WO2011099380A1

WO2011099380A1 - Essential data management system, calculator, essential data management program, recording medium, and communication method

Info

Publication number: WO2011099380A1
Application number: PCT/JP2011/051712
Authority: WO
Inventors: 英治松尾
Original assignee: 三菱電機株式会社
Priority date: 2010-02-10
Filing date: 2011-01-28
Publication date: 2011-08-18
Also published as: JPWO2011099380A1; JP5355725B2

Abstract

Provided is a data backup system which does not require a shared disc or a designated server for backup. One server (S) functioning as a reception server inputs registration instructions which instruct to save data that is essential for the data processing executed by the processing execution unit (300) of said server (S). Once the registration instruction is entered, the reception server determines a server which acts as master in relation to the registration instruction between the other servers by means of communication, and transmits the registration instruction to the master server. The master server selects, on the basis of a predetermined selection rule and from among a plurality of servers, a server to which the essential data specified by the registration instruction is stored, and transmits data arrangement information indicating the correspondence between the selected server and the essential data to be stored to a retention server (for retaining the essential data) specified in the registration instruction. Upon receiving the data arrangement information from the master server, the retention server transmits the essential data to the selected server indicated in the data arrangement information.

Description

Essential data management system, computer, essential data management program, recording medium, and communication method

The present invention relates to an essential data management system, a computer, an essential data management program, and a communication method for backing up essential data as a countermeasure against a failure of a server device.

When a failure occurs in a computer (hereinafter referred to as a server), the processing continued on the spare server is restarted by restarting the processing that was running on the server that was stopped due to the failure. When constructing this configuration with a non-shared disk, it is necessary to duplicate in advance data relating to the processing of the spare server (hereinafter, essential data).

In addition, in order to take over such processing in a short time, all data groups to be arranged (data such as essential data and their differences, hereinafter referred to as preliminary data) must be properly managed prior to the failure. The problem of having to occur arises. This means that if the required data does not exist on the server that has taken over the processing, it is necessary to restore the required data on the server that takes over the processing. Because.

JP 2009-080695 A (Synchronization in operation and standby systems) JP 2008-257576 A (Quick storage by backup server) JP 2008-217302 A (Determination of processing takeover destination)

In Patent Document 1, regarding the takeover of processing, the spare server and the active server are operated at 1: 1, and data is replicated by synchronizing the data of the spare server and the active server (FIG. 40). However, since the conventional technique operates at 1: 1, a spare server that matches the active server is required, which increases the cost.

In the prior art of Patent Document 2, it is possible to cope with data duplication by backing up data and restoring the data. In addition, by limiting the data to be backed up to only the difference from before, the network load at the time of backup and the disk capacity required for backup are reduced. However, the conventional technique requires a dedicated server for backup, and the cost becomes a problem (FIG. 41). Further, the speed at the time of restoration is not taken into consideration, and there remains a problem that the processing cannot be taken over in a short time.

In the prior art of Patent Document 3, the process takeover itself is completed in a short time by selecting a process takeover destination. However, a shared disk is assumed, and data takeover accompanying process takeover is not considered. Therefore, when applied in a non-shared disk environment, problems remain in handling data.

The essential data management system of this invention is
A plurality of computers including a process execution unit that executes a predetermined process, a storage unit that stores information, and an essential data management unit are connected via a network,
Any of the calculators
A storage instruction for instructing storage of essential data that is partial data constituting part of execution use data used for execution of the process by the process execution unit of any of the computers, the essential data Input a storage instruction including the required data specifying information for specifying the stored computer specifying information for specifying the held computer that is the computer that holds the required data as a save instruction input computer,
The essential data management unit of the storage instruction input computer is:
When the save instruction is input, a master for the save instruction is communicated with the essential data management unit of another computer by communicating with the essential data management unit of the other computer via the network. Determine the computer to be, and transmit the storage instruction via the network to the storage master management unit which is an essential data management unit of the computer determined as the master,
The storage master management unit
A storage destination of the essential data specified by the essential data specifying information of the storage instruction is selected from the plurality of computers based on a predetermined selection rule, and the selected selection computer and the selection computer are stored Transmitting data allocation information indicating the correspondence with the essential data to the possessed computer specified by the retained computer specifying information of the storage instruction,
The essential data management unit of the possessed computer is:
When the data arrangement information is received from the storage master management unit, the essential data is transmitted to the selection computer indicated by the data arrangement information via the network,
The selection calculator is
The essential data transmitted from the possessed computer is stored in the storage unit of the selected computer.

According to the essential data management system of the present invention, it is possible to provide a data backup system that eliminates the need for a spare server, a backup dedicated server, and a shared disk.

1 is a configuration diagram of a spare image management system 1000 according to Embodiment 1. FIG. 6 is a flowchart showing an outline of operation of the spare image management system 1000 according to the first embodiment. FIG. 3 is a diagram illustrating a configuration related to registration of a spare image of the spare image manager 100 according to the first embodiment. 5 is a detailed flow of a preliminary image registration process in the first embodiment. FIG. 10 shows a data table 10A in the first embodiment. The figure which shows server table 20A in Embodiment 1. FIG. The figure which shows 30 A of data arrangement tables in Embodiment 1. FIG. FIG. 3 is a diagram illustrating an example of a difference created by a data differentiator 110 in the first embodiment. FIG. 3 is a diagram showing a configuration relating to the construction of a spare image of the spare image manager 100 in the first embodiment. 4 is a detailed flow of a spare image construction process in the first embodiment. The figure which made the flowchart of FIG. 2 a sequence. FIG. 2 is a block diagram showing an internal configuration of a preliminary image manager 100 in the first embodiment. FIG. 10 is a diagram illustrating a configuration related to registration of a spare image of the spare image manager 100 according to the second embodiment. The figure which shows data synchronization with the other server in Embodiment 2. FIG. The processing flow in the registration process in Embodiment 2. The flow which shows the content of S404 of FIG. The figure which shows VM table 40A in Embodiment 2. FIG. The figure which shows the data table 10B in Embodiment 2. FIG. The figure which shows the data arrangement table 30B in Embodiment 2. FIG. FIG. 10 is a diagram illustrating a configuration related to registration of a spare image in the third embodiment. 10 is a flow at the time of registration processing of the data placement destination determination device 120 in the third embodiment. The figure which shows the structure which incorporated the specific determination device in the data arrangement location determination device 120 in Embodiment 3. FIG. The flow at the time of incorporating the determination device of FIG. 22 into the process. The processing flow of the failure guarantee determination unit 122 in S511 of FIG. The processing flow of the data relationship determination device 123 in S513 of FIG. The processing flow of the resource reference | standard determination device 124 in S514 of FIG. The figure which shows the data table 10C in Embodiment 3. FIG. The figure which shows the data arrangement | positioning table | surface 30C in Embodiment 3. FIG. The figure which shows VM table 40B in Embodiment 3. FIG. FIG. 10 is a diagram illustrating an operation example of a spare image management system 1000 according to the third embodiment. The figure which shows the specific operation example on the assumption of FIG. The figure which shows the case where a failure generate | occur | produces in server Si in the state of FIG. The figure which shows the case where a failure generate | occur | produces in server Sk in the state of FIG. 10 is a flow showing a construction process of the spare image manager 100 in the fourth embodiment. 10 is a flow showing relocation processing of the spare image manager 100 in the fifth embodiment. The flow which refined | miniaturized FIG. FIG. 18 shows a processing flow in the sixth embodiment. FIG. 20 is an external view of an operational server S in a seventh embodiment. FIG. 18 shows hardware resources of an active server S in the seventh embodiment. The figure which shows a prior art. Another figure which shows a prior art.

Embodiment 1 FIG.
The meanings of main terms used in the following Embodiments 1 to 7 are as follows.
(Definition of words)
(1) Operational server (also simply referred to as a server): a computer provided with a CPU, a storage device and the like.
(2) Essential data: Data used by processing (3) Preliminary data: Data saved by the spare image manager (4) Original: Base data of difference when essential data is differentially processed (original data is included in the preliminary data) include)
(5) Patch: Data of the difference portion when the essential data is differentially processed (the patch is included in the preliminary data)
(6) Chunk: A piece of data obtained by dividing spare data (chunk is included in spare data)
(7) Mandatory data consisting of spare data: Mandatory data used when restoring spare data (8) Processing related to spare data: Processing requiring essential data consisting of spare data (9) Resource: CPU performance required for processing , Memory capacity, required devices, I / O throughput performance, etc.

In the first embodiment, a spare image manager that manages data (hereinafter referred to as a VM image) required for starting a VM as essential data in an environment in which a virtual machine (hereinafter referred to as a VM) is started in each server as processing. Will be described.

FIG. 1 shows a configuration of a spare image management system 1000 (essential data management system) according to the first embodiment. The spare image management system 1000 includes a plurality of operational servers S1 to SN. Each active server S includes a spare image manager 100 (essential data management unit), a storage device 200 (storage unit) that stores information, and a process execution unit 300 that executes a plurality of processes. The processes operating in the process execution unit 300 of each server are not the same, but are individual processes. That is, the processes operating on the active server S1, the active server S2,... Are different. Each server is connected to the network 400 and communicates with other computers via the network 400.

There are no restrictions on the number of processing and storage devices for each active server. As shown in FIG. 1, the configuration of each active server is the same, and each active server is equivalent. Note that there are resource differences between the servers.

FIG. 2 is a flowchart showing an outline of the operation of the spare image management system 1000. This flow is composed of two parts: steps S101 to S105, which are “registration of a spare image”, and steps S106 to S110, which are “construction of a spare image”.
FIG. 11 is a diagram showing the flow of FIG. 2 as a sequence. With particular reference to FIG.

(Overview of spare image registration operation: S101 to S105)
First, the outline (S101 to S105) of the preliminary image registration operation will be described with reference to FIGS.

(Server role)
The server that appears in the description of FIG. 2 is defined. In the description of FIG. 2, for simplicity, four systems of servers S1 to S4 are assumed.
(1) Reception server: a server that accepts user instructions (referred to as S1),
(2) possessed server: a server (referred to as S4) that possesses the essential data d to be stored;
(3) Storage destination server: A server (referred to as S2) that is a storage destination of essential data to be stored (registered).
(4) Construction destination server: a server (referred to as S3) on which a spare image using the essential data d is constructed,
(5) Master server: A server that is the main component of the registration operation of the spare image (assuming that S1 also serves).
The spare image manager 100 of the master server is called a master manager. For example, the master manager executes a selection process of a storage destination server.

(Difference between components)
The spare image manager 100 of the server S1 is distinguished by attaching “−1” as “spare image manager 100-1”. “-1” is also given to the constituent elements of the “preliminary image manager 100-1” appearing below. Although the spare image manager 100 of each server is the same, in the following embodiment, it is assumed that any one of the spare image managers 100 functions as a master when executing the processing of this system. Which spare image manager 100 becomes the master depends on the implementation. The master spare image manager 100 (M) controls the spare image manager 100 of each server, and executes processing related to the entire system, for example, determination of a storage destination server to be described later. (M) indicates a master.
In the following description, the server S1 is set as the master as described above. The server S1 also serves as a reception server.

(S101: The user instructs the reception server)
In step S <b> 101, the user inputs “registration instruction” (storage instruction) to the reception server using the server S <b> 1 as the reception server. Each server is equal, and the user may input a registration instruction to any server. If “registration instruction” is input to the server S2, the server S2 becomes a receiving server (storage instruction input computer).
"Registration instructions"
(1) Information (owned computer specifying information) for specifying the owned server S4 (owned computer),
(2) Information (essential data specifying information) for specifying essential data d (data to be registered),
(3) Includes three registration instructions for essential data d.
In response to the “registration instruction”, the spare image manager 100-1 of the server S1 (accepting server) accepts an instruction to save essential data d (data necessary for processing itself or data that is different from other data). In this example, the spare image manager 100-1 of the server S1, which is a reception server, functions as a master.
Note that the reception server is a master server. Any server can be the master. For example, when the registration instruction is input, the spare image manager 100-1 of the receiving server (server S1) communicates with the spare image manager 100 of the other server via the network 400 to thereby reserve the spare server of the other server. A server serving as a master may be determined with respect to the storage instruction with the image manager 100, and a registration instruction may be transmitted via the network to the spare image manager 100 of the server determined as the master.

(S102: Manager 100-1 (M) starts processing)
In step S102, the spare image manager 100-1 (M) causes the holding server S4 (user input) to start storing the essential data d in accordance with an instruction from the user. In other words, the spare image manager 100-1 (M) transmits a notification to start saving processing of the essential data d to the holding server S4 (user input). Receiving server start notification, possession server S4 starts the preservation | save process of the essential data d. This “save process” means a difference process or the like (described later) by the holding server S4. In the difference sentence process, the possessing server S4 communicates with the other servers S1 to S3 to acquire information about the difference, and performs the difference process.

(S103: Manager 100-1 (M) determines storage destination server)
In step S103, the spare image manager 100-1 (M) selects a server suitable for the storage destination of the essential data d held by the server S4 from the servers S1 to S4 including its own server S1. Select based on selection rules. The “predetermined selection rule” will be described in detail in the second and subsequent embodiments. Embodiments 2 to 6 are embodiments of a specific determination method of the storage destination server. Here, it is assumed that spare image manager 100-1 (M) determines server S2 (selected computer) from the “predetermined selection rule” as the storage destination server.

(S104: Manager 100-4 stores data in selected server S2)
In step S104, the spare image manager 100-1 (M) communicates with the spare image managers 100-4 and 100-2 of the holding server S4 and the storage destination server S2. Through this communication, the backup image managers 100-4 and 100-2 are notified that the server S2 has been selected as the storage destination server. The spare image manager 100-4 of the possessing server S4 transmits the difference data to the spare image manager 100-2 of the storage destination server S2. The spare image manager 100-2 stores the received difference data.

(S105: Completion notification)
In step S105, the preliminary image manager 100-1 (M) notifies the user who issued the instruction in S101 of the completion of storage. This is triggered by the fact that the spare image manager 100-1 (M) receives a transmission completion notification from the spare image manager 100-4 or receives a reception completion notice from the spare image manager 100-2. To do.
The above is the process up to registration of the spare image.

(Construction operation: S106 to S110)
Next, an outline (steps S106 to S110) of the preliminary image construction operation will be described. In S106 to S110, since the holding server S4 is down, the construction destination server S3 uses the essential data d (original data of VM (A) or difference data of VM (A)) stored in the storage destination server S2. Assume that a spare image (a spare image of the VM (A) operating on the owned server S4) is constructed. Described below is the operation of re-operating the processing that was running on the server where the failure occurred (server S4) on another server (server S3) (operation for constructing a spare image using the essential data d). To do.

(S106: The user instructs the server S1)
In step S106, the user issues a construction instruction to the server S1 (receiving server). As with the registration process, each server is equivalent, and the user may instruct any server to construct. The server that has received the construction instruction becomes the acceptance server (construction instruction input computer).
"Build instruction"
(1) Information for specifying a preliminary image to be constructed (execution use data specifying information),
(2) Information (construction computer identification information) for identifying a construction destination server (in this example, server S3),
(3) Construction command,
Of three. The user does not need to recognize the storage destination server.
What is necessary is to recognize the VM image to be constructed and the construction destination server to construct the VM image.

(S107: Start construction processing)
In step S107, the preliminary image manager 100-1 starts the corresponding data construction process in accordance with the construction instruction from the user. The spare image manager 100-1 notifies the construction destination server S3 of the contents of the construction instruction.

(S108: specific)
In step S108, the spare image manager 100-3 of the construction destination server S3 that has received the notification is necessary for constructing data of the spare image to be constructed (in this example, VM (A) below) based on the notice. The essential data d (original data a0, difference data a1 and the like shown below) and the storage destination server (server S2) where the essential data d is stored are specified.
The spare image manager 100-3 specifies essential data such as original data a0 and difference data a1 necessary for data construction from a data table 10A (FIG. 5) described later, and a data arrangement table 30A (FIG. 7) described later. From the above, specify the storage destination server of essential data. The relationship between the VM (A) in this case and the essential data d necessary for the construction of the VM (A) will be briefly described. As will be described later with reference to FIG. 8, it is assumed that VM (A) is constructed from original data a0, difference data a1, difference data a3, and difference data a11 as shown in FIG.
Here, the original data a0 and the like are as follows.
(1) Original data a0: Original data created when VM (A) was first registered,
(2) difference data a1: difference data from the original a0 created when the VM (A) is updated and registered again,
(3) Difference data a3: Difference data a3 from difference data a1 created when VM (A) is updated and registered again
(4) Difference data a11: Difference data from difference data a3 created when VM (A) is updated and registered again Original data a0, difference data a1, etc. are the difference processing of holding server S4 in FIG. Alternatively, the data stored in the storage destination server when the servers S1 to S3 are owned servers.

(S109: Construction)
In step S109, the spare image manager 100-3 of the spare image construction destination (server S3) communicates with the spare image manager between the servers, and the essential data necessary for construction is transferred to the server for storing the essential data ( Acquired from the server S2) and constructs data (VM (A)).
That is, the construction destination server S3 acquires essential data storage destination server essential data from the identification results of the data table 10A and the data arrangement table 30A, and constructs a preliminary image (VM (A)) using the obtained essential.

(S110: Manager 100-1 notifies completion of construction)
In step S110, the construction destination server S3 transmits a construction completion notification to the receiving server S1, and the spare image manager 100-1 notifies the construction completion to the user who issued the “construction instruction” (for example, the display device completes the display device). Display). As described above, S106 to S110 are the preliminary image construction processing.

In this way, essential data (a0, a1, a3, a11, etc.) is stored in another operational server (server S2, etc.), and when a certain server (server S4) fails, a certain server (server S4) Data construction (for example, construction of the spare image VM (A) using the essential data d) is performed on the server (server S3) other than the server in which the failure has occurred, based on the essential data d serving as a backup of the processing executed in step S2. By doing so, it is possible to take over data in the system.

When applied to a VM, the VM is activated as “processing” on each server, and the VM stores its actual VM data on a non-shared disk for each server. The spare image manager of each server obtains a VM image (execution usage data) in the storage device according to an instruction from the system user, transfers it to another server, or similarly stores the stored VM image according to the instruction. Restore on the target server.

(Details of spare image registration operation)
Next, details of the preliminary image registration operation (S101 to S105 in FIG. 2) will be described.
FIG. 12 is a block diagram showing an internal configuration of the preliminary image manager 100. As shown in FIG. 12, the preliminary image manager 100 includes a data differentiator 110, a data placement destination determiner 120, a data distributor 130, and a data compounder 140.
(1) When the data differentiator 110 is a holding server, the data differentiator 110 creates difference data for registration in the storage destination server.
(2) The data placement destination determination unit 120 selects a storage destination server when it is a master server.
(3) The data distributor 130 exchanges data with other preliminary image managers 100.
(4) The data compounder 140 constructs a spare image when it is a construction destination server.

FIG. 3 shows components related to the registration of the spare image with respect to the internal configuration of the spare image manager 100 shown in FIG. In FIG. 3, the data differentiator 110 and the data distributor 130 are mainly owned servers, and the data placement destination determiner 120 is mainly a master server.
FIG. 3 shows both the component part of the owned server and the component part of the master server. That is, FIG. 3 shows a state in which the master server selects a storage destination server, and the holding server transmits data to the selected storage destination server.

FIG. 3 shows a series of flow from generation of difference data stored in the storage destination server to distribution of difference data.
(1) The data difference unit 110 of the owned server cooperates (communications) with the spare image manager 100 of another server, and refers to the data table 10A (FIG. 5) based on the input data 5 (VM image). Then, a difference is calculated with the existing registered data (information is acquired by cooperation), the result is reflected in the data table 10A, and the difference data 7 is output.
(2) The master data location determination unit 120 (M) cooperates with the spare image manager 100 of another server, and based on the difference data 7, the data table 10A, and the data location table 30A created by the holding server. Then, a server suitable for storing the difference data 7 is selected, and the result is reflected in the data table 10A and the data arrangement table 30A (data arrangement information) (FIG. 7).
(3) The data distributor 130 of the owned server transfers the difference data 7 to the storage destination server in accordance with the data arrangement table 30A (latest state).

(Synchronization of data table 10A and data arrangement table 30A)
The data table 10A and the data arrangement table 30A are synchronized with other servers by the spare image manager 100.

(Specific processing of registration)
FIG. 4 shows a detailed processing flow in the preliminary image manager 100 in the registration processing (S101 to S105 in FIG. 1). As in FIGS. 2 and 11, the server S1 is the accepting server, the server S2 is the storage destination server, the server S3 is the construction destination server, and the server S4 is the holding server. Steps S201 and S202 are processes performed in step S102, steps S203 and S204 are performed in step S103, and step S205 is a process performed in step S104.

(S201)
First, in step S201, the data differentiator 110-4 of the possession server S4 performs difference data (for example, a1) of data (VM (A)) for which registration is instructed (the possession server S4, VM (A) is input by the user). ) Is generated.

(S202)
Next, in step S202, the data differentiator 110-4 of the holding server S4 reflects the information regarding the difference in the data table 10A-4 (FIG. 5).

(S203: Selection by master)
In step S203, the data placement destination determination unit 120-1 (M) acquires a server group as a storage destination candidate from the server table 20A-1 (FIG. 6). The server table 20A-1 in FIG. 6 is address information of each server.

(S204)
In step S204, the data placement destination determination unit 120-1 (M) selects a server suitable for storing the difference data 7 from the candidates, and reflects the result in the data placement table 30A-1 (FIG. 7). The data allocation table 30A-1 is synchronized with other servers.

(S205)
Finally, in step S205, the data distributor 130-4 of the holding server S4 updates the difference data 7 in synchronization with the data arrangement table 30A-1 updated by the data arrangement destination determination unit 120-1 (M). The data is transferred to the storage destination server S2 according to the data arrangement table 30A-4.

FIG. 8 shows an example of the difference created by the data difference unit 110 of the owned server.
a0 is data (original) created when VM (A) is first registered.
a1 is difference data from the original a0 created when the VM (A) is updated and registered again.
a2 is difference data from the original a0 created when VM (D) was first registered as a derivative from VM (A).
a3 is difference data from the difference data a1 created when the VM (A) is again updated and registered again.
As described above, the spare image manager 100 performs management by dividing the data of a plurality of VMs into original data and differential data according to differences.

(Details of spare image construction operation)
Next, the details of the preliminary image construction operation will be described.

FIG. 9 shows a configuration relating to the construction of a preliminary image in the internal configuration of the preliminary image manager 100 in FIG. FIG. 9 shows the configuration of the construction destination server.
(1) The data distributor 130 receives specific difference data 13 from another server. In this example, the construction destination server is the server 3.
(2) The data compounder 140 combines the inputted difference data 13 to construct a preliminary image. Regarding the construction process, it is the construction destination server that performs data composition. In this example, it is the server 3.

FIG. 10 shows a detailed processing flow in the spare image manager 100-3 of the construction destination server S3 in the construction processing. Step S301 is processing performed in step S108, and steps S302 to S304 are processing performed in step S109.

(S301)
In step S301, the spare image manager 100-3 of the construction destination server S3 performs the data group (a0, a1, a3, a11) necessary for constructing the “spare image instructed by the user” (for example, VM (A)). Is obtained from the data table 10A-3. Thereafter, in accordance with the tree structure of the data group, steps S302 to S304 are repeated for any difference data sequentially from the original.

(S302)
First, in step S302, the data distributor 130-3 of the construction destination server specifies a storage destination server in which the difference data of the data group acquired from the data arrangement table 30A is stored.

(S303)
Next, in step S303, the data distributor 130-3 refers to the server table 20A (address information), cooperates with (communicates with) the data distributor 130 of another server, and is necessary for constructing the VM (A). The difference data is acquired from the possession server having such data.

(S304)
Thereafter, in step S304, the data composite unit 140-3 of the construction server reflects the acquired difference data in the preliminary image to be constructed. When all the difference data has been reflected, the construction process is completed. In the example of VM (A), the data compounder 140-3 reflects the difference data a1, a3, a11 in the original data a0, and constructs VM (A).

For processes other than the VM, the same effect can be used in an environment where other processes are performed by replacing the portion corresponding to the VM with arbitrary processing and the portion corresponding to the VM image with arbitrary essential data. .

Embodiment 2. FIG.
In the second embodiment, an example corresponding to the primary failure is shown by improving the first embodiment and judging the operation state of the VM. When the master spare image manager 100 (M) selects a storage destination server for essential data, the server in which the VM (A) related to the essential data d (A) is operating is excluded from the storage destination candidates. Indicates. For example, when selecting the storage destination server of the difference data a3 of VM (A), the server in operation of VM (A) is excluded from the storage destination server. This is to avoid collapsing between the operating VM (A) and the difference data a3 necessary for constructing the VM (A) when the server operating the VM (A) goes down. The system of the servers S1 to S4 is the same as in the first embodiment, and the receiving server is the same.
Embodiments 2 to 6 are embodiments relating to selection of a storage destination server by the spare image manager 100 (M) as a master.

Hereafter, the processing at the time of registration of the spare image will be mainly described.

FIG. 13 shows a configuration relating to registration of a spare image in the second embodiment.
FIG. 14 is a diagram illustrating synchronization with other servers. FIG. 14 shows a master spare image manager 100 (M).

(1) As in the case of the first embodiment, the data difference unit 110 of the owned server cooperates with the spare image manager of another server, and creates an existing registration based on the input data 5 (VM image). A difference is taken from the completed data (for example, a3 for a1), the result is reflected in the data table 10B (FIG. 18), and difference data 7 is output. In the data table 10B, in addition to the data table 10A created by the data differentiator 110 of the first embodiment, a VM name related to the data 5 (VM image) is also registered. This is the operation of the holding server.
(2) The data placement destination determination unit 120 (M) cooperates with the spare image management unit 100 of another server, and based on the information in the data table 10B, the difference data 7, the server table 20A, and the VM table 40A (FIG. 17). Then, a server suitable for storing the difference data 7 is selected, and the result is reflected in the data arrangement table 30B (FIG. 19) and the data table 10B. In addition, the data placement destination determination unit 120 (M) determines to provide a copy (described later) of the difference data 7 if necessary. When the data placement destination determination unit 120 (M) determines that a copy should be provided, the data difference unit 110 of the holding server creates the difference data 19 according to the determination. The “duplication” of the difference data 7 will be described later. This data location determination unit 120 (M) is the data location determination unit 120-1 (M) of the master server. The data distributor 120-4 of the holding server transfers the difference data 7 to the storage destination server according to the data allocation table 30B-4. The data allocation table 30B-4 in FIG. 19 has a replication number indicating the number of replications. The copy number “0” is a case where no copy is made. For example, the difference data a3 is transmitted to one storage destination server. The copy number “1” indicates that one copy should be created. The data difference unit 110 of the holding server generates one copy of the difference data a3 and transmits the two difference data a3 to the storage destination server selected by the data placement destination determination unit 120-1 (M). (3) The operation VM determination unit 121 (M) exists in the data placement destination determination unit 120. The operation VM determination unit 121 determines an arrangement destination of essential data based on the operation state of the VM.

(Synchronized)
The data table 10B and the data arrangement table 30B are synchronized with other servers by the spare image manager.

The VM table 40A is information acquired from the server's own VM monitor in a timely manner, and indicates the VMs operating on the server.

FIG. 15 shows a processing flow in the preliminary image manager in the registration process, and the contents are the same as those in FIG. 4 except for the step number.
FIG. 16 shows a detailed operation example of S404.

(S401)
In step S401, the data differentiator 110-4 of the owned server S4 generates a difference of the data d instructed to be registered.

(S402)
In step S402, the data differentiator 110-4 reflects the information regarding the difference in the data table 10B-4.

(S403)
In step S403, the data placement destination determination unit 120-1 (M) acquires a server group as a storage destination candidate from the server table 20A-1.

(S404)
In step S404, the data location determination unit 120-1 (M) selects a storage destination server (server S2) suitable for storing the difference data 7 from the storage destination candidates, and the result is stored in the data allocation table 30B-1. And reflected in the data table 10B-1.

(S405)
In step S405, the data distributor 120-4 of the holding server S4 transfers the difference data 7 to another server according to the data arrangement table 30B-4 (after synchronization).

Next, detailed operations (S411 to S417) in step S404 (selection of storage destination server) will be described with reference to FIG.

(S411)
In step S411, the data location determination unit 120-1 (M) initializes the server candidate list. Next, steps S412 to S413 are performed for all servers in the server table 20A-1. In S411 of FIG. 16, “OUT” indicates a server candidate list.

(S412)
In step S412, the operation VM determination unit 121-1 of the data placement destination determination unit 120-1 (M) is a VM table of VMs operating on the target server (in this case, all of the servers S1 to S4). 40 (processing operation information) is acquired. At this time, the operation VM determination unit 121-1 performs step S413 only when there is no VM related to the difference data 7 in the acquired VM table 40, and adds the target server to the server candidates. In other words, when the difference data 7 to be saved (essential data, for example, difference data a3) is related to the VM (A), only the server on which the VM (A) is not operating becomes the save destination candidate (outS).
After S412 to S413 are completed for all the servers, in step S414, the data placement destination determination unit 120-1 (M) performs S415 and S416 only when the server candidate is empty.

(S415)
In S415, the data placement destination determination unit 120-1 (M) sets “data replication count” to “2” and sets all servers in the server table 20A-1 as server candidates. The fact that the storage destination candidate (outS) is “none” in S414 means that the VM (A) is operating in all the servers S1 to S4 in the previous example.

(S416)
In S416, the data location determination unit 120-1 (M) updates the data table 10B-1 and updates the number of data copies to 2. The effect of the number of data replications being “2” is that the same difference data 7 regarding VM (A) is stored in “two” servers. The number of copies “2” is an example. If the number of data replications is “k”, it is stored in “k” servers. In this sense, the phrase “number of copies” is used.

(S417)
In S417, the data placement destination determination unit 120-1 (M) selects servers for the number of replicas from the server candidates and reflects the selection result in the data placement table 30B-1. As for the copy number in FIG. 18, when the number of copies is “2”, two difference data 7 of “0” and “1” are registered.

As a result, the data is stored on the server where the VM related to the data to be registered is not operating. Alternatively, for example, when the difference data a3 related to the VM (A) is to be stored and the VM (A) is operating on any server, the difference data a3 is duplicated. Therefore, the replicated data is stored in separate servers. Therefore, data recovery can be reliably performed for a single failure.

Embodiment 3 FIG.
Similar to the first embodiment, in the preliminary data management in which the VM image is essential data in the environment activated on the server with the VM as a process, optimal data storage is performed so that data transfer does not occur as much as possible when the VM is constructed. An example of determining the destination will be shown. In this example, the same data related to construction is collected on the same server, and data transfer is minimized.
The third embodiment also relates to selection of a storage destination server by the master, as in the second embodiment.

As in the first embodiment, the system of the servers S1 to S4 is used, and the receiving server is the same.

FIG. 20 shows a configuration of the data placement destination determination unit 120 related to the registration of the spare image in the third embodiment. The data placement destination determination unit 120 outputs placement candidates (in the case of a master) based on inputs such as the data table 10C, the server table 20A, the data placement table 30C, the VM table 40B, and placement data (difference data). Output the actual data to be placed (in case of owning server). In addition, the data placement destination determination unit 120 updates the input data.

(Each judger)
The determination units 1 (first selection algorithm) to determination unit n (nth selection algorithm) existing in the data placement destination determination unit 120 are different from the operation VM determination unit 121 and have different indexes (selection criteria). Determine the location (storage destination server). The data placement destination judging device 120 selects placement candidates by repeatedly making decisions while changing the judging device. In this example, the data placement destination determination unit 120-1 (M) operates as a master.

(Registration process A flow)
FIG. 21 shows a processing flow at the time of registration processing in the data placement destination judging device 120-1 (M).

(S501)
First, in step S501, the data arrangement destination determination unit 120-1 (M) acquires an initial arrangement candidate using the operation VM determination unit 121-1. This step S501 is the content of FIG. “OutS” (number of servers) and “OutCount” in S501 are the results obtained in S417.

Thereafter, the data placement destination determination unit 120-1 (M) repeats S502 to S506 while changing the determination units by the number of copies (the number of OutCounts) acquired in step S501. This repetition is repeated by the number of “OutCount”, that is, the number of copies (the number of storage destination servers). That is, one difference data storage destination server is selected in one loop.

(S502)
First, in step S502, the data arrangement destination determination unit 120-1 (M) copies the temporary candidate (TmpS (1)) from the candidate list OutS and initializes the division number (PartCount) to 1. Here, the “number of divisions” means the number of division of the difference data to be saved. The number of divisions = 1 means that the difference data is not divided, the number of divisions = 2 means that the difference data is divided into two, and the number of divisions = k means that the essential data is k pieces. It means to divide.

(S503)
Next, in step S503, the determiner 1 set to operate next to the operation VM determiner 121-1 is selected based on the arrangement candidate (S501: OutS), the arrangement data, and the number of divisions (S502). Select the server.

(S504)
Thereafter, in step S504, the data placement destination determination unit 120-1 (M) performs S506 only when the number of candidate results (TmpS (2)) in S503 is the same as the “number of divisions” (NO in S504). , End and proceed. For example, when the number of candidates in S503 (the number of candidates for the storage destination server) is one and the “number of divisions” of the essential data d is “1”, the process proceeds to S506 and the end.

(S505)
If the condition is not satisfied (YES in S504), in step S505, the data placement destination determination unit 120-1 (M) switches the determination unit (changes the server selection algorithm) and returns to S502 again. That is, the data placement destination determination unit 120-1 (M) determines whether or not “number of divisions” = “number of server candidates” by another determination unit (server selection algorithm).

(S506)
Finally, in step S506, the data placement destination determination unit 120-1 (M) reflects the server of the temporary candidate (TempS (2)) in the data placement table 30B-1 as the storage destination, and the candidate “ The server that is a temporary candidate (TempS (out)) is deleted from “OutS”. As the final determination device n, a determination device (simple determination device 125 described later) that is NO in S504 is prepared.

FIG. 22 shows a configuration example in which several types of specific determiners are actually incorporated in addition to the operation VM determiner 121. In FIG. 22, in addition to the operation VM determination unit 121, four determination units including a failure guarantee determination unit 122 to a simple determination unit 125 are incorporated.
(1) The failure guarantee determination unit 122 selects a server capable of VM recovery when an unspecified server is stopped.
If the failure guarantee determination unit 122 selects a storage destination server and assumes that another server is down, can the VM operating in the down server be constructed as a storage destination server? To simulate. However, the VM constructability related to the data stored in the server to be stored is simulated.
(2) The data relation determination unit 123 selects a server in which a lot of essential data related to the same spare data is stored among the selected servers. For example, the original data a0 and the difference data a1, a3, and a11 are required to construct the VM (A) as the preliminary data. When a0 is stored in the server S1 and a1, a3 are stored in the server S2, the data relationship determination unit 123 includes the servers S1 and S2 as candidates, and when the difference data a11 is stored, the related data Is selected as the storage destination of the difference data a11.
(3) The resource criterion determination unit 124 selects a server having the most free resources from the selected servers.
(4) The simple determiner 125 selects servers for the appropriate number of divisions from the candidates. Therefore, in the determination by the simple determiner 125, S504 is always NO, and the process proceeds to S506.

FIG. 23 is a flow in the case where these four determiners are incorporated into the processing with respect to FIG. With reference to FIG. 23, a case where the determiners 121 to 125 are incorporated will be described.

In steps S511 and S512, the failure guarantee determination unit 122 performs processing. In step S513, the data relationship determination unit 123 performs processing. In step S514, the resource criterion determination unit 124 performs processing. In step S515, the simple determiner 125 performs processing. In step S512 of this example, the failure guarantee determination unit 122 and the data placement destination determination unit 120 perform processing in cooperation.
The processing outline of FIG. 23 is as follows.
When a plurality of storage destination candidate numbers N5 are selected by the processing of the failure guarantee determination unit 122, the data relationship determination unit 123 further narrows down from the plurality of storage destination candidate numbers N5. When a plurality of storage destination candidate numbers N4 are selected by the processing of the data relationship determination unit 123, the resource reference determination unit 124 narrows down from the number of storage destination candidate numbers N4.
When a plurality of storage destination candidate numbers N3 are selected by the resource criterion determination unit 124, the simple determination unit 125 finally determines a predetermined number of storage destination candidates from the plurality of storage destination candidate numbers N3.

(S511)
In step S511, the failure guarantee determination unit 122 is a temporary candidate (TempS (2)) from the candidate (TmpS (1)) passed from the operation VM determination unit 121 only for a server that can recover the VM when an unspecified server is stopped. ) To select. If the result (temporary server candidate) is smaller than the number of divisions (S502), the failure guarantee determination unit 122 increments the “number of divisions” by 1 in step S512 and repeats step S511. For example, when out (TempS) = 0, the “number of divisions” is incremented by 1 from 1 and becomes 2.

(S513)
If the number of server candidates is greater than the “number of divisions”, the process proceeds to step S513. In step S513, the data relationship determination unit 123 selects, from the servers selected from the server candidates (TempS) passed from the failure guarantee determination unit 122, servers that have a lot of other data related to the data to be stored as candidates.

(S514)
In S514, when the number of server candidates selected in S513 is different from the number of divisions, the resource criterion determination unit 124 uses the resources (CPU, memory, disk capacity, device, I / I) from the server candidates passed from the data relationship determination unit 123. A server having an available computer resource such as O throughput is selected as a storage destination candidate server.

(S515)
When the number of candidates processed in S514 is different from the number of divisions, in step S515, the simple determiner 125 appropriately selects servers for the number of divisions from the candidates passed from the resource criterion determiner 124.

Next, the operations of the failure guarantee determination unit 122, the data relationship determination unit 123, and the resource reference determination unit 124 will be described with reference to FIGS.

(Specific processing of the fault guarantee determination unit 122)
FIG. 24 is a flowchart showing specific processing performed by the failure guarantee determination unit 122 in step S511 shown in FIG. In FIG. 24, the failure guarantee determination unit 122 is an operation subject. FIG. 24 is a flow for determining whether or not spare data can be constructed in the candidate server by the failure guarantee determination unit 122. That is, the failure guarantee determination unit 122 confirms whether the storage destination server can also be the construction destination server.

The process consists of four nested loops.
In step S511, the failure guarantee determination unit 122 simulates that an unspecified server has stopped, and determines whether or not recovery can be supported even when data is stored.

(1) First, in step S521, the failure guarantee determination unit 122 acquires a VM related to data to be stored.
(2) Next, in step S522, the failure guarantee determination unit 122 initializes the temporary candidate NewTempS.
(3) After that, in step S523, the failure guarantee determination unit 122 covers combinations of the number of divisions (PartCount) from among the server candidates passed from the operation VM determination unit 121, and sets each as a storage destination candidate group C. Steps S524 to S530 are performed. Specifically, for example, it is assumed that there are four server candidates passed and the number of divisions is two. In that case, two pieces of data are stored in two of the four servers. Therefore, as the candidate group C, “ ₄ C ₂ ways” can be combined.
(4) In step S524, the failure guarantee determination unit 122 covers all combinations when the VMs related to the data to be stored are divided into division numbers, and steps S525 to S530 are used as VM combination candidates D to be restored. I do. In steps S525 to S529, a case where a specific server sj is stopped is simulated.
(5) First, in step S525, the failure guarantee determination unit 122 acquires the VM list Va operating on the server sj.
(6) Next, in step S526, the parameter cnt is initialized. Thereafter, steps S527 to S529 are repeated for the number of elements of D.
(7) In step S527, the failure guarantee determination unit 122 acquires a VM list Vs related to data stored in the cnt-th server of the candidate group C.
(8) Next, in step S528, among the sum of the cnt-th VM group of the candidate group D and Vs, the one included in Va is set as Vu.
(9) After that, it is determined whether or not there is an available resource in the cnt-th server of the candidate group C that can build all the VM groups Vu. In step S530, the combination C in which vacancies exist for all the failure determinations of sj is added to the storage destination candidate to be output. After this is executed for all combinations of C and D, the output candidate thrives as a temporary candidate in step S531.

(Specific processing of the data relationship determination unit 123)
FIG. 25 shows a detailed flow of step S513. These are all processed by the data relationship determination unit 123. The processing includes a first half (steps S541 to S545) for calculating the data relation level of each server and a second half (steps S547 to S551) for selecting a server having a high degree of relation from among the first half.
(1) First, in step S541, the data relationship determination unit 123 initializes the data relationship level (Rates) of each server, and stores VM groups related to data to be stored in Vd.
Thereafter, Steps S542 to S545 are performed for all the candidate servers.
(2) In step S542, the data relationship determination unit 123 acquires a data group stored in the server si to be determined.
(3) Next, in step S544, the data relationship determining unit 123 sets the degree of relationship to 0, and then performs step S545 on all Vd VMs as VMVs.
(4) In S545, the data relationship determination unit 123 acquires the ratio that the data necessary for the construction of the VM V is stored in the server si, and adds it to the relationship level.
(5) Thereafter, in S546, the data relationship determination unit 123 stores the previous relationship level as the relationship level of the server si.
(6) Next, the data relationship determination unit 123 selects a candidate server based on the calculated degree of relationship. First, in step S546, the data relationship determiner 123 initializes the number of candidates and candidate groups that have been output.
(7) In step S547, the data relationship determiner 123 acquires the largest candidate group from the relationship rate Ratings.
(8) Thereafter, in steps S548 and 549, the data relationship determiner 123 adds the previous candidate to the output candidate, and deletes the candidate from the relationship list in step S550. (9) Thereafter, when the number of candidates to be output is smaller than the number of divisions in step S551, the process returns to step S548.

(Specific processing of the resource criterion determiner 124)
FIG. 26 shows a detailed flow of step S514. These are all processing of the resource standard determination unit 124.
(1) First, in step S561, the resource criterion determination unit 124 initializes the number of candidates to be output and the candidates.
(2) Next, in step S562, the resource criterion determination unit 124 obtains a server having the largest free resource amount from the input candidates in cooperation with the spare image manager of another server.
(3) Thereafter, in steps S563 and S564, the acquired server is added to the output candidates, and the candidates added from the candidates input in step S565 are deleted.
(4) After that, if the number of candidates to be output is smaller than the number of divisions in step S566, the process returns to step S548.
(5) Finally, in step S567, the output candidate is reflected in the final candidate.

With the combination of these determiners, in the third embodiment, data transfer is prevented by aggregating preliminary data necessary for constructing essential data so that data transfer does not occur as much as possible. In aggregation, on the basis of whether there are enough resources to start the process to be taken over on that server, leaving the room for processing on each server, the relationship between the spare data that has been placed, Considering the relationship between the processes related to the data, the data placement destination is determined.

In this example, the data table and the data position table are as shown in FIGS. 27 and 28 in order to correspond to the division. Further, since it is necessary to determine the resources necessary for starting the VM, the VM table 40 is as shown in FIG.

Next, an actual operation example of this example is shown.
30 to 33 show the processing results of the operation VM determination unit 121, the failure guarantee determination unit 122, the data relationship determination unit 123, the resource reference determination unit 124, and the simple determination unit 125 described above.
Assume that VMs A to J are operating on servers i to l (FIG. 30). Further, it is assumed that VMs A to J are composed of preliminary data shown in FIG. In this example, spare data is stored one by one in the order from left to right in FIG. 8 (a0 → b0 → a1 → a2 → b1 → b2 → a3 → a4 → a5...) In the environment where the VM is already operating. Describe the case. Each server is assumed to be a disk of 270 GB, a VM of 40 GB, a0 and b0 of 40 GB, a1 to 14, and b1 to b9 of 5 GB, and resource determination is performed on the disk capacity.

(Fig. 31)
In this case, determination of the arrangement of a0 to a14 and b0 to b9 is performed as shown in FIG. In this example, a0 is collectively stored in one server, and b0 is divided and stored in two servers. From a1 and b1 onward, each server is assigned based on the degree of relevance of the data, and finally the configuration at the bottom of FIG. 31 is obtained.
“Si to Sl” in the upper part of FIG. 31 indicates the following.
<Upper row>
In addition, when considering the free space at the time of storage and the data relationship, there is still room for constructing three VMs at the present time, even if all of the a0 groups are stored in the most free S1. When all of a0 are aggregated in Sj, when the b0 group is stored, the VMs related to b0 are Si and Sj that are not operating. 150GB is required to recover the VM related to b0, but no server is satisfied at the time of storing b0. Startup is performed on multiple servers. Equally distribute the minimum number of servers required at startup.
“Si to S1” in the middle of FIG. 31 indicates the following.
<Middle stage>
In the storage of a1, since there is 150 GB of free space after storing in Sl having a strongly related a0, it is stored in the server l. The same applies to a2. In the storage of b1, since the related data amount and the free capacity are the same amount, it is appropriately stored in Si. In the storage of b2, b2 and b1 are irrelevant and the related data amount is the same, but since the free capacity of Sj is large, it is stored in Sj. Hereinafter, similarly, the storage destination is determined based on the availability of VM activation, the amount of related data, the free capacity, and the like.
“Si to S1” in the lower part of FIG. 31 indicates the following.
<Lower row>
When Si is stopped, when A, B, and C are reconstructed, data transfer does not occur when S1 is constructed. When Sk stops, when reconstructing G, H, and I, when G is constructed in Si and H and I are constructed in Sj, a total of 50 GB is transferred.

(Fig. 32)
In this configuration, when the VM is recovered with the data flow as shown in FIG. 32 for the failure of the server i and with the data flow as shown in FIG. 33 for the failure of the server k, the transfer amount is the highest. There are few and the start in the shortest becomes possible. In FIG. 32, when Si is stopped, if A, B, and C are reconstructed in S1, data transfer does not occur. In FIG. 33, when Sk is stopped, when reconstruction of G is performed by Si and reconstruction of B and C is performed by Sj, a total of 50 GB is transferred. In FIG. 33, b0 / 1 in Sj is necessary for the construction of G in Si, and b0 / 0 in Si is necessary for the construction of H and I in Sj.

Embodiment 4 FIG.
As in the first embodiment, when an instruction is issued to restore a VM image at a specific server in the environment in which the VM image is processed as an indispensable data in an environment in which the VM is running as a process, the specification is performed. An example in which there is not enough disk space to construct a VM image on the server of FIG.

In this example, by re-moving the data that accompanies the process takeover, the resource shortage at the actual startup is resolved and the next failure is dealt with. This configuration conforms to FIGS. 9 and 20.

FIG. 34 shows a processing flow at the time of constructing the spare image manager in this example.
(1) First, in step S601, the data compounder 12 determines whether there are enough resources available for construction.
(2) If there is no room, in step S602, the spare image manager is improperly placed on its own server for data not related to the construction among the data stored in the construction destination server. Select some data.
(3) In this selection, the data arrangement destination determination unit 22 determines again an appropriate arrangement destination for all the target data, and treats data different from the server whose result is already stored as a candidate.
(4) As soon as candidates are found, the preliminary image manager moves data to another server using the data distributor in step S603 by an appropriate number from the candidates.
Thereafter, the process is repeated until the resource becomes available.

Embodiment 5 FIG.
As in the first embodiment, the processing for optimizing the arrangement of data in the server when there is a margin in processing such as late at night in the spare data management in which the VM image is essential data in the environment in which the server is started with the VM as a process Show about.

In this example, the spare image manager performs data rearrangement using the data arrangement destination determination unit 22 and the data distributor 19 provided therein. The spare image manager directly uses the failure guarantee determination unit 29 in the data arrangement destination determination unit 22.

The spare image manager passes a provisional table to the data placement destination judgment unit 22 and the failure guarantee judgment unit 29, and acquires the placement destination as provisional. Further, the result is reflected in the data arrangement table 30, and the data distributor 19 is instructed to redistribute. The data placement destination determination unit 22 calculates a temporary placement destination from the input information passed as temporary. The failure guarantee determination unit 29 returns whether or not the designated server is included in the selection candidate from the temporary input information passed from the spare image manager. The data distributor 19 transfers data according to the data arrangement table 30 determined by the preliminary image manager.

FIG. 35 shows a processing flow at the time of rearrangement processing of the spare image manager in this example. The flow consists of three steps, S701, S702, and S703.
(1) In step S701, the spare image manager activates the data placement destination judgment unit 22 using the placement patterns of all the data managed by the spare image manager as provisional input information, and moves the data at the most reconstruction time. Select an arrangement pattern with a small amount.
(2) Next, in step S702, the spare image manager selects a combination of data that satisfies the conditions selected by the operation guarantee determination unit 29 and the optimal placement destination server from among the data that does not conform to the optimum placement. Data is transferred to the target server using the data distributor 19.
(3) Finally, in step S703, the spare image manager unconditionally transfers the remaining data that does not conform to the optimal arrangement to the target server using the data distributor 19.

The flow in FIG. 36 is a detailed modification of the flow in FIG. Step S701 corresponds to steps S711 to S715. Step S702 corresponds to steps S721 to S729. Step S703 corresponds to steps S731 to S736.

First, in step S711, the spare image manager determines whether the process has been resumed from the interrupted state or whether it is the initial start based on whether re-verification is necessary. When the determination is necessary, the preliminary image manager changes the forced movement flag force to false in step S712, and selects an optimum arrangement destination using the data arrangement destination determination unit 22 in step S713. In step S714, if the forced movement flag force is true, that is, if the previous forced movement was performed in the interruption process, the process jumps to step S732. In step S715, it is determined whether the current arrangement is equal to the optimum arrangement. If they are equal, the process ends. If not, the process jumps to step S721.

In step S721, the preliminary image manager initializes the data movement flag flag to false, and then moves the data that satisfies the conditions of the failure guarantee determination unit 29 among the data that is not optimally arranged.

First, in step S722, it is determined whether or not to interrupt, and only when not interrupted, the process jumps to step S723 to acquire the placement destination server sk of the data d that is a candidate for current movement verification. Next, the process jumps to step S725 only when the placement destination server is not equal to the current server, and jumps to step S726 only when the condition of the operation guarantee device is satisfied even if the data d is moved to the server sk. In step S726, the preliminary image manager updates the data arrangement table 30, instructs the data distributor 19 to move the data, and moves the data d to the server sk. If even one piece of data has been moved, in step S727, the spare image manager truly updates the data movement flag flag. Thereafter, in step S728, it is determined whether the current arrangement and the optimum arrangement are equal. If they are equal, the process ends. If not, the process jumps to step S729. In step S729, the spare image manager jumps to step S721 only when the data is moved under the condition of the failure guarantee determination unit 29 and tries again whether the data can be moved under the condition of the failure guarantee determination unit 29.

If there is no data moved in step S726, in step S731, the spare image manager sets the forced movement flag force to true and shifts to the forced movement state. In the forced movement state, first, in step S732, the spare image manager determines in step S732 whether or not to interrupt, and jumps to step S733 only when not interrupted. In step S733, the preliminary image manager selects one of the data different from the optimum pattern, and in step S734, the data arrangement table 30 is updated, and the data distribution unit 19 moves the data to optimize the selected data. Move according to the arrangement pattern. Thereafter, in step S735, the spare image manager determines whether the current arrangement and the optimum arrangement are equal. If they are not equal, the process returns to step S732. If they are equal, the forcible movement flag is set to false in step S736 to force movement. The state is released and the process is terminated.

Embodiment 6 FIG.
As in the first embodiment, an example will be described in which an index is shown to the user when data is restored in the spare data management in which the VM image is essential data in an environment where the VM is used as a process in the server.

The processing flow in this example is shown in FIG. The flow is when a presentation of an index for the construction of a specific VM is issued by the user.
In FIG.
size (D) = total data size of data group D,
ready (si) = data group stored in the server si,
src (A) = data group necessary to construct VM A,
It is. In this example, for all servers operating as the spare image manager, first, in step S801, the rate in which data necessary for building a specific VM is stored is calculated as a rate. .

Next, in step S802, when starting a specific VM on the server, the extent of resource excess or deficiency is calculated as res. Finally, the rate and res of all servers are returned to the user and the process is terminated. From the viewpoint of the user, rate indicates the transfer amount load at the time of construction, and res indicates the remaining capacity of the server after construction.

Embodiment 7 FIG.
In the seventh embodiment, the hardware configuration of the active server S1, which is a computer, will be described. FIG. 38 is a diagram illustrating an example of the appearance of the active server S1 that is a computer. FIG. 39 is a diagram illustrating an example of hardware resources of the active server S1. Since the other active servers are the same as the active server S1, the active server S1 will be described.

In FIG. 38 showing the appearance, the active server S1 includes a system unit 830, a display device 813 having a CRT (Cathode / Ray / Tube) or LCD (liquid crystal) display screen, and a keyboard 814 (Key / Board: K / B). And hardware resources such as a mouse 815, FDD 817 (Flexible Disk Drive), compact disk device 818 (CDD: Compact Disk Drive), and printer device 819, which are connected by cables and signal lines. The system unit 830 is connected to another operational server S via a network.

In FIG. 39 showing hardware resources, the active server S1 includes a CPU 810 (Central Processing Unit) for executing a program. The CPU 810 includes a ROM (Read Only Memory) 811, a RAM (Random Access Memory) 812, a display device 813, a keyboard 814, a mouse 815, a communication board 816, an FDD 817, a CDD 818, a printer device 819, and a magnetic disk device 820 via a bus 825. And control these hardware devices. Instead of the magnetic disk device 820, a storage device such as an optical disk device or a flash memory may be used.

The RAM 812 is an example of a volatile memory. Storage media such as the ROM 811, the FDD 817, the CDD 818, and the magnetic disk device 820 are examples of nonvolatile memories. These are examples of a storage device or a storage unit, a storage unit, and a buffer. The communication board 816, the keyboard 814, the FDD 817, and the like are examples of an input unit and an input device. The communication board 816, the display device 813, the printer device 819, and the like are examples of an output unit and an output device.

The communication board 816 is connected to a network (such as a LAN). The communication board 816 may be connected not only to the LAN but also to a WAN (wide area network) such as the Internet or ISDN.

The magnetic disk device 820 stores an operating system 821 (OS), a window system 822, a program group 823, and a file group 824. The programs in the program group 823 are executed by the CPU 810, the operating system 821, and the window system 822.

The program group 823 stores programs that execute the functions described as “˜unit” or “˜device” in the description of the first to sixth embodiments. The program is read and executed by the CPU 810.

The file group 824 includes, as described in the above embodiment, “determination result”, “calculation result”, “extraction result”, “generation result”, and “processing result”. The described information, data, signal values, variable values, parameters, and the like are stored as items of “˜file” and “˜database”. The “˜file” and “˜database” are stored in a recording medium such as a disk or a memory. Information, data, signal values, variable values, and parameters stored in a storage medium such as a disk or memory are read out to the main memory or cache memory by the CPU 810 via a read / write circuit, and extracted, searched, referenced, compared, and calculated. Used for CPU operations such as calculation, processing, output, printing, and display. Information, data, signal values, variable values, and parameters are temporarily stored in the main memory, cache memory, and buffer memory during the CPU operations of extraction, search, reference, comparison, operation, calculation, processing, output, printing, and display. Is remembered.

In the description of the embodiment described above, data and signal values are stored in the memory of the RAM 812, the flexible disk of the FDD 817, the compact disk of the CDD 818, the magnetic disk of the magnetic disk device 820, other optical disks, mini disks, DVDs (Digital). -It records on recording media, such as Versatile and Disk. Data and signals are transmitted on-line via the bus 825, signal lines, cables, and other transmission media.

In the above description of the embodiment, what has been described as “to part” and “to instrument” may be “to means”, “to circuit”, and “to device”. It may be “step”, “˜procedure”, “˜processing”. That is, what has been described as “˜unit” and “˜device” may be realized by firmware stored in the ROM 811. Alternatively, it may be implemented only by software, only hardware such as elements, devices, substrates, wirings, etc., or a combination of software and hardware, and further a combination of firmware. Firmware and software are stored as programs in a recording medium such as a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, and a DVD. The program is read by the CPU 810 and executed by the CPU 810. In other words, the program causes the computer to function as the “˜unit” and “˜device” described above. Alternatively, it causes a computer to execute the procedures and methods of “to part” and “to instrument” described above.

In the above embodiment, the spare image management system and the active server have been described. However, it is also possible to grasp the operation of the active server as an essential data management program that causes the computer to execute. Alternatively, it can be grasped as a computer-readable recording medium in which the essential data management program is recorded. Furthermore, it is possible to grasp the operation of the active server S as a communication method.

In the above embodiment, the case of the VM in the active server S has been described. However, this is an example, and all processes for storing spare data are targeted.

In the above embodiment, the items (1) to (14) have been described for the following major items A to E.

<A. Entire system>
The following (1) to (3) relate to the first, second and third embodiments.
(1) Necessary to take over processing in an environment where the processing (continuous processing) becomes difficult due to some failure, and when it becomes difficult to continue processing on other active servers The technology related to the transfer of data used by the process (hereinafter referred to as essential data) has been described. This technology relates to a spare data manager that takes over essential data even when a storage device (hereinafter referred to as a disk) is used for each server, instead of a shared storage device such as a SAN. The spare data manager appropriately determines the server on which the essential data is to be arranged, reduces the transfer of the essential data at the time of takeover, and makes it possible to take over in a short time. Unlike normal backup devices, the spare data manager operates on the active server, eliminating the need for a server dedicated to takeover of processing, and eliminating the need for a server dedicated to data takeover when passing essential data. . The spare data manager operates in an operational server constituting the system, and the spare data manager exchanges data with each other. Indispensable data is copied to another server by the spare data manager in accordance with an instruction from the user. At this time, the spare data manager uses a server that can actually take over the processing that requires the essential data (a server that has sufficient capacity to perform processing when takeover occurs) as a copy destination. The essential data is pre-aggregated in a server that is preferentially handled and processed. As a result, the transfer amount associated with the collection of essential data at the time of taking over the processing at the time of failure is reduced, and the time required for taking over is reduced. Hereafter, the arranged replica data is referred to as spare data.
(2) In the above (1), the update of the essential data corresponds to the update of the essential data by updating the stored data under the management of the preliminary data manager in accordance with the update of the original essential data.
(3) In (1) above, a difference is taken between essential data, and data generated by the difference is used for saving, thereby reducing the amount of disk used for saving the copy. The data generated by the difference consists of an original that is the source of the difference and a patch that is the difference from the original. At this time, a difference is taken when the essential data is registered, and data generated by the difference is arranged as spare data. Further, when the essential data is required, the essential data is restored on the server where the process is started before the process is started. Unless otherwise stated, the preliminary data includes not only the essential data itself but also the original and the patch.

<B. Judgment at the placement destination judging device
The following (4) relates to the second and third embodiments, and (5) relates to the third embodiment.
(4) The preliminary data manager has a placement destination judgment device for judgment of the placement destination. The placement destination determination unit determines the server on which the spare data is to be placed based on the information on the spare data to be placed, the information on the processing related to the essential data, and the information on the server, and outputs the index to others. The information includes not only the data size but also server resources (hereinafter referred to as resources) such as CPU performance required for processing, memory capacity, required devices, and I / O throughput performance. The spare data manager manages spare data based on the output result. The determination of the placement destination in the placement destination judging device consists of an indispensable condition and an additional condition. The essential conditions must satisfy the decision, and the additional conditions are further refined after the essential conditions. Additional conditions can be omitted. Moreover, when using an additional condition, it carries out combining at least 1 and multiple. It should be noted that although there is selection, it is not determined that the choice is arranged / not arranged, but it is assumed that a weight of 0 to 1 is given to the server.
<Requirement> (3 or 4 should satisfy either)
1. A server on which processing related to a spare image that is a placement target is not currently operating is selected as a placement destination. For example, when the spare image is a patch and there are a plurality of essential data that require the patch, a server that does not operate all of the plurality of processes that use the essential data is a candidate. If the corresponding server does not exist, the spare data is duplicated and distributed to the two servers. Each storage location is determined according to other criteria.
2. If the spare data is allocated, a server that can secure a resource sufficient to start all the processes related to the spare data allocated to the server is selected.
3. Temporarily, after the spare data is placed, the processing related to the spare data stored in the placement destination server in the processing group running on a certain server is targeted, and all the processing can be started. Select the server that has the resource. As a result, resources that can continue processing when a certain server stops are secured in all servers.
4). For example, if a server is stopped for the purpose of placing spare data, the data placement destination server has enough resources to start at least one of the processes running in that server. In addition, a server that can start the remaining processing on another server is selected.
<Additional conditions>
1. Similarly, for a process related to spare data to be arranged, a server having more related data is selected. In determining whether there is a lot of data, the target data group consisting of the preliminary data to be arranged is targeted, and the arrangement target server has all the necessary data for the total size of all the data required to construct the required data. The amount of data included in the data is used as an index.
2. A process instructed to operate on the same server as a process a is defined as a process closely related to the process a. When processing related to the processing related to the spare data to be arranged is included in the processing related to data already arranged in the server, a server having a large number of the processing is selected.
3. A process instructed to operate on a server different from a process a is a process having a weak relationship with the process a. When processing related to the processing related to the spare data to be arranged is included in the processing related to data already arranged in the server, a server having a small number of processing is selected.
4). Select the server with the least resources.
5. When the user inputs which server the process is to be activated (such as the activation rate of the process for each server), the server is selected based on that value.
(5) In 4 above, the placement destination determiner has a function of dividing data. The timing at which the division is performed is when the determination destination is determined, and there are the following two.
1. When there is no server that satisfies the essential condition (4), if the server that satisfies the condition is generated by dividing the spare data, the data is divided and stored. The divided data is equally stored in another server. The determination of the storage destination server is made by making the following changes to the server determination in the determination methods 2 and 3. Rather than determining whether or not the processing related to the spare data can be started by one server, whether or not the processing can be started when all the servers that are divided and stored are combined is used as a criterion.
2. In the above (4), when the process activated on the other server is also related to the data arranged in itself, the data is divided into other servers. Divided data is treated as a chunk.

<C. Movement of arranged data (unplanned movement and planned movement)>
The following (6) relates to the fourth embodiment, and the following (7) to (9) relate to the fifth embodiment. (6) When processing is performed on a certain server, the spare data manager needs the spare data in the server if the server does not have enough free disk space to construct the essential data related to the processing. Transfer as much as possible to another server and move to secure free space. At this time, in selecting the spare data to be moved, a server to be arranged is obtained for each data stored in the server by the same method as in the above (4), and the server is currently selected. Select data different from the saved server. Further, the server moves to the server to be stored determined at that time.
(7) When a process is performed on a certain server, the spare data management unit restores the next data among the spare data arranged in the server after the essential data is restored or the server process has a margin. At some point, move to another server. This addresses further obstacles. The data to be transferred is data related to the processing that has been performed. Also, when processing is performed on a certain server, if data that does not satisfy the arrangement method determined in (4) above occurs in the data in that server, the data that does not satisfy that condition is moved to another server To do. The destination server is determined in the same manner as (4) above.
(8) The spare data manager moves the data stored in the server to another server when there is room in the server at midnight or the like.
One of the following two is used as a data destination.
1. For data movement, a server suitable for storage is checked by means similar to (4) above, and if a server other than the server where the data is currently stored corresponds, the server moves to the corresponding server.
2. Simulates that when a server stops due to a failure, processing starts on another server. In the simulation, it is determined how quickly the entire process can be activated when the data is distributed. Data is moved based on this determination result, and the optimal arrangement is changed. In addition, instead of moving the data all at once, it is timely determined whether there is room to perform the movement process, and if there is no room, temporarily stop the movement process and restart at the next timing, Reduce the processing load. In addition, when preliminary data is newly registered or moved after the previous pause, the simulation is performed again. Also in data movement destination determination 2, data movement is performed within a range that satisfies the essential determination of (4) above.
(9) When the data is a chunk in the movement of data, the backup data manager moves the data to a server that does not have data that is the same as the chunk. Even when the chunks are combined and combined as one data, if there is a server that satisfies the above requirement (4), the chunks are combined and stored in that server.

<D. Countermeasures for secondary failures using replicas>
The following (10) to (12) relate to the second and third embodiments.
(10) The spare data manager saves the replica separately in a separate server when saving the spare data. The data that is the basis of the replica is the master. In storing a replica, a storage destination server is determined by the same method as in (4) above. The storage destination is stored in a server different from the master and other replicas that share the same master.
(11) When there is a replica and the required data is restored on a certain server, if the spare data manager does not have enough free space for the restoration, the spare data manager deletes as much data on the server as necessary. To do. When the data to be deleted is a master, a replica by the master is set as a new master. As in the above (6), the data to be deleted is obtained as a server where each data is arranged, and the server is targeted for data different from the server currently stored.
(12) The backup data manager restores the replica from the master or replica to another server immediately after deletion according to (11) above or when there is room in post-deletion processing. The server to which the replica is restored is obtained in the same manner as (4) above.

<E. Presenting ease of continuation of processing>
The following (13) and (14) relate to the sixth embodiment.
(13) The preliminary data manager presents the easiness of continuation in each server as an index for a certain process based on the data amount stored in each server and the resource amount of each server.
(14) The preliminary data manager presents the ease of continuation of each process as an index for a certain server based on the data amount stored in each server and the resource amount of each server.

S-1, S-2, SN active server, 10A, 10B, 10C data table, 20A server table, 30A, 30B, 30C data allocation table, 40A, 40B VM table, 100 spare image manager, 110 data Differentiator, 120 Data placement destination determiner, 121 Operation VM determiner, 122 Fault guarantee determiner, 123 Data relation determiner, 124 Resource criterion determiner, 125 Simple determiner, 130 Data distributor, 140 Data compounder, 200 Storage device, 300 processing execution unit, 400 network, 1000 spare image management system.

Claims

A plurality of computers including a process execution unit that executes a predetermined process, a storage unit that stores information, and an essential data management unit are connected via a network,
Any of the calculators
A storage instruction for instructing storage of essential data that is partial data constituting part of execution use data used for execution of the process by the process execution unit of any of the computers, the essential data Input a storage instruction including the required data specifying information for specifying the stored computer specifying information for specifying the held computer that is the computer that holds the required data as a save instruction input computer,
The essential data management unit of the storage instruction input computer is:
When the save instruction is input, a master for the save instruction is communicated with the essential data management unit of another computer by communicating with the essential data management unit of the other computer via the network. Determine the computer to be, and transmit the storage instruction via the network to the storage master management unit which is an essential data management unit of the computer determined as the master,
The storage master management unit
A storage destination of the essential data specified by the essential data specifying information of the storage instruction is selected from the plurality of computers based on a predetermined selection rule, and the selected selection computer and the selection computer are stored Transmitting data allocation information indicating the correspondence with the essential data to the possessed computer specified by the retained computer specifying information of the storage instruction,
The essential data management unit of the possessed computer is:
When the data arrangement information is received from the storage master management unit, the essential data is transmitted to the selection computer indicated by the data arrangement information via the network,
The selection calculator is
The essential data management system, wherein the essential data transmitted from the possessed computer is stored in the storage unit of the selected computer.
Any of the calculators
A construction instruction for instructing construction of the execution usage data used for execution of the processing by the processing execution section of any one of the computers, execution usage data specifying information for specifying the execution usage data, and the plurality A construction instruction including construction computer specifying information for identifying a construction computer designated as a construction destination of the execution use data from among the computers is input as a construction instruction input computer,
The essential data management unit of the construction instruction input computer is:
Command that the construction computer specified by the construction computer specification information included in the input construction instruction should construct the execution usage data specified by the execution usage data specification information included in the construction instruction Sending construction command information to the network,
The essential data management unit of the construction computer is
When the construction command information is received, the essential data necessary for construction of the execution use data commanded by the construction command information is acquired by communicating with the other computer as the selection computer via the network. 2. The essential data management system according to claim 1, wherein the execution use data is constructed using the acquired essential data.
The storage master management unit
Processing operation information indicating the processing in operation from each computer of the plurality of computers when the storage destination of the essential data specified by the essential data specifying information of the storage instruction is selected from the plurality of computers. The computer according to claim 1 or 2, wherein the computer that does not perform the processing operation related to the essential data to be collected and stored is selected, and the selected computer is selected from the specified computers. The mandatory data management system described in any one.
The storage master management unit
If it is not possible to identify the computer that does not operate the processing related to the essential data to be stored, select at least two computers as the selected computer from the plurality of computers, and select all the selected computers. 4. The data allocation information of the selected computer and the essential data having the selected computer as a storage destination is transmitted to the possessed computer specified by the retained computer specifying information of the storage instruction. The required data management system described.
The storage master management unit
When selecting the storage destination of the essential data specified by the essential data specifying information of the storage instruction from the plurality of computers, the first selection algorithm to the nth selection algorithm (n is 2) with different selection criteria. The selection process of the selection computer is executed while switching the selection algorithm until a selection computer is selected by any of the selection algorithms in the order of the integers above. The required data management system described.
The essential data management unit of the possessed computer is:
The difference data of the execution use data is transmitted as the essential data via the network to the selection computer indicated by the data arrangement information when the data arrangement information is received from the storage master management unit. 1. The essential data management system according to 1.
The essential data management unit of the construction computer is
When receiving the construction command information, as the essential data necessary for construction of the execution utilization data commanded by the construction command information by communicating with the other computer that is the selected computer via the network, 3. The essential data management system according to claim 2, wherein difference data of execution use data is acquired.
In a computer that is connected to a network together with a plurality of other computers and communicates with the other plurality of computers that execute predetermined processing via the network,
A storage instruction for instructing storage of essential data that is partial data constituting a part of execution use data used for execution of processing of any one of the other computers. The storage instruction via the network from the computer to which the storage instruction including the required data specifying information for specifying the required data and the storage computer specifying information for specifying the held computer that is the computer that holds the required data is input. And the storage destination of the required data specified by the required data specifying information of the received storage instruction is selected from among itself and a plurality of other computers based on a predetermined selection rule. Data allocation information indicating a correspondence relationship between the selected computer and the essential data with the selected computer as a storage destination is specified by the possessed computer specifying information of the storage instruction. Computer and comprising the required data management unit to be transmitted to the held computer that.
A computer connected to a network together with a plurality of other computers and communicating with the plurality of other computers executing predetermined processing via the network,
A storage instruction for instructing storage of essential data that is partial data constituting a part of execution use data used for execution of processing of any one of the other computers. The storage instruction via the network from the computer to which the storage instruction including the required data specifying information for specifying the required data and the storage computer specifying information for specifying the held computer that is the computer that holds the required data is input. And the storage destination of the required data specified by the required data specifying information of the received storage instruction is selected from among itself and a plurality of other computers based on a predetermined selection rule. Data allocation information indicating a correspondence relationship between the selected computer and the essential data with the selected computer as a storage destination is specified by the possessed computer specifying information of the storage instruction. The holdings required data management program for causing to function as an essential data management unit to be transmitted to the computer that.
A computer-readable recording medium on which the essential data management program according to claim 8 is recorded.
In a communication method performed by a computer that is connected to a network together with a plurality of other computers and communicates with the other plurality of computers that execute predetermined processing via the network,
Mandatory data management department
A storage instruction for instructing storage of essential data that is partial data constituting a part of execution use data used for execution of processing of any one of the other computers. The storage instruction via the network from the computer to which the storage instruction including the required data specifying information for specifying the required data and the storage computer specifying information for specifying the held computer that is the computer that holds the required data is input. And the storage destination of the required data specified by the required data specifying information of the received storage instruction is selected from among itself and a plurality of other computers based on a predetermined selection rule. Data allocation information indicating a correspondence relationship between the selected computer and the essential data with the selected computer as a storage destination is specified by the possessed computer specifying information of the storage instruction. Communication method and transmits the that the held computer.