 240C.

  The service type 240A is a data item similar to the service type 230A in FIG. The hardware type 240B is a data item similar to the hardware type 22C in FIG.

  The driver storage location 240 </ b> C is information indicating the storage location of the driver Dr corresponding to the hardware specific to the server 30 when the service is provided by the server 30. In the example of FIG. 5, when “service A” is provided by the server 30 of “hardware 1”, the driver Dr stored in “file: //path/to/A1.lib” corresponds. In addition, when “service B” is provided by the server 30 of “hardware 2”, the driver Dr stored in “file: //path/to/B2.lib” corresponds.

  Here, the driver storage location 240 </ b> C is not particularly limited as long as it is accessible to the resource manager 2 and the server 30 in the resource pool 3. In the present embodiment, the driver repository 24 includes a storage (not shown), and the driver Dr is stored in this storage.

  The driver repository 24 searches the driver storage table 240C for the driver storage location 240C using the service and hardware type included in the driver acquisition request as search conditions (search keys). Then, the driver repository 24 acquires the driver Dr stored at the position indicated by the searched driver storage location 240 </ b> C and outputs the driver Dr to the server activation processing function unit 25.

  The driver acquisition request is information indicating that the driver repository 24 is requested for the driver Dr, and is input from the server activation processing function unit 25. The driver correspondence table 240 is manually set by a maintenance person.

  Hereinafter, the server activation processing function unit 25 will be described with reference to FIG. 6 (see FIG. 2 as appropriate). The server activation processing function unit 25 deploys the template disk image Tdi and the driver Dr to the server 30 in the resource pool 3 and constructs an environment in which the service can be provided in the server 30. Therefore, as shown in FIG. 6, the server activation processing function unit 25 includes a service addition request input unit 251, a standby server selection unit 252, a template image acquisition unit 253, a driver acquisition unit 254, and a template image driver. And an output unit 255.

  The service extension request input unit 251 receives the service extension request from the maintenance person terminal 5. Then, the service extension request input unit 251 outputs the service type included in the service extension request to the template image acquisition unit 253 and the driver acquisition unit 254.

  The standby server selection unit 252 inquires of the server database 22 about the server 30 whose server state 22B is “standby”. Then, the standby server selection unit 252 acquires the server ID and hardware type of the server 30 that is “standby” as the inquiry result. In the example of FIG. 3, the standby server selection unit 252 acquires the server IDs and hardware types of “server 2”, “server 3”, and “server 6” from the server database 22.

  Here, when there are a plurality of “standby” servers 30, the standby server selection unit 252 selects one server 30 from them. For example, the standby server selection unit 252 selects one server 30 that is first entered in the server database 22 (“server 2” in FIG. 3). Further, for example, the standby server selection unit 252 randomly selects one server 30 in the server database 22 whose server state 22B is “standby”.

  In addition to the simple selection method described above, the standby server selection unit 252 can also select one server 30 that minimizes communication delay. Specifically, the standby server selection unit 252 manages server location information such as the network topology or the physical location of the server 30 in the resource pool 3. Based on this server location information, the standby server selection unit 252 then selects the server 30 with the least communication delay in the resource pool 3, such as being connected to the same hub or directly connected via a network cable. Select one. Thus, the cluster system 1 can improve the service providing efficiency by selecting the server 30 that minimizes the communication delay.

  Then, the standby server selection unit 252 outputs the hardware type of the selected one server 30 to the driver acquisition unit 254. In the server database 22, when there is one server 30 in which the server state 22B is "standby", it goes without saying that the standby server selection unit 252 selects the server unconditionally.

  The template image acquisition unit 253 receives the service type from the service addition request input unit 251. Then, the template image acquisition unit 253 generates a template disk image acquisition request including this service type and outputs it to the template image repository 23. Furthermore, in response to this template disk image acquisition request, the template image acquisition unit 253 acquires the template disk image Tdi from the template image repository 23. Thereafter, the template image acquisition unit 253 outputs the acquired template disk image Tdi to the template image driver output unit 255.

  The driver acquisition unit 254 receives a service from the service expansion request input unit 251 and a hardware type from the standby server selection unit 252. Then, the driver acquisition unit 254 generates a driver acquisition request including this service and hardware type, and outputs the driver acquisition request to the driver repository 24. Further, in response to this driver acquisition request, the driver acquisition unit 254 acquires the driver Dr from the driver repository 24. Thereafter, the driver acquisition unit 254 outputs the acquired driver Dr to the template image / driver output unit 255.

  The template image / driver output unit 255 receives the template disk image Tdi from the template image acquisition unit 253 and the driver Dr from the driver acquisition unit 254. Then, the template image / driver output unit 255 outputs the template disk image Tdi and the driver acquisition unit 254 to the server 30 selected by the standby server selection unit 252.

  Returning to FIG. 2, the description will be continued. The server status monitoring function unit 26 monitors all the servers 30 in the resource pool 3 and updates the server status 22B of the server database 22. Specifically, the server status monitoring function unit 26 outputs a status query indicating a server status query to each server 30, and updates the server status 22B of the server database 22 in response to the status query.

<Server>
Hereinafter, the server 30 will be described. The server 30 provides various services to the user, and includes an activation agent 31, a local storage unit 32, and a calculation processing function unit 33.

  The activation agent 31 communicates with the resource manager 2. Specifically, in response to a status inquiry from the server status monitoring function unit 26, a response indicating that the service type being executed on the server 30 or no service is provided and “standby” is provided. To do.

  In addition, the activation agent 31 deploys the template disk image Tdi and the driver Dr input from the server activation processing function unit 25. For example, the activation agent 31 applies the setting specific to the server 30 input from the server activation processing function unit 25 to the template disk image Tdi. Then, the activation agent 31 stores the template disk image Tdi and the driver Dr in the local storage unit 32 as an activation disk image.

  At this time, the activation agent 31 can deploy the driver Dr at a predetermined deployment position (for example, “/ lib / modules /”). Further, the activation agent 31 may mount the driver Dr at a deployment position (for example, “/ lib / modules /”) when the operating system is activated after the driver Dr is deployed at another location.

  Further, the activation agent 31 applies a setting specific to the server 30 to the template disk image Tdi, and then outputs an activation instruction to the calculation processing function unit 33. The settings unique to each server 30 include, for example, network settings such as an IP address, netmask, and DNS server, middleware settings such as buffer size and the number of simultaneous connections, and operating system parameters.

  The local storage unit 32 stores the above-described startup disk image, and is, for example, a nonvolatile memory such as a hard disk or a volatile memory. In addition to the boot disk image, the local storage unit 32 may store various data necessary for providing the service.

  The calculation processing function unit 33 is an arithmetic device including a CPU (Central Processing Unit) and a main storage device. When the activation instruction is input from the activation agent 31, the calculation processing function unit 33 activates the operating system using the activation disk image stored in the local storage unit 32 and provides a service.

[Cluster system operations]
Hereinafter, the operation of the cluster system 1 of FIG. 2 will be described with reference to FIG. 7 (see FIGS. 2 and 6 as appropriate). Here, an example in which “service A” is added to “server 2” using the server database 22 of FIG. 3, the template disk image correspondence table 230 of FIG. 4, and the driver correspondence table 240 of FIG. .

  First, when the maintenance person instructs to add “service A”, the maintenance person terminal 5 generates a service addition request indicating “service A” and inputs it to the server activation processing function unit 25. That is, the server activation processing function unit 25 receives a service expansion request including “service A” from the maintenance person terminal 5 by the service expansion request input unit 251 (step S1: service expansion request input step).

  The server activation processing function unit 25 uses the standby server selection unit 252 to query the server database 22 for the server 30 whose server status is “standby” (step S2: standby server inquiry step).

  In accordance with the processing in step S2, the server database 22 outputs server IDs for three servers “server 2”, “server 3”, and “server 6” to the standby server selection unit 252 (see FIG. 3). Then, the server activation processing function unit 25 selects, for example, “server 2” that is the first entry in the server database 22 from the three servers 30 by the standby server selection unit 252 (step S3: standby server). Selection step).

  The server activation processing function unit 25 generates a template disk image acquisition request including “service A” by the template image acquisition unit 253, and outputs the template disk image acquisition request to the template image repository 23 (step S4: template image acquisition request step).

  In accordance with the processing in step S4, the template image repository 23 searches the template disk image correspondence table 230 for the storage location of the template disk image using “service A” as a search condition (search key), and stores this location in the storage location. The stored template disk image Tdi is output. That is, the server activation processing function unit 25 uses the template image acquisition unit 253 to acquire the template disk image Tdi stored in “file: //path/to/serviceA.img” from the template image repository 23 (step S5). : Template image acquisition step).

  The server activation processing function unit 25 generates a driver acquisition request including “service A” and “hardware 2” by the driver acquisition unit 254, and outputs the driver acquisition request to the driver repository 24 (step S6: driver acquisition request step).

  In accordance with the processing of step S4, the driver repository 24 uses the “service A” and “hardware 2” as search conditions, and stores the driver Dr storage location “file: //path/to/A2.lib” from the driver correspondence table 240. ”And the driver Dr stored in this storage location is output. That is, the server activation processing function unit 25 acquires the driver Dr stored in “file: //path/to/A2.lib” from the driver repository 24 by the driver acquisition unit 254 (step S7: driver acquisition step). ).

  The server activation processing function unit 25 selects the template disk image Tdi acquired by the template image acquisition unit 253 and the driver Dr acquired by the driver acquisition unit 254 by the template image / driver output unit 255 by the standby server selection unit 252. Output to the server 2 "(step S8: template image driver output step).

Server 30 _2, by the activation agent 31, by applying the specific settings to the server 30 ₂ to the template disk image Tdi, stores the template disk image Tdi and driver Dr as a startup disk image locally storing unit 32. Then, the server 30 _2, by the activation agent 31, and outputs an activation instruction to the calculation processing function unit 33. Thereafter, the server 30 _2, by the calculation processing function unit 33, using the boot disk image stored in the local storage unit 32 activates the operating rate ting systems to provide services.

  In FIG. 7, the server activation processing function unit 25 has been described as executing the processes of steps S6 and 7 after executing the processes of steps S4 and 5, but the present invention is not limited to this. For example, the server activation processing function unit 25 may execute the processes in steps S4 and 5 after executing the processes in steps S6 and S7.

  As another example, consider a case where “server 6” is selected instead of “server 2” in the process of step S3. In this case, the server activation processing function unit 25 stores the template disk image Tdi corresponding to “service A”, that is, “file: //path/to/serviceA.img” for “server 6”. A template disk image Tdi is output. Further, the server activation processing function unit 25 outputs the driver Dr corresponding to “service A” and “hardware 3”, that is, the driver Dr stored in “file: //path/to/A3.lib”. Will do.

  As described above, the cluster system 1 according to the embodiment of the present invention includes the template disk image Tdi corresponding to the service and the server that provides the service even when the server 30 of different hardware types is included in the resource pool 3. By combining with the driver Dr corresponding to the hardware type, the number of template disk images Tdi stored in the storage can be reduced. As a result, the cluster system 1 can be configured in a simple manner without requiring a storage having a large storage capacity, and even when the software is updated, the management cost associated with this update can be suppressed.

  In the resource manager 2 described above, the management interface 21, the server database 22, the template image repository 23, the driver repository 24, the server activation processing function unit 25, and the server state monitoring function unit 26 include an input / output interface, a communication interface, and storage means. You may implement | achieve by the computer provided with. At this time, each function of the resource manager 2 may be realized by program execution processing by the CPU, or may be realized by dedicated hardware.

DESCRIPTION OF SYMBOLS 1 Cluster system 2 Resource manager 21 Management interface 22 Server database 23 Template image repository 24 Driver repository 25 Server start processing function part 251 Service expansion request input part 252 Standby server selection part 253 Template image acquisition part 254 Driver acquisition part 255 Template image driver Output unit 26 Server state monitoring function unit 3 Resource pool 30 Server 31 Startup agent 32 Local storage unit 33 Calculation processing function unit

Claims (4)

A cluster system comprising: a resource pool composed of a plurality of servers that provide services to users; and a resource manager that manages the services provided by all servers in the resource pool,
The resource manager
A server identifier for uniquely identifying the server, a service type indicating a service provided by the server or a server state indicating that the server is waiting, and a hardware type indicating hardware specific to the server A server database that stores
A template disk image correspondence table including the service type and a storage location of a template disk image including software providing the service indicated by the service type is stored, and a template disk image acquisition request indicating the template disk image request is stored. A template image repository for searching the template disk image storage location from the template disk image correspondence table using the included service type as a search condition, and outputting the template disk image stored in the searched template disk image storage location; ,
Storing a driver correspondence table including the service type, the hardware type, and the storage location of the driver corresponding to the hardware specific to the server; and the service type included in the driver acquisition request indicating the driver request, and Using the hardware type as a search condition, search for the storage location of the driver from the driver correspondence table, and output the driver stored in the storage location of the searched driver,
A server status monitoring function unit that monitors all servers in the resource pool and updates the server status of the server database;
A service addition request indicating the service type to be added is input, the template disk image acquisition request including the service type of the service addition request is output to the template image repository, and the template disk image is extracted from the template image repository. And acquiring the driver acquisition request including the service type of the service addition request and the hardware type of the one server that is in the standby state in the server database to the driver repository, A server activation processing function unit that acquires the driver from a driver repository and outputs the driver to the one server;
The server
A cluster system comprising: a startup agent that receives a template disk image and a driver from the resource manager and deploys the input template disk image and the driver to the server. The server activation processing function unit
A service expansion request input unit to which the service expansion request is input;
A standby server selection unit that selects one of the servers from the server in which the server state of the server database is the standby;
Generating and outputting the template disk image acquisition request to the template image repository, and in response to the template disk image acquisition request, a template image acquisition unit that acquires the template disk image from the template image repository;
A driver acquisition unit that generates the driver acquisition request and outputs the driver acquisition request to the driver repository, and acquires the driver from the driver repository according to the driver acquisition request;
A template disk image acquired by the template image acquisition unit and a driver acquired by the driver acquisition unit; a template image / driver output unit that outputs the server selected by the standby server selection unit;
The cluster system according to claim 1, further comprising:   The said standby server selection part selects the said 1 server from which the communication delay becomes the minimum from the said server in which the said server state of the said server database is the said standby. Cluster system. A software deployment method in the cluster system according to claim 2,
The server activation processing function unit is
A service expansion request input step in which the service expansion request is input;
A standby server query step for querying the server database for the server whose server status is said to be standby.
A standby server selection step of selecting one of the servers from the servers in which the server status of the server database is the standby, based on a query result for the server database;
A template image acquisition request step of generating the template disk image acquisition request and outputting it to the template image repository;
In response to the template disk image acquisition request, a template image acquisition step of acquiring the template disk image from the template image repository;
A driver acquisition request step of generating the driver acquisition request and outputting it to the driver repository;
In response to the driver acquisition request, a driver acquisition step of acquiring the driver from the driver repository;
A template image driver output step for outputting the template disk image acquired by the template image acquisition unit and the driver acquired by the driver acquisition unit to one server selected by the standby server selection unit;
The software deployment method characterized by performing.

Images