JP2012146093A - Software image distribution method, repository device, server, and system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable servers to be quickly incorporated even when many servers are incorporated in a cluster from a resource pool.SOLUTION: When a repository device 2100 in a cluster system 2000 distributes a software image to servers 2310, the distribution is performed by multicasting via a management network 2200. Then, when the management network 2200 does not support multicasting, first, the repository device 2100 distributes the software image to a server 2310 as a child node in a binary tree among servers 2310 to be added, and thereafter the server 2310 having received the software image transfers the software image to a server as a child node of the server itself.

Description

  The present invention relates to a software image distribution technique to each server of a cluster system.

  In recent years, in the field of Web server hosting and the like, a form called IaaS (Infrastructure as a Service) etc. in which server management providers such as data centers collectively manage servers and host service providers have become common. (See Non-Patent Document 1). Here, it is necessary to prepare a server that includes a certain amount of standby surplus resources in order to respond quickly to requests when service demand increases or when a server providing that service leaves due to a failure. There is. In recent years, network server cluster architectures that can add and delete server resources dynamically without interruption have emerged, and further operational flexibility and cost reduction of facilities that make up communication networks are expected ( Non-patent document 2).

  In response to such a request, the surplus resources of each network server cluster are aggregated as a resource pool, and when the demand increases or when resources lost due to failures or disasters are replenished, the necessary servers from the resource pool are added to each cluster. In addition to this, a technique for improving the overall facility utilization efficiency is being studied.

  When a server is added to a cluster from a shared resource pool in which surplus resources are aggregated in this way, it is not known in advance what software configuration the server held in the resource pool will take. For this reason, the server of the resource pool is generally in a state that has no software. Therefore, when adding a resource, a software image including a software configuration corresponding to each service is transferred from the repository to a server to be added. Then, after making the necessary settings for this server, it is necessary to incorporate it into the cluster.

Amazon Elastic Compute Cloud (Amazon EC2), [online], [Search December 3, 2010], Internet, <URL: http: //aws.amazon.com/ec2/> Mobicents, [online], [Search December 3, 2010], Internet, <URL: http://www.mobicents.org/>

  Since the software image includes all of the operating system, various middleware, data, and the like, the software image has a large capacity ranging from several to several tens of GB and requires a certain amount of time for transfer. When it is necessary to incorporate a large number of servers from the resource pool into the cluster at the same time, it is necessary to distribute a large-capacity image from the repository to a large number of servers at the same time. There is a risk. In such a case, it is not possible to quickly add resources according to the situation, and it is impossible to satisfy non-stop and responsiveness required for highly public communication services as infrastructure.

  In view of the above, an object of the present invention is to solve the above-described problems and provide means for quickly incorporating even a large number of servers from a resource pool into a cluster.

  In order to solve the above-described problem, according to the first aspect of the present invention, a repository apparatus that distributes a software image for operating a server to a server added to the cluster system via the management network performs the following processing. It was a method. That is, the repository apparatus accepts an input of an addition instruction including the number of additional servers and the type of software image to be distributed to the servers. Based on the input addition instruction, the server information indicating the ID of each server, whether or not the server can be added, and the address of the end point for remotely turning on the server is read. Next, the number of servers that can be added is selected from the servers indicated in the read server information. Then, a remote power-on instruction to each selected server is transmitted via the management network via the endpoint of the selected server. As a result, the selected server is powered on. After that, the repository device refers to the software image indicating the storage location of the software image of the type or the software image of the type for each type of software image, and sends the software image of the type indicated in the addition instruction via the management network To deliver to the selected server. Then, each node constituting the management network or each selected server copies and transfers the received software image to each other, thereby distributing the software image to the selected server.

  In this way, in the repository device, each node or server constituting the management network copies and transfers the software image to each destination server, so that the repository device unifies the software image to each server. There is no need to send in a cast. Thereby, congestion of the repository apparatus at the time of software image distribution can be prevented.

  According to a second aspect of the present invention, in the software image distribution method according to the first aspect, when the repository device distributes the software image to the selected server, the software image is distributed by multicast via the management network. .

  According to a seventh aspect of the present invention, a repository apparatus that distributes a software image for operating a server to a server added to the cluster system via the management network has the following configuration. In other words, the repository device receives a management interface unit that accepts an input of an addition instruction including the number of servers added and the type of software image to be distributed to the server, each server ID, whether the server can be added, and whether A storage unit that stores server information indicating an address of an endpoint that performs remote power-on, and software image information that indicates a software image of the type or a storage location of the software image of the type for each type of software image; Have. Further, the repository device selects the number of servers that can be added from the servers indicated in the server information based on the input addition instruction, and selects the selected server via the selected server's endpoint. A server activation unit that transmits a remote power-on instruction to each via a management network; and an image distribution unit that distributes the software image indicated in the addition instruction to each of the servers selected by multicast via the management network. Prepare.

  By doing so, the repository device distributes the software image to each server to be added to the cluster system by multicast, and therefore, the congestion of the repository device at the time of software image distribution can be prevented.

  According to a third aspect of the present invention, in the software image distribution method according to the first aspect, the following processing is performed after transmitting a remote power-on instruction to each of the selected servers via the management network. That is, the repository device creates transfer ID information indicating the transfer ID of the server of the selected ID for each selected server ID. Then, the created transfer ID information is distributed to each of the selected servers via the management network. Then, the repository device distributes the software image to the transfer server with the transfer ID in the first order or the last order among the transfer IDs indicated in the transfer ID information by unicast. Thereafter, each selected server receives the transfer ID information and the software image. Then, each of the selected servers refers to the transfer ID information and distributes the software image to the next server. That is, each selected server specifies its own server transfer ID based on the received transfer ID information and its own server ID. Then, referring to the specified transfer ID and transfer ID information, when it is determined that there is a server having a transfer ID next to the transfer ID of its own server, the received software image is transferred to the server having the next transfer ID. Copy and transfer. When there is no next server when viewed from its own server, the server does not transfer the software image.

  The invention described in claim 8 is a server used in a cluster system, and the server stores a remote management interface unit that performs a power-on process based on an instruction from an external device and a received software image. And a software image stored in the storage unit, a rebooting unit that reads the software image from the storage unit and starts up the OS of the software image, and an image transfer unit. The image transfer unit receives a software image for operating its own server and transfer ID information indicating the transfer ID of the server of each ID to be added to the cluster system. Based on the server ID of the server, the transfer ID of its own server is specified. Then, with reference to the specified transfer ID and transfer ID information, it is determined whether or not there is a server having a transfer ID next to the transfer ID of its own server. When it is determined that there is a server having a transfer ID next to the transfer ID of its own server, the received software image is copied and transferred to the server having the next transfer ID. When there is no next server when viewed from its own server, the server does not transfer the software image.

  In this way, the repository device distributes the software image to the server that is the root of the binomial tree among the servers to be added to the cluster system, and the server that receives the software image copies the software image to other servers. Forward to. Therefore, it is not necessary for the repository apparatus to transmit the software image to all the servers to be added by unicast, and the repository apparatus can be prevented from being congested when distributing the software image.

  According to a fourth aspect of the present invention, in the software image distribution method according to the first aspect, for each ID of the selected server, the repository device creates transfer ID information indicating the transfer ID of the server of that ID. Then, the created transfer ID information is distributed to each of the selected servers via the management network. Further, the repository device distributes the software image to a server that is a child node of the repository device when the selected server group is a node having a binary tree structure having the repository device as a root. Next, each selected server receives the transfer ID information and the software image. Then, each of the selected servers refers to the transfer ID information and distributes the software image to the next server. That is, each of the selected servers specifies the transfer ID of its own server based on the transfer ID information and its own server ID. Then, with reference to the specified transfer ID and transfer ID information, it is determined whether or not there is a child server as seen from its own server in the binary tree. Here, when it is determined that there are child servers when viewed from its own server, the server copies and transfers the received software image to each of the child servers. When there is no child server when viewed from its own server, the server does not transfer the software image.

  The invention according to claim 9 is a server used in the cluster system, and the server stores a remote management interface unit that performs a power-on process based on an instruction from an external device, and a received software image. A boot unit that performs startup processing of its own server according to the received software image, and an image transfer unit. The image transfer unit receives a software image for operating its own server, and transfer ID information indicating the transfer ID of the server corresponding to the ID of the server selected as the server to be added to the cluster system. Further, the image transfer unit specifies the transfer ID of its own server based on the transfer ID information and its own server ID. Then, referring to the specified transfer ID and transfer ID information, when the selected server group is a node of a binary tree having a repository device as a root, it is viewed from its own server in the tree structure. Determine if there is a child server. When it is determined that there is a child server, the image transfer unit performs a software image transfer process of copying and transferring the received software image to each child server. When there is no child server when viewed from its own server, the server does not transfer the software image.

  The invention according to claim 13 is used in a system including the server according to claim 9, and distributes a software image for operating the server to a server newly added to the cluster system via the management network. The repository device includes a management interface unit that receives an input of an additional instruction including the number of additional servers and the type of software image to be distributed to the server, and a storage unit. This storage unit includes server information indicating the ID of each server, whether or not the server can be added, and the address of the endpoint that remotely powers on the server, and the software image of the type for each type of software image Alternatively, software image information indicating a storage location of the type of software image is stored. Further, the repository device selects the number of servers that can be added from the servers indicated in the server information based on the input addition instruction, and selects each of the selected servers via the selected server's endpoint. A server activation unit that transmits a remote power-on instruction to the computer via the management network. In addition, the repository device creates transfer ID information indicating the transfer ID of the server of the selected server ID for each selected server ID, and distributes the created transfer ID information to each of the selected servers via the management network. A server that is a child node of a repository device when a server group activated by unicasting a software image of the type indicated in the add instruction is a binary tree tree node rooted in the repository device And an image distribution unit that distributes to

  In this way, the repository device distributes the software image to the servers that are the children of the root of the binomial tree among the servers to be added to the cluster system, and the server that receives the software image copies the software image and others. To the server. Therefore, it is not necessary for the repository apparatus to transmit the software image to all the servers to be added by unicast, and the repository apparatus can be prevented from being congested when distributing the software image. Each server recognizes a server that is a child node of its own server when each server of the cluster system is considered in the binary tree structure based on the transfer ID information transmitted from the repository device, and the child node The software image to other servers. Thereby, a software image can be distributed efficiently.

  According to a fifth aspect of the present invention, in the software image distribution method according to the third or fourth aspect, each of the selected servers stores a transfer OS (operating system) and a transfer program in the storage unit in advance. Yes. When the transfer ID information is received from the repository device and the software image is received from the repository device or another server, the server first loads the OS for transfer and the transfer program from the storage unit and starts up. Thereafter, the transfer program is executed by the activated transfer OS, receives transfer ID information from the repository apparatus, and receives a software image from the repository apparatus or another server. Then, the server executes a software image transfer step, and stores the software image in the software image storage unit of the storage unit of its own server. Thereafter, the server is restarted and the OS of the software image is started.

  By doing so, the server can execute the transfer program and transfer the software image by the transfer OS provided in the server.

  According to a tenth aspect of the present invention, the image transfer unit in the server according to the eighth or ninth aspect reads the transfer OS and transfer program from the storage unit by the boot unit, starts the transfer OS, and starts the transfer OS This is realized by executing the transfer program.

  As described above, when each server temporarily activates the transfer OS, the transfer program is executed, and the function of the image transfer unit of the server according to claim 8 or 9 can be realized.

  According to a sixth aspect of the present invention, in the software image distribution method according to the third or fourth aspect, each of the selected servers receives the transfer ID from the repository device, and the software image from the repository device or another server. First, a transfer program is acquired from an external device and activated. Then, a software image transfer step is executed, and the received software image is stored in the software image storage unit of the storage unit. Thereafter, the server activates the OS of the software image in the software image storage unit by the transfer program.

  In this way, even when each server does not have a transfer program, the software image transfer process can be executed and the OS of the software image can be activated.

  The invention according to claim 11 is realized by the image transfer unit in the server according to claim 8 or 9 executing the transfer program received from the external device by the boot unit.

  By doing in this way, even if each server is not provided with the transfer program, the function of the image transfer unit of the server according to claim 8 or claim 9 can be realized.

  The invention described in claim 14 is a system including the repository device and the server described in claim 7.

  By doing so, a system including the repository device and the server according to claim 7 can be realized.

  The invention described in claim 15 is a system including the repository device described in claim 12, and the server described in any one of claims 8, 10, and 11.

  By doing so, it is possible to realize a system including the repository device according to claim 12 and the server according to any one of claims 8, 10, and 11.

  The invention described in claim 16 is a system including the repository apparatus described in claim 13 and the server described in any one of claims 8, 10, and 11.

  By doing so, it is possible to realize a system including the repository apparatus according to claim 13 and the server according to any one of claims 8, 10, and 11.

  According to the present invention, it is not necessary for the repository apparatus to transmit the software image to each server added to the cluster system by unicast, so that the repository apparatus can be prevented from being congested during software image distribution. As a result, even when a large number of servers are incorporated into the cluster from the resource pool, they can be quickly incorporated.

It is the figure which showed the outline | summary of the system of 1st Embodiment. It is a figure showing an example of composition of a system of a 1st embodiment. It is the figure which illustrated the software image management table in software image DB of FIG. It is the figure which illustrated the server information in server DB of FIG. It is the flowchart which showed the process sequence of the system of FIG. It is the figure which showed the outline | summary of the system of 2nd Embodiment. It is the figure which showed the structural example of the system of 2nd Embodiment. It is the figure which illustrated the transfer ID information which the repository apparatus of FIG. 7 delivers. It is the figure which showed notionally the delivery of the software image in the system of FIG. It is the flowchart which showed the process sequence of the system of FIG. It is the figure which showed the structural example of the system of 2nd Embodiment. FIG. 8 is a flowchart showing a processing procedure when a server of the system of FIG. 7 determines a transfer destination of a software image. It is the figure which showed the structural example of the system in other embodiment. It is the flowchart which showed the process sequence of the system of FIG.

  Hereinafter, modes for carrying out the present invention (hereinafter referred to as embodiments) will be described by dividing them into a first embodiment and a second embodiment. In the following description, a software image is assumed to be a file including an OS (operating system), middleware, software for providing various services by a server, data, and the like. Further, the number of repository apparatuses 2100 and servers 2310, 2320, and 2330 is not limited to the number shown in each drawing.

<First Embodiment>
First, the outline of the first embodiment will be described with reference to FIG. As shown in FIG. 1, the cluster system 2000 supports the repository apparatus 2100, the server 2310 (servers 2310A, 2310B, 2310C) of the resource pool 2300, and multicast distribution (for example, IP (Internet Protocol) network) management. And a network 2200. The repository device 2100 and the server 2310 of the resource pool 2300 are connected via the management network 2200. The repository apparatus 2100 distributes the software image by multicast via the management network 2200. As a result, congestion of the repository apparatus 2100 during software image distribution can be prevented. That is, in the management network 2200, the router in the management network 2200 duplicates and distributes data (software image), so that no congestion occurs in the repository apparatus 2100 during software image distribution. Therefore, the repository apparatus 2100 can quickly distribute the software image to each server 2310.

  In the repository apparatus 2100, when the management network 2200 is a network of the same segment, distribution of the software image to the broadcast address is multicast.

<Configuration>
Next, a configuration example of the system according to the first embodiment will be described with reference to FIG. As shown in FIG. 2, the cluster system 2000 is connected to the maintenance person terminal 1000 via a maintenance network 4000 (LAN (Local Area Network) or IP network), and is connected to the user terminal 3000 via a service network 5000. The service provider inputs a maintenance command (described later) for the cluster system 2000 from the maintenance person terminal 1000. A user uses a user terminal 3000 to use a service provided on the cluster system 2000 over a service network 5000 (LAN or IP network).

  The cluster system 2000 includes a resource pool 2300 in which a plurality of servers 2310 that are computing resources are aggregated, and a repository apparatus 2100, which are connected by a management network 2200.

  The repository apparatus 2100 starts up the server 2310 of the resource pool 2300 based on the maintenance command input from the maintenance person terminal 1000 and distributes the software image to the server 2310. The repository device 2100 includes a management interface unit 2110, a software image DB (database) 2120, a server DB 2130, an image distribution unit 2140, and a server activation unit 2150. The repository apparatus 2100 is realized by a computer including an input / output interface, a communication interface, a CPU (Central Processing Unit), a memory, and the like.

  The management interface unit 2110 serves as an interface with the maintenance person terminal 1000 and receives a maintenance command. For example, the management interface unit 2110 receives a maintenance command to be added to the server 2310 operated in the cluster system 2000. This maintenance command includes at least two of the number of servers to be added and the image ID of the software image to be distributed. The management interface unit 2110 is realized by a communication interface, an input / output interface, and a program execution process by the CPU.

  The software image DB 2120 includes a software image management table (software image information, see FIG. 3). The software image management table is information indicating the storage location of the software image for each type of software image (image ID) necessary for executing a predetermined service. The software image is a file including a software configuration necessary for providing a service, and is a file distributed to the server 2310. When the software image DB 2120 receives an input of an image ID as a main key from the management interface unit 2110 or the like, the software image DB 2120 returns a path indicating the storage location of the software image. The path may be on a local disk of the repository apparatus 2100 or a remote storage as long as the repository apparatus 2100 can access the path.

  The server DB 2130 includes server information (see FIG. 4). This server information is information indicating the address of the remote power management endpoint of the server 2310 of the server ID and the state of the server 2310 (whether it can be used as an additional resource) for each server ID. The address of the remote power management endpoint varies depending on the server interface that remotely powers on the server 2310. For example, the IP address is IPMI (Intelligent Platform Management Interface), and the MAC address is WoL (Wake on LAN). (Media Access Control) address. Further, the state of the server 2310 only needs to be able to determine whether or not the server 2310 can be used as an additional resource. In the example of FIG. 4, the state that can be used as an additional resource is “standby”, and the other unusable states are divided into “use” and “failure” for each reason. For example, when receiving an inquiry from the image distribution unit 2140 or the server activation unit 2150 to the server 2310 that is available as an additional resource, the server DB 2130 is a server whose status is “standby” in the server information. A predetermined number 2310 is selected, and the server ID of the server 2310 and the end point are returned.

  The server DB 2130 and the software image DB 2120 include program execution processing by the CPU and storage means for storing server information and a software image management table (RAM (Random Access Memory), HDD (Hard Disk Drive), flash memory, etc.). It is realized by.

  In response to the server addition request received by the management interface unit 2110, 2150 inquires the server DB 2130 about the server 2310 in the “available” state. Next, the server activation unit 2150 selects the server 2310 to be added from the available servers 2310 obtained as a response to the inquiry. Then, the remote power management end point of the selected server 2310 is obtained, and the selected server 2310 is turned on remotely.

  The image distribution unit 2140 specifies a necessary software image by making an inquiry to the software image DB 2120 in response to a server addition request received by the management interface. Then, the image distribution unit 2140 distributes the identified software image to each server 2310 to be added.

  The server activation unit 2150 and the image distribution unit 2140 are realized by a program execution process by the CPU.

  Next, the server 2310 will be described. After the server 2310 is turned on, the server 2310 receives the software image distributed from the repository device 2100 and restarts the server 2310 with this software image. The server 2310 includes a boot unit 2312, a remote management interface unit 2313, and a software image storage unit 2314. The server 2310 is also realized by a computer including an input / output interface, a communication interface, a CPU, a storage unit, and the like.

  The boot unit 2312 restarts the server 2310 after receiving the software image distributed from the repository apparatus 2100. Then, the received software image is read and the OS of this software image is activated. As a result, the server 2310 can provide a service via the service network 5000.

  The remote management interface unit 2313 is an interface that accepts remote power-on from a server (not shown) that performs remote power-on from the server activation unit 2150 of the repository device 2100.

  The software image storage unit 2314 stores the software image distributed from the repository device 2100. The software image storage unit 2314 is realized by storage means (not shown) of the server 2310.

<Processing procedure>
Next, processing procedures of the repository apparatus 2100 and the server 2310 of FIG. 2 will be described with reference to FIG.

  The management interface unit 2110 of the repository apparatus 2100 receives an addition request (addition instruction, including an image ID and the number of servers to be added) from the maintenance person terminal 1000 of FIG. 2 via the maintenance network 4000 (S1). Thereafter, the management interface unit 2110 notifies the server activation unit 2150 of the number of servers to be added (S2). After S2, the server activation unit 2150 inquires of the server 2310 available to the server DB 2130, and acquires the remote power management endpoints of the necessary number of servers (S3). Then, each server 2310 is remotely powered on via this remote power management endpoint (S4). Each server 2310 is activated by a boot unit 2312 (for example, a boot loader such as GRUB (GRand Unified Bootloader)) when remote power-on is performed. Here, it is assumed that the server 2310 to be activated is selected by a predetermined algorithm, such as a method of selecting in ascending order of server IDs or a method of selecting a server having the longest standby time.

  Also, the management interface unit 2110 of the repository device 2100 notifies the image distribution unit 2140 of the image ID of the software image to be distributed and the number of servers to be added (S5). Then, the image distribution unit 2140 makes an inquiry to the software image DB 2120 to acquire the corresponding software image (S6), and prepares for distribution.

  After S6, the image distribution unit 2140 of the repository device 2100 distributes the software image acquired in S6 to each server 2310 by multicast via the management network 2200 (S11). Each server 2310 stores the received software image in the software image storage unit 2314 (S12), and each server 2310 is restarted (S13). When restarting, the boot unit 2312 of the server 2310 loads the software image from the software image storage unit 2314 (S14), and starts the OS in the software image (S15). Thereby, the activation of the server 2310 requested to be added is completed.

  Note that the method of adding the activated server 2310 to the cluster system 2000 differs depending on the service and the cluster system 2000, and the logic is assumed to be included in the software image. Assume that you have completed adding server resources.

  In S2 described above, the number of servers to be added is determined by a predetermined algorithm based on measurement data such as the current usage rate of the processor of the server 2310, the number of past accesses, the number of data processes, and the like. It may be.

  By doing so, the repository device 2100 does not cause congestion on the repository device 2100 side even when distributing the software image to a large number of servers 2310. Thereby, even when many servers 2310 are incorporated into the cluster (cluster system 2000) from the resource pool 2300, they can be incorporated quickly.

<Second Embodiment>
The outline of the second embodiment will be described with reference to FIG. Constituent elements similar to those of the above-described embodiment are denoted by the same reference numerals and description thereof is omitted. The second embodiment is characterized in that all servers do not simply receive a software image from the repository apparatus 2100 but transfer a predetermined distribution order between the repository apparatus 2100 and the server. Specifically, in order to transfer the repository device 2100 and the server in a predetermined distribution order, the repository device 2100 allocates a transfer ID of each server, and transfers the transfer ID information indicating the transfer ID of each server. Distribute to the server. Each server refers to the transfer ID information and transfers the software image in a predetermined distribution order. For example, when transferring software images in ascending order of transfer IDs, the repository apparatus 2100 first distributes the transfer IDs indicated in the transfer ID information to the server with the highest transfer ID order. Software images are transferred in ascending order of transfer ID. When transferring the software image in the descending order of the transfer ID, the repository device 2100 first distributes the transfer ID indicated in the transfer ID information to the server with the latest transfer ID order, and then each server receives the transfer ID. Software images are transferred in descending order. As an example of an efficient distribution order, the repository apparatus 2100 and the server are configured by a binary tree having the repository apparatus 2100 as a root (root), a child node from the root of the binary tree, and further child nodes of the child node. It can be considered that the order is as follows. An embodiment in this case will be described below.

  As illustrated in FIG. 6, the resource pool 2300 of the cluster system 2000 according to the second embodiment includes a server 2320 (servers 2320A, 2320B, and 2320C). In such a system, the repository apparatus 2100 transmits a software image to a predetermined server 2320 (not all the servers 2320) in the resource pool 2300 via the management network 2200. The server 2320 that has received the software image stores the received software image, copies the software image, and transfers it to the other server 2320 to be added via the management network 2200. That is, the addition target servers 2320 cooperate to transfer software images. In this way, even when the management network 2200 does not support multicast, the software image can be quickly distributed to each server 2320.

  Next, an example of the overall configuration of the system according to the second embodiment will be described with reference to FIG. The repository apparatus 2100 of the system according to the second embodiment includes an image distribution unit 2141 instead of the image distribution unit 2140 of FIG. The resource pool 2300 includes a server 2320. The server 2320 further includes an image transfer unit 2321 in addition to the configuration of the server 2310 of FIG.

  The image distribution unit 2141 distributes the software image to the server 2320 (server 2320 to be added to the cluster system 2000) activated by remote power-on. Here, the server 2320 to which the software image is distributed is a binary tree whose root is the repository device 2100 for the server 2320 group to be added (the binary tree of degree 0 has one node and the binomial of degree k. A tree has one root, and its child nodes are trees with the characteristic that they are the parent nodes of a binary tree of degree k-1, k-2,. The server 2320 is a child node of the repository apparatus 2100. That is, the image distribution unit 2141 does not distribute the software image to all the servers 2320 to be added, but distributes it to some of the servers 2320.

  The image distribution unit 2141 includes a transfer ID information creation unit 2142. When the transfer ID information creation unit 2142 receives registration of a server ID from the server 2320 (addition target server 2320) activated by remote power-on, the transfer ID information generation unit 2142 indicates the transfer ID of the server 2320 for each server ID. ID information (see FIG. 8) is created. This transfer ID is obtained by assigning unique numbers 1 to N to all servers 2320 to be added (for example, N servers 2320). If a unique number from 1 to N is assigned to each server 2320, the order does not matter. For example, when numbers are assigned in ascending order of server IDs, the transfer ID information shown in FIG. 8 is obtained. The image distribution unit 2141 distributes the transfer ID information to each of the servers 2320 to be added. Then, the server 2320 that has received this transfer ID information identifies its own transfer ID from this transfer ID information, and determines whether there is a server 2320 that is the transfer destination of the software image. If there is a server 2320 that is a transfer destination of the software image, the software image is transferred to the server 2320. Details of the determination process of the transfer destination server 2320 at this time will be described later with reference to FIG. As described above, each server 2320 grasps the server 2320 group to be added as a binary tree, so that it can be easily determined whether there is a child node to which the software image is to be transferred.

  When the image transfer unit 2321 of the server 2320 receives the transfer ID information from the repository device 2100 and receives the software image from the repository device 2100 or another server 2320, the image transfer unit 2321 refers to the transfer ID information and the server ID of its own server 2320. To identify its own transfer ID. Based on this transfer ID, when the addition target server 2320 group is considered as a binary tree structure (see FIG. 9), it becomes a child node (transfer destination of the software image) of its own server 2320. It is determined whether there is a server 2320. If there is a server 2320 as a child node, the software image is copied and transferred to the server 2320. If there is no server 2320 as a child node, the software image is not transferred and the process is terminated. The image transfer unit 2321 is realized by the boot unit 2312 reading and starting the transfer OS and transfer program, and executing the transfer program by the started transfer OS.

Here, an outline of software image distribution processing will be described with reference to FIGS. Here, as shown in FIG. 9, an example of processing when the repository apparatus 2100 distributes a software image to the server 2320 with the transfer ID “1-7” will be described. In this algorithm, when all servers 2320 are mapped to a binary tree structure with the repository device 2100 as a root, the number of servers 2320 is N, and distribution is performed in log ₂ (N) transfer steps. Can be completed. In other words, the load of the software image distribution process can be greatly reduced as compared with the case where the repository apparatus 2100 distributes the software image to all the servers 2320 by unicast. An outline of processing when distributing software images along the binomial tree of FIG. 9 will be described below.

First, the repository apparatus 2100 allocates transfer IDs having numbers 1 to N to all servers 2320 (N is assumed here) to be activated. If different numbers from 1 to N are assigned to each server 2320, the order is not limited. For example, when the server IDs are allocated in ascending order, the result is as shown in FIG. The image distribution unit 2141 of the repository apparatus 2100 transmits the transfer ID information illustrated in FIG. 8 to all the servers 2320. This transmission may be unicast for the number of servers. Upon receiving this transfer ID information, each server 2320 recognizes the transfer ID allocated to itself from the transfer ID information and the server ID. Here, a method for setting the transfer ID of the repository apparatus 2100 to “0” and distributing the software image of the repository apparatus 2100 to the servers 2320 from the transfer IDs “1” to “N” based on the binomial tree will be described. Each server 2320 and repository apparatus 2100 to which a transfer ID is assigned has the following two expressions s = log ₂ (own transfer ID + 1)... (1)
id = own transfer ID + 2 ^s (2)
The software image transfer destination server 2320 is determined based on the flowchart shown in FIG.

  The repository device 2100 and each server 2320 (hereinafter collectively referred to as a “node”) refer to the transfer ID information and its own server ID to identify its own transfer ID. If the node has a software image (Yes in S101), the process proceeds to S2. On the other hand, since the server 2320 that has not received the software image proceeds to No in S101, it does not proceed to the previous state.

According to the calculation formula of S102 (s = log ₂ (own transfer ID + 1)) and the calculation formula of S103 (id = own transfer ID + 2 ^s ), each node has a server image to be transmitted based on the binomial tree. The value of the transfer ID (initially the transfer ID is “1”) is calculated. If the value calculated here exceeds the maximum value of the transfer ID (No in S104), it means that there is no child node in the binomial tree, and therefore the transfer of the software image is completed (S107). . On the other hand, when the calculated value is equal to or less than the maximum value of the transfer ID (Yes in S104), the software image is transferred to the node (server 2320) of the calculated transfer ID (id) (S105). Thereafter, in order to perform transmission processing for another child in the binary tree, the node increments the value of s (S106), and executes the processing from S103 onward again.

  By performing the processing as described above, for example, the repository apparatus 2100 (transfer ID “0”) in FIG. 9 transfers the software image to the server 2320 with the transfer IDs “1”, “2”, and “4”. The server device 2320 with the transfer ID “1” transfers the software image to the server 2320 with the transfer IDs “3” and “5”. Further, the server 2320 with the transfer ID “3” transfers the software image to the server 2320 with the transfer ID “7”.

  In this way, the repository apparatus 2100 in the cluster system 2000 can significantly reduce the load of the transfer process compared to distributing the software image to each of the addition target servers 2320.

  Details of the server 2320 having such a transfer function will be described with reference to FIG. The storage means (not shown) of the server 2320 includes a management OS storage unit 2315 in addition to the software image storage unit 2314. These may be physically different disks or different areas on the same disk. The image transfer unit 2321 described above is realized by a transfer APL (transfer application, transfer program) 2322 of the management OS storage unit 2315 and a transfer OS 2323 that operates the transfer APL 2322.

  The software image distribution order (transfer order) is, for example, the repository device 2100 → the server 2320 with the transfer ID “0” → the server 2320 with the transfer ID “1” → the server 2320 with the transfer ID “2” →. The servers 2320 may be logically arranged in a line in the order of transfer IDs (in ascending order or descending order), and the software images may be transferred in order from the top. In this case, the repository apparatus 2100 first distributes the software image to the server 2320 having the transfer ID in the first order (or the latest) among the transfer IDs indicated in the transfer ID information. Thereafter, the server 2320 that has received this software image refers to the transfer ID information and its own server ID, and identifies the transfer ID of its own server 2320. Then, referring to the transfer ID information, if there is another server 2320 whose transfer ID is next to the transfer ID of its own server 2320 in ascending order (or descending order), the software image is transferred to this server 2320. On the other hand, if there is no other server 2320 whose transfer ID is next to the transfer ID of its own server 2320 in ascending order (or descending order), the software image is not transferred and the process ends. Also by such a method, the processing load when the repository apparatus 2100 distributes the software image can be reduced, and congestion can be effectively prevented.

<Processing procedure>
Next, processing procedures of the repository apparatus 2100 and the server 2320 of FIG. 7 will be described with reference to FIG. Steps up to S6 in FIG. 12 are the same as those in the first embodiment, and thus description thereof will be omitted.

  When the server 2320 is turned on remotely, the boot unit 2312 reads the transfer OS 2323 from the management OS storage unit 2315 (S7) and activates the transfer OS 2323 (S8). When the transfer OS 2323 is activated, each server 2320 executes the transfer APL 2322 by the transfer OS 2323 (S9), and the server ID of its own server 2320 is assigned to the image distribution unit 2141 of the repository apparatus 2100 by the transfer APL 2322. Register (S10). That is, it notifies the repository apparatus 2100 that its own server 2320 has started. The image distribution unit 2141 of the repository apparatus 2100 receives the registered server ID and stores it in a predetermined area of the storage unit. Thereafter, the image distribution unit 2141 of the repository device 2100 transfers, to each activated server 2320 (the server 2320 registered in S10), a transfer ID indicating the transfer ID of the server 2320 for each server ID of each activated server 2320. ID information (see FIG. 8) is created and distributed to each server 2320. Further, the image distribution unit 2141 distributes the software image to the server 2320 corresponding to the child node of the binomial tree. Thereafter, each server 2320 that received the software image grasps its own transfer ID from the received transfer ID information, and if there is a server 2320 corresponding to the child node, copies the software image to the server 2320 and transfers it to the server 2320. By repeating, the software image is distributed to each server 2320 (S11). The details of the processing of S11 are as described with reference to FIG. Moreover, since the process of S12-S15 is the same as that of 1st Embodiment, description is abbreviate | omitted.

  In this way, even when the management network 2200 does not support multicast, the repository device 2100 can quickly distribute a software image to each server 2320 without causing congestion.

<Other embodiments>
Note that the function of the image transfer unit 2321 (see FIG. 11) of the server 2320 may be provided in advance by the server 2320 itself as in the second embodiment, or the network boot server 2400 shown in FIG. The bootstrap program 2430 may be distributed to each server 2330 (2330A, 2330B, 2330C).

  This network boot server 2400 is realized by a computer having a CPU, storage means, a communication interface with the management network 2200, and the like, and is a server connected to the management network 2200. The network boot server 2400 includes an address distribution unit 2410 and a bootstrap program distribution unit 2420, and stores the network bootstrap program 2430 in a predetermined area of the storage unit. The network bootstrap program 2430 includes an image transfer program 2331 for realizing the function of the image transfer unit 2321.

  When the address distribution unit 2410 receives a DHCP (Dynamic Host Configuration Protocol) request from the activated server 2330, the address distribution unit 2410 issues an IP address to be assigned to the server 2330 in response to the request. Further, the address of the network boot server 2400 and a path indicating the storage location of the network bootstrap program 2430 are transmitted to the server 2330.

  The bootstrap program distribution unit 2420 distributes the network bootstrap program 2430 to the server 2330.

  In the following description, each server 2330 reads and starts a remote software image according to a network boot standard such as PXE (Preboot eXecution Environment).

  Next, processing procedures of the repository device 2100, the network boot server 2400, and the server 2330 of FIG. 13 will be described with reference to FIG. Steps up to S6 in FIG. 14 are the same as those in the first embodiment and the second embodiment, and thus description thereof will be omitted, and description will be made from S25.

  When the server 2330 is powered on in S4, the boot unit 2312 in FIG. 13 starts network boot according to the PXE net boot standard (S25). Then, the boot unit 2312 transmits a DHCP request by a broadcast packet of the management network 2200 (S26). The address distribution unit 2410 of the network boot server 2400 that has received this DHCP request issues an IP address to the server 2330 that is the source of the DHCP request. Also, the address of the network boot server 2400 and the path indicating the storage location of the network bootstrap program 2430 are notified (S27). The boot unit 2312 of each server 2320 acquires the network bootstrap program 2430 on the network boot server 2400 using a protocol such as TFTP (Trivial File Transfer Protocol) (S28), and executes the network bootstrap program 2430 on the server 2330. (S29). When the function of the image transfer unit 2321 (see FIG. 11) is realized by the executed network bootstrap program 2430, the server 2330 causes the image of the repository device 2100 to be displayed by the image transfer unit 2321 as in S10 of FIG. Registers itself in the distribution unit 2141 (S30). Thereafter, the repository apparatus 2100 distributes the software image to each server 2330 in the same manner as S11 in FIG. 5 and S11 in FIG. However, in the present embodiment, the server 2330 does not temporarily store the software image in a storage area such as an HDD based on the network bootstrap program 2430, and directly starts up the OS in the software image in the main memory. (S31).

  Thus, even when the server 2330 does not include the image transfer unit 2321, the boot unit 2312 can execute the transfer program received from the network boot server 2400, thereby realizing the function of the image transfer unit 2312. . As a result, the system can perform the software image transfer process as described in the second embodiment.

1000 Maintenance terminal 2000 Cluster system 2100 Repository device 2110 Management interface unit 2120 Software image DB
2130 Server DB
2140, 2141 Image distribution unit 2142 Transfer ID information creation unit 2150 Server activation unit 2200 Management network 2300 Resource pool 2310, 2320, 2330 Server 2312 Boot unit 2313 Remote management interface unit 2314 Software image storage unit 2315 Management OS storage unit 2321 Image transfer unit 2322 Transfer APL
2323 OS for transfer
2331 Image transfer program 2400 Network boot server 2410 Address distribution unit 2420 Bootstrap program distribution unit 2430 Network bootstrap program 3000 User terminal 4000 Maintenance network 5000 Service network

Claims (16)

A repository device that distributes the software image for operating the server to the server to be added to the cluster system via the management network.
Receiving an input of an additional instruction including the number of additional servers and the type of software image to be distributed to the servers;
Based on the input addition instruction, the server information indicating the ID of each of the servers, whether the server can be added, and the address of the endpoint for performing remote power-on to the server is read, and the read Of the servers indicated in the server information, the number of servers that can be added is selected, and the remote power-on instruction to each of the selected servers is selected via the endpoint of the selected server. Transmitting over the network;
For each type of the software image, refer to the software information indicating the type of software image or the storage location of the type of software image, and the type of software image indicated in the addition instruction via the management network Delivering to the selected server;
Each of the nodes constituting the management network or each of the selected servers performs the step of distributing the software image to the selected server by copying and transferring the received software image to each other. A software image distribution method characterized. The repository device is
In the step of distributing the software image to the selected server,
2. The software image distribution method according to claim 1, wherein the software image is distributed to the selected server by multicast via the management network. The repository device is
After executing a step of transmitting a remote power-on instruction to each of the selected servers via the management network,
Creating transfer ID information indicating the transfer ID of the server of the ID for each of the selected server IDs;
Further delivering the created transfer ID information to each of the selected servers via the management network;
Instead of delivering the software image to the selected server,
The step of delivering the software image to the transfer server with the transfer ID in the first or last transfer ID among the transfer IDs indicated in the transfer ID information by unicast,
Each of the selected servers delivering the software image to each other;
Each of the selected servers is
Receiving the transfer ID information from the repository device and receiving the software image from the repository device or another server;
Based on the received transfer ID information and the ID of the own server, the transfer ID of the own server is specified, and the transfer ID of the own server is determined by referring to the specified transfer ID and the transfer ID information. The software image transfer step of copying and transferring the received software image to the server with the next transfer ID when it is determined that there is a server with the next transfer ID. Software image distribution method. The repository device is
After executing a step of transmitting a remote power-on instruction to each of the selected servers via the management network,
Creating transfer ID information indicating the transfer ID of the server of the ID for each of the selected server IDs;
Further delivering the created transfer ID information to each of the selected servers via the management network;
Instead of delivering the software image to the selected server,
Executing the step of distributing the software image to a server that is a child node of the repository device when the selected server group is a binary tree tree node rooted in the repository device;
Each of the selected servers delivering the software image to each other;
Each of the selected servers is
Receiving the transfer ID information from the repository device and receiving the software image from the repository device or another server;
Based on the received transfer ID information and the ID of the own server, the transfer ID of the own server is specified, and on the basis of the specified transfer ID, a child server is seen from the own server in the binomial tree. The software image transfer step of executing a software image transfer step of copying and transferring the received software image to each of the child servers when it is determined whether or not there is the child server. 2. A software image distribution method according to 1. Each of the selected servers is
A storage OS (operating system) and a transfer program are stored in the storage unit in advance.
In the step of receiving the transfer ID information from the repository device and receiving the software image from the repository device or another server,
The transfer OS and the transfer program are read from the storage unit and activated, the transfer program is executed by the activated transfer OS, the transfer ID information is received from the repository device, the repository device or the like Receiving the software image from a server of:
The software image transfer step;
Storing the software image in the software image storage unit of the storage unit;
further,
After the step of storing the received software image in the software image storage unit of the storage unit is completed, the step of restarting the server and starting the OS of the software image of the software image storage unit is executed. The software image distribution method according to claim 3 or 4. Each of the selected servers is
In the step of receiving the transfer ID information from the repository device and receiving the software image from the repository device or another server,
Obtaining and starting a transfer program from an external device; and
Receiving the transfer ID information from the repository device by the transfer program, and receiving the software image from the repository device or another server;
The software image transfer step;
Storing the received software image in a software image storage unit of a storage unit,
further,
5. The software image distribution method according to claim 3, wherein a step of starting an OS of the software image in the software image storage unit is executed by the transfer program. A repository device that distributes a software image for operating a server to be added to a cluster system via a management network,
A management interface unit that accepts an input of an additional instruction including the number of additional servers and the type of software image to be distributed to the servers;
The server image indicating the ID of each server, whether or not the server can be added, and the address of the endpoint that remotely powers on the server, and the software image of the type or the type for each type of the software image A storage unit for storing software image information indicating a storage location of the software image;
Based on the input addition instruction, the number of servers that can be added is selected from the servers indicated in the server information, and the selected server is sent to each of the selected servers via an endpoint of the selected server. A server activation unit that transmits a remote power-on instruction via the management network;
A repository apparatus comprising: an image distribution unit that distributes the software image indicated by the addition instruction to each of the selected servers via multicast via the management network. A server used in a cluster system,
A remote management interface unit that performs power-on processing based on an instruction from an external device;
A software image for operating its own server, and transfer ID information indicating the transfer ID of the server of that ID for each server ID added to the cluster system are received, and the transfer ID information and its own server are received. Based on the ID of the server, the transfer ID of the own server is identified, and it is determined that there is a server having a transfer ID next to the transfer ID of the own server with reference to the identified transfer ID and the transfer ID information. An image transfer unit that copies and transfers the received software image to the server of the next transfer ID;
The storage unit for storing the received software image;
A server comprising: a boot unit that reboots after the software image is stored in the storage unit, reads the software image from the storage unit, and starts an OS of the software image. A server used in a cluster system,
A remote management interface unit that performs power-on processing based on an instruction from an external device;
A software image for operating its own server, and transfer ID information indicating the transfer ID of the server of that ID for each server ID added to the cluster system are received, and the transfer ID information and its own server are received. Based on the ID of the server, the transfer ID of the own server is specified, and based on the specified transfer ID, a server group selected as a server to be added to the cluster system is a binary tree whose root is the repository device. When it is a tree-structured node, it is determined whether or not there is a child server viewed from its own server in the tree structure, and when it is determined that there is a child server, the child server receives the received server An image transfer unit for performing a software image transfer process for copying and transferring a software image;
The storage unit for storing the received software image;
A server comprising: a boot unit that reboots after the software image is stored in the storage unit, reads the software image from the storage unit, and starts an OS of the software image. The image transfer unit, the boot unit,
The load OS and the transfer program are read from the storage unit and activated, and the transfer program is executed by the activated transfer OS. server. The image transfer unit, the boot unit,
The server according to claim 8 or 9, which is realized by executing a transfer program received from an external device. A software image used in a system including the server according to any one of claims 8, 10, and 11 for operating the server via a management network to a server added to the cluster system. A repository device that distributes
A management interface unit that accepts an input of an additional instruction including the number of additional servers and the type of software image to be distributed to the servers;
The server image indicating the ID of each server, whether or not the server can be added, and the address of the endpoint that remotely powers on the server, and the software image of the type or the type for each type of the software image A storage unit for storing software image information indicating a storage location of the software image;
Based on the input addition instruction, the number of servers that can be added is selected from the servers indicated in the server information, and the selected server is sent to each of the selected servers via an endpoint of the selected server. A server activation unit that transmits a remote power-on instruction via the management network;
For each ID of the selected server, transfer ID information indicating the transfer ID of the server of that ID is created, and the created transfer ID information is distributed to each of the selected servers via the management network, and the addition A software image of the type indicated in the instruction, and an image distribution unit that distributes to the server with the first or last transfer ID among the transfer IDs indicated in the transfer ID information by unicast. Feature repository device. A repository that is used in a system including a server according to any one of claims 9 to 11 and that distributes a software image for operating the server to a server added to the cluster system via a management network. A device,
A management interface unit that accepts an input of an additional instruction including the number of additional servers and the type of software image to be distributed to the servers;
The server image indicating the ID of each server, whether or not the server can be added, and the address of the endpoint that remotely powers on the server, and the software image of the type or the type for each type of the software image A storage unit for storing software image information indicating a storage location of the software image;
Based on the input addition instruction, the number of servers that can be added is selected from the servers indicated in the server information, and the selected server is sent to each of the selected servers via an endpoint of the selected server. A server activation unit that transmits a remote power-on instruction via the management network;
For each ID of the selected server, transfer ID information indicating the transfer ID of the server of that ID is created, and the created transfer ID information is distributed to each of the selected servers via the management network, and the addition The software image of the type indicated in the instruction is a child node of the repository device when the selected server group is a binary tree tree node rooted at the repository device by unicast. A repository apparatus comprising: an image distribution unit that distributes to a server.   A system comprising the repository device according to claim 7 and the server.   A system comprising: the repository device according to claim 12; and the server according to any one of claims 8, 10, and 11.   A system comprising the repository device according to claim 13 and the server according to any one of claims 9 to 11.

Images