WO2006013992A1 - ネットワークシステム、管理コンピュータ、クラスタ管理方法およびコンピュータプログラム - Google Patents
ネットワークシステム、管理コンピュータ、クラスタ管理方法およびコンピュータプログラム Download PDFInfo
- Publication number
- WO2006013992A1 WO2006013992A1 PCT/JP2005/014467 JP2005014467W WO2006013992A1 WO 2006013992 A1 WO2006013992 A1 WO 2006013992A1 JP 2005014467 W JP2005014467 W JP 2005014467W WO 2006013992 A1 WO2006013992 A1 WO 2006013992A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- class
- computer
- computers
- evening
- management
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/505—Clust
Definitions
- the present invention relates to an efficient distributed computing technique performed by a plurality of computers connected to a computer network in a broadband environment.
- the present invention relates to a network system capable of distributed computing, its components, and a class management method in a network.
- the server identifies the size of the job load and the surplus processing capacity (computation resources) of each network connected to the network when performing distributed processing.
- the server determines the surplus processing capacity according to the load.
- the assigned computers are assigned sequentially, and job execution results are received from the assigned computers.
- the present invention provides a distributed processing mechanism that can solve such conventional problems.
- the main challenge is to provide Disclosure of the invention
- the present invention provides a network system capable of efficiently distributing and executing information processing, which is expected to change from moment to moment, by a plurality of computers, in various sizes, types, and numbers of jobs to be executed.
- the above-mentioned problems are solved by the components and class management method.
- Computer here refers to a device that includes a processor that is operated by a computer program, but a device, processor port, processor chip, or processor itself that does not take the form of a device is also included in the concept of “computer”. It is. For example, individual processors, such as multiprocessor systems, can also be “computers”.
- the network system of the first configuration provided by the present invention is a network system in which a plurality of computers that can be clustered with other combi- nations can freely join and leave.
- this network system there is a first table in which information on whether or not each computer is capable of class setting is recorded, and a cluster that is already formed by one or more computers. And a second table for recording additional ease information indicating how easily other computers can be added.
- any one of the computers forms a class including the self and other computers whose clusterability information in the first table indicates a clusterable state, and all computers included in the formed cluster
- the clustering means for updating the clustering availability information for information to information indicating a state where class setting is impossible and further recording the additional ease information about the formed class setting in the second table.
- the computer forming the cluster records in the second table whether or not to add the candidate computer evening to the class evening when there is a candidate computer to be added to the class evening. It has a cluster growth means that is determined based on the added ease information.
- one of the computers autonomously forms a class based on the clustering availability information of each computer recorded in the first table, and adds the class evening.
- the number of computers included in the class evening may be the default number determined at the time the class evening is formed, depending on the size of the job that was actually submitted or expected to be submitted in the near future. It may be a number determined by flexure. In any case, a computer whose clusterability information is in a clusterable state is selected and a cluster is formed.
- the additional ease information is information for the computer that formed the class to autonomously determine whether it is appropriate to add the candidate computer, and various types of information can be used. Add only the computers that have applied for the addition first, or add only computers with identification information that is generated by random numbers. The conditions such as' can also be adopted as additional ease information. wear. However, from the viewpoint of simplifying the process of deciding whether or not to add, the ease of addition and when the scale is quantified numerically, and the numerical value can be compared with this numerical value. It is desirable to make it easier to add candidate computers to the class evening.
- the numerical value in this case is a random response only once to five inquiries about whether or not it can be added at random (the numerical value in this case is 15: 20%)
- Candidate computers can be added only during the time period of 10% of the system operating time of the day in total (in this case, 2 4 (hours) X 6 0 (minutes) X 6 0 (Seconds) X 0.1 X (probability value)).
- the numerical value may be held at a constant value regardless of the addition of the candidate computer, or may be variable after being recorded in the second table.
- the candidate computer was added in the class evening. By moving the class, it is easy to gather candidate candidates, and it is easy to gather other candidate computers, and it is easy to easily form clusters of sizes that meet actual demand. .
- the computer forming the cluster deletes the cluster when the job execution by the cluster is completed, and is recorded in the second table.
- Class event extinction means for returning the clusterability information recorded in the first table to the state before the class event formation for all the computer events belonging to the extinction cluster and the extinction information. It has further.
- the cluster grows based on the additional ease information described above. However, the cluster disappears when the job execution is completed. Compared to the case where a fixed class is prepared. Effective use of computers will become possible.
- a plurality of management computers each of which owns one or a plurality of computers capable of classifying with other combination evenings, freely join and leave. It is a network system that can do this.
- Managing one or more computers as computers under their own control means controlling the operation of a plurality of computers connected to the computer and monitoring their operating states.
- a first table in which information about whether or not each computer can be clustered is recorded, and a class already formed by one or more computers.
- a second table that records additional ease information indicating how easy it is to add another evening.
- at least one of the management computers is a computer under its own control.
- a cluster including a computer in which the clustering availability information of the first table indicates that the class can be changed is formed.
- the management computer evening that forms the class evening is recorded in the second table as to whether or not the candidate computer evening is added to the class evening when there is a candidate computer to be added.
- Class evening growth means is determined based on the additional ease information.
- one of the management computers autonomously forms a cluster based on the information on whether or not the combination is recorded in the first table.
- the management computer that created the class evening records them in the second table. Decide whether to add the candidate computer to the class based on the added ease of use information.Classification availability information, handling of additional ease information, class sunset, and cluster growth criteria are:
- the network system of the first configuration is the same as the network system of the first configuration.
- class event erasure unit that restores the recorded information to the state prior to cluster formation.
- the class event grows based on the additional ease information described above until the job is executed. Since it disappears when the class ends, the computer can be used more effectively than if the class evening was prepared in a fixed manner.
- the second table may exist at an arbitrary place, but normally, it will be held by any management computer that will generate the second table. .
- the management computer has the second table of the maximum number of computers under its control. Become.
- the second table is a master table that is generated for the first computer when the management computer that owns it forms a first class evening that includes the first computer that is under its control, and the management that owns it.
- a computer supervises and controls the operation of the second combination evening that is added to the second class evening created by another management computer evening, the slave table generated for that second computer is created. At least one of the bulls. Additional ease information is recorded in the master table.
- the management computer that owns the master table behaves as a master management computer that takes the lead in information processing related to the formation of the first cluster, the change in the number of computers in the first class evening, and the disappearance of the first class evening.
- a management computer that owns a slave management computer for the second class evening is a master table that is generated for the first computer when the management computer that owns it forms a first class evening that includes the first computer that is under its control, and the management that owns it.
- the master management computer searches for candidate computers to be added to the first class by inquiring whether there is a computer that can be clustered from any of the management computers. It is assumed to have a search means for Also, one of the management computers determines whether or not to add a candidate computer belonging to the management computer to the first cluster based on the additional ease information about the first cluster formed by the mass management computer. Configure to determine.
- the network system of the third configuration provided by the present invention is a network system in which a plurality of computers capable of classifying with other computer evenings can freely join and leave; Including a table that lists, for each cluster formed by one or a plurality of computers, identification information of other clusters to which a computer belonging to the class is associated; the computer forming the class Based on the identification information restored in the table, whether or not to add a candidate computer to another cluster related to its own class when there is a candidate computer to be added.
- This is a network system having a class evening growth means for determining.
- “Relevant” means, for example, a state in which it is possible to communicate with each other and to perform collaborative processing. Say. In a network system with such a configuration, it is possible to perform clustering according to the power distribution regardless of the additional ease information.
- the present invention also solves the above problem by providing a management computer for configuring the network system of the second configuration, for example.
- Each of these management computers is a management computer that owns one or more computers that can be clustered with other computers; they can freely join and leave with other management computers with similar functions.
- the computer and other management computers under its own umbrella Among the computers, the cluster setting availability information in the first table forms a cluster including computers indicating a class setting possible state, and the computers for all the computers included in the formed class setting.
- a class setting unit for updating the class setting information to information indicating a non-clusterable state, and recording the additional ease information for the class setting in the second table;
- Cluster growth means for determining whether or not to add the candidate combination to the class evening when there is a computer based on the ease of addition information recorded in the second table. .
- the class evening is deleted, and the additional ease information recorded in the second table and the class belonging to the deleted class evening are all stored. It is also possible to further include a class deletion means for returning the clustering availability information recorded in the first table to the state before cluster formation for all computers.
- the present invention also solves the above problems by a cluster management method executed by a plurality of computers included in the network systems of the first configuration and the third configuration described above, for example.
- the first class management method is a class management method in a network system in which a plurality of computers that can be clustered with other computers can freely join and leave; A step of recording in the first table the class setting availability information indicating whether or not the own state is a state where class setting is possible; A class evening is formed that includes another computer whose clustering availability information indicates a clusterable state, and the class evening availability information for all the combination evenings included in the created class evening Easier to add to the information indicating the status of non-configurable, and how easily the cluster can add other computers The step of recording information in the second table; and the ease of addition recorded in the second table as to whether the candidate computer is added to the cluster when there is a candidate computer to be added.
- a class decision management method comprising: determining based on information.
- the method may further include the step of returning the recorded information of the first table for the computer to the state before the class setting.
- the second cluster management method is a class management method in a network system in which a plurality of computers each capable of classifying with other computers can freely join and leave;
- a computer that forms a cluster with a plurality of computers lists the identification information of other classes that are associated with computers belonging to the class in a predetermined table; Based on the identification information recorded in the table, whether or not to add the candidate computer to another cluster associated with the class evening when there is a candidate computer evening to be added.
- a cluster management method comprising:
- the present invention also solves the above problems by providing a computer program for giving a computer a predetermined function.
- Each of the first computer programs is read and executed by any computer in the network system in which a plurality of computers that can be clustered with other computers can freely join and leave.
- a first table that records whether or not each computer can be clustered, and a cluster that has already been formed by one or more computers.
- Table management means that allows access to the second table to record additional ease information that indicates how easy it is to add a computer; clusterability information on the first table can be clustered Forms a cluster that includes self and other computers representing the state of At the same time, the class setting availability information in the first table for all combinations included in the formed class evening is updated to information indicating a non-clusterable state, and the additional ease for the cluster is further updated.
- Class evening forming means for recording information in the second table; whether there is a candidate computer to be added as a candidate computer; whether the candidate computer is added to the cluster is recorded in the second table Class determined based on additional ease information A computer program for functioning as an evening growth means.
- the second computer program is read and executed by a management computer that owns one or more computers that can be clustered with other computers, thereby executing the management computer;
- Network connection means for connecting to a computer network that can freely join and leave with a functioning management computer;
- class setting that indicates whether the state of each computer can be set to class A first table recording availability information, and a second table for recording additional ease information indicating how easily a class already formed by one or more computers can add another computer.
- Table management means to allow access to the table; under its own umbrella A class evening including a computer in which the class evening availability information in the first table of the computers under the control of the computer and the other management computer indicates a class evening possible state is formed and formed.
- Class Classifying the clustering availability information for all computers included in the evening A class evening forming means for updating the information indicating that the sunset is impossible, and recording the additional ease information for the class evening in the second table; when there is a candidate computer to be added
- Each of the third computer programs is read and executed by one of the computers in the network system in which a plurality of computers capable of classifying with other computers can freely join and leave.
- Cluster management means for listing, in a predetermined table, identification information of other classes that are associated with computers belonging to the cluster when the class is already formed by one or more computers. When there is a candidate computer to be added, whether to add the candidate computer to another class associated with the own cluster is determined based on the identification information listed in the table.
- Class evening growth means; Computer program.
- These computer programs are recorded on a portable recording medium and distributed through the market, or downloaded from a program server or the like through a computer network accessible by a computer or a management computer.
- classes of various sizes are formed in any one of a plurality of computers in a network system, and grow based on recorded information in a predetermined table. You can easily get a class evening. As a result, the size, type, and number of jobs to be executed vary, and even information processing with uncertainty that one of them is expected to change from moment to moment can be executed efficiently. The unique effect is that distributed computing that can do this is realized.
- FIG. 1 is an overall view of a network system to which the present invention is applied.
- FIG. 2 is a diagram showing a general architecture of the management computer according to the present embodiment.
- Figure 3 is an illustration of a large-scale information processing integration.
- FIG. 4 is a functional configuration diagram of the management computer according to the present embodiment.
- Figure 5 is an actual measurement plot with the number of links on the horizontal axis and the number of nodes with links on the vertical axis.
- (A) is an example of a random connection
- (b) is a selective connection. This is an example.
- Figure 6 (a) shows the contents of the master table
- Figure 6 (b) shows an example of the contents of the slave table.
- FIG. 7 is a schematic procedure diagram of distributed computing according to this embodiment.
- FIG. 8 is an explanatory diagram of the cluster formation and growth procedure by the management computer.
- FIG. 9 is a diagram for explaining the procedure for cluster formation and growth by the management computer.
- FIG. 10 is an explanatory diagram of job submission and execution procedures by the management computer.
- FIG. 11 is an explanatory diagram of the processing procedure when the management computer extinguishes the cluster.
- 12 (a) to 12 (d) are explanatory diagrams showing the process of formation, growth, and disappearance of a class evening according to this embodiment.
- Fig. 13 (a) is a schematic diagram of a network system realized by a multiprocessor
- Fig. 13 (b) is a schematic diagram of a network system realized by a multi-core 'processor alone.
- FIG. 14 is a diagram showing an example of the contents of the master table used in the duplication mode.
- FIG. 15 is a diagram showing an example of operation in the duplication mode.
- Figures 16 (a) and 16 (b) are diagrams showing other examples of operation in the duplication mode.
- FIG. 17 is a diagram showing another example of operation in the duplex mode.
- FIG. 1 is an overall view of a network system 10 1 to which the present invention is applied.
- This network system 10 1 includes a computer network 10 4.
- Examples of computer 'networks 104 include local' area 'networks (LAN), global networks such as the Internet, or other computer networks.
- the computer network 10 4 can be connected to management computers 10 6 having one or more computers under their own control, each at an arbitrary timing. Each management computer 10 6 can perform two-way communication with another management computer 1 0 6 and another computer 10 8 under the control of the other management computer 1 0 6. . Each computer 10 8 can be part of any management computer 10 6 at any time. In other words, these computers 1 0 6 and 1 0 8 can participate in the network system 1 0 1 in such a way that they can freely leave at any time.
- Computer in this embodiment refers to an apparatus including a processor that operates by a computer program, but a device, a processor board, or a set of these that does not take the form of an apparatus is also included in the concept of “computer”. . Also, “belonging to one or a plurality of computers 10 8” means controlling the operation of a plurality of computers 10 8 connected to itself and monitoring their operation state.
- Examples of the management computer 106 include a computer having a server function, a game console with a communication function, a computing device, and a processor board.
- Examples of the computers 10 8 under the management computer 10 6 include personal computers, game consoles with communication functions, and other wired or wireless computers, computing devices, processor boards, and the like.
- a plurality of computers 1 0 8 are directly connected around a management computer 1 0 6 connected to a computer network 1 0 4.
- Evening connection type multiple computers 1 0 8 are connected via a local network to which management computers 1 0 6 are connected, and multiple computers 1 0 8 are connected via a computer network 1 0 4 Examples include a wide area network type connected to the management computer 10 6.
- local networks there are various types of local networks in the local network type, and there are some networks connected to the computer network 10 4 and the director, while the computer network 10 4 is connected to the network 10 10 through the management computer 10 6. Some networks are connected to each other.
- each management computer 10 6 includes a storage device 1 0 61, a communication device 1 0 6 3, a semiconductor memory 1 0 6 5, etc. connected to each other through a bus B 11. Including processor 1 0 6 7.
- the management computer 10 6 having such hardware resources allows the processor 1 0 6 7 to read and execute a computer program loaded into the semiconductor memory 1 0 6 5 through a recording medium such as a CD-ROM.
- the storage device 1 0 6 1 stores a plurality of management tables 2 0 6 described later, a table management unit 2 1 2, a class management unit 2 1 6, The functions of the communication control unit 2 3 6 and job execution unit 2 5 6 are built in the computer.
- the table management unit 2 1 2 accesses the management table 2 06 and updates the recorded contents of the management table 2 0 6.
- Class evening management department 2 1 6 is the information related to the formation, growth and disappearance of class evening to which the computer 1 0 8 under its control and the computer 1 0 8 under the other management computer 1 0 6 belong. Process. That is, the cluster management unit 2 16 functions as cluster forming means, class evening growth means, and cluster extinguishing means according to the change in the class evening state.
- the communication control unit 2 3 6 enables communication between its own computer 10 8 and other management computers 1 0 6 through the communication device 1 0 6 3.
- the job execution unit 2 5 6 submits the job to the class including the computer 1 0 8 under its own control and executes it.
- the job execution unit 2 5 6 may be provided with a function for calculating the job size, processing time, etc. in order to easily identify the optimal cluster size, that is, the number of computers.
- the computer 10 8 has the same storage device, communication device, semiconductor memory and processor as the management computer 10 6. In this embodiment, a status table described later is stored in the storage device.
- various functions related to class management and job execution are built in the computer body by the processor reading and executing the computer program loaded into the semiconductor memory through a recording medium such as a CD-ROM.
- the other combinations 10 8 and the instruction set architecture SA) are all considered to be the same or the same, and the required processing can be executed according to the same instruction set.
- the number of computers 10 8 under each management computer 10 6 is arbitrary. In a certain management computer 10 6, the number of computers 10 8 is allocated depending on the processing capability necessary to execute jobs given by various applications.
- Each management computer 106 forms a class class including one or more of its own computers 10 08 or one of the other management computers 10 8 of the management computer 10 6. Enable job execution in units. Differences in performance and restrictions in forming at least one class evening between the computer under its own umbrella 10 10 and the other computer 10 0 6 Etc. does not exist. By handling in this way, it is not so important which computer 10 8 of which management computer 10 6 executes the job. It is only necessary to specify the recipient of the job execution result as the management computer 10 6 that requested the job, the computer 10 8 under its management, or any computer 10 8 that executes the subsequent job. Therefore, individual jobs can be easily executed in a distributed manner among the computers 10 8 belonging to a plurality of management computers 10 6 connected to the computer network 10 4.
- Each computer 10 8 in the large-scale information processing integration WO is physically managed by a management computer 10 6 to which the computer belongs, and can operate as a single computer or belong to the management computer 10 6.
- the class is set up together with some other computer 1 0 8 and works together.
- class setting in such a form one job can be distributedly executed by a plurality of combinations 10 8 belonging to the same class setting.
- the present invention provides an efficient distributed computing mechanism using the architecture of the network system 101 described above.
- each of a plurality of computers 10 8 constituting a network such as the large-scale information processing integrated body WO shown in FIG. It will be possible to repeat “growth” in which the number of nodes increases with the passage of time and “selective connection” as a method of connecting links between nodes.
- “Selective connection” refers to a node that already has more links as a criterion for selecting a link destination node when a newly joined node establishes a link. It is easier to connect to the card.
- the two characteristics of growth and selective connectivity make the network grow and scale-free. By repeating such growth and selective connection, the distribution of the number of links in a node becomes a power distribution. For example, Albert Barabasi, Reka Albert, Hawoong "Mean-field theory for scale-free random networks" Shown in detail.
- Figure 5 is a graph plotting measured values with the horizontal axis representing the number of links of nodes when nodes are randomly connected and the case of selective connections, and the vertical axis representing the number of nodes with the number of links. is there.
- Figure 5 (a) shows an example of connecting nodes randomly. In this example, a typical node with a typical number of links appears.
- Figure 5 (b) shows an example of selective connection. A typical node does not appear, and the number of links extends over a wide range. In other words, it is called scale free in the sense that a typical scale does not appear.
- the distribution shown in Fig. 5 (b) is expressed as follows. When written in a log-log graph, the distribution is very wide and forms a straight line.
- the present invention applies the above two features, interprets each node as a class, and interprets a link as two combinations 108 included in a cluster corresponding to nodes connected to each other.
- clustering is performed in a form that can have any distribution, of course, the power distribution shown in Fig. 5 (b), thereby enabling efficient distributed computing.
- the number of links is the number of computers 108 included in the class evening, that is, the class evening size, or the class evening computing capacity. If the horizontal axis plots the number of computers 108 in the cluster and the vertical axis plots the number of classes or the frequency of clustering, it should be a distribution similar to the power distribution shown in Figure 5 (b). .
- each computer 10 8 is currently managed by a computer class status (clustering availability information) indicating whether or not its own state is in a state where class setting is possible.
- the computer 'class' status is recorded on the status table, and is updated by the computer 10 8 or the management computer 10 6 managing it as a subsidiary, following changes in the state of the computer 10 8.
- the status table is sufficient if it exists in an arbitrary memory area accessible by the computer 10 8 or the management computer 10 6, but may be provided in the storage device of the computer 10 8.
- C lus tered indicates that the computer has already been clustered and is currently waiting for job execution.
- “Run” indicates that the computer is executing a job.
- “F ree” indicates that the computer is not operating.
- “c lus tered” or “run” is selected, the computer 10 8 cannot classify. Therefore, “c lus tered” and “run” are information indicating a non-clusterable state.
- “free” is information indicating that the class can be changed.
- the management computer 10 6 One of the roles of the management computer 10 6 is to manage the computer 10 8 under its own control as described above.
- the management table 206 is used for that purpose.
- the management table 2 06 is stored in, for example, the storage device 1 0 6 1 included in the management computer 1 0 6.
- This management computer 10 6 has as many management tables 2 0 6 as the number of computers 1 0 8 at its maximum.
- the management table 2 0 6 categorized into one type is used for the management computer 1 0 6 to record information on the state of the class when the management computer 1 0 6 first forms a cluster including its own computer 1 0 8. It is a “mass evening table”. The other is a “slave table” for recording information about the state of the class when there is a computer 10 8 added to the cluster formed by other management computers 10 6.
- Management computer 1 0 6 is a class that is managed by the master table. When monitoring and controlling the computers 10 8 included in, it behaves as a “master management interview”. On the other hand, when a slave table 10 8 belonging to a cluster managed by another management computer 10 6 is monitored and controlled by the slave table, it behaves as a “slave management computer”. In other words, even though it is one management computer 10 6, it will behave as a management computer of the number of computers 1 0 8 at its maximum.
- the master table is generated by the management composition that acts as the master management computer, and the slave table is generated by the management computer that acts as the slave management computer.
- the existence of a slave table means that a corresponding master table exists in some management computer 106.
- Fig. 6 (a) shows an example of the contents of the cell table 2 1 6 1
- Fig. 6 (b) shows an example of the contents of the slave table 2 1 6 2.
- the master table 2 1 6 1 contains the cluster ID (ID is the identification information; the same shall apply hereinafter) 2 3 6 1, cluster size 2 3 6 2, computer list 2 3 6 3, class setting 'status 2 3 6 4, Maximum number of computers 2 3 6 5, Minimum number of computers
- Class evening ID 2 3 6 1 is a unique ID given to the class evening. As long as the class class continues to exist, it should be unique, so for example, the ID of the mass management computer and the ID of the combination class that will be included in the class class at the beginning. And the class evening ID.
- Class evening size 2 3 6 2 is the total number of computers 10 8 newly added or added to the class evening. This can be identified by the master management computer by measuring the number of computers in the cluster.
- the computer list 2 3 6 3 is a list of identification information of the combination evening 10 8 that has been included in the class evening.
- Class evening status 2 3 6 4 is information indicating the current state of class evening.
- three types of states “idle”, “run”, and “wai t” are used as a cluster state.
- I dl e” is a state in which the cluster is not executing a job. When a class evening is in this state, a computer 10 8 can be added to that class evening.
- “Run” is a state in which the class is executing a job.
- “Wai t” is the state in which the number of retained computers 10 8 has reached the maximum number of computers. In this state, computers 10 8 cannot be added to the class evening.
- the maximum number of computers 2 3 6 5 is the maximum number of computers that the class can hold, and is defined by the user or defined as a constant held by the system. In the example shown in the figure, it is shown that the cluster can hold up to 200 computers.
- the minimum number of computers 2 3 6 6 is the minimum number of computers that must be held as a class, and is defined by the user or a constant held by the network system 1 0 1. In the example shown in the figure, it is shown that three computers 1 0 8 make a class evening.
- the total number of computers 2 3 6 7 is the total number of computers 108 that may be included in the class, and can be obtained from the expected value based on statistics. In the example shown, it is shown that 48 computers 10 8 may be included.
- connection ratio 2 3 6 8 is information that is an example of additional ease information.
- the connection ratio is a value that determines the ease of addition by a probability value. The higher this number is, the easier it will be to add computers 108 in the class, and the easier the cluster will grow.
- the connection ratio 2 3 6 8 can be a constant value or a variable value. A constant value can be replaced with a variable value afterwards. And vice versa. This may be set automatically as the default value when creating a class evening, or an application program that executes a job in a class evening as one of the parameters of the master table 2 1 6 A connection ratio of 1 may be set to 2 3 6 8.
- the horizontal axis represents the number of computers 10 8 in the cluster
- the vertical axis represents the distribution when the number of classes or the frequency of clustering is plotted. It depends largely on how you do it.
- Figure 1 the horizontal axis represents the number of computers 10 8 in the cluster
- Figure 2 the vertical axis represents the distribution when the number of classes or the frequency of clustering is plotted. It depends largely on how you do it.
- connection ratio When trying to obtain a normal distribution such as (a), the connection ratio is a constant value. As a result, the above distribution becomes a normal distribution centered on the representative value.
- the connection ratio may change as the computer evening is added and the class evening size increases. For example, the connection ratio may be increased each time one or several computers are added. In this way, the power distribution as shown in Fig. 5 (b) can be easily obtained.
- Class evening ID 2 4 6 1 is an ID for identifying the class evening (same as class evening ID 2 3 6 1 recorded in mass evening table 2 1 6 1).
- Computer i d 2 4 6 2 is an ID for identifying the computer 1 0 8 included in the class evening.
- Slave table 2 1 6 2 is cluster ID 2 4 6 1 and computer i d 2 4
- the way of holding the above components (values) in the slave table 2 1 6 2 is also an example, and can be increased or decreased as appropriate.
- Step S 2 an example of an operation mode of distributed combining by the network system 100 of this embodiment.
- the job is distributedly executed while repeating the three-stage class change cycle.
- the state change in the first stage is caused by the formation and growth of a new cluster (Step SD o
- the state change in the second stage is caused by the input of a job requested by an application program etc. (Step S 2).
- the three-stage state change is caused by the disappearance of the class setting after job execution (step S 3) .
- the class setting disappears, all computers 10 8 belonging to the cluster up to that point It becomes a candidate computer (node) that can newly belong to another class, and the state change in the first stage in this cycle occurs regardless of whether or not there is a job.
- It becomes a management computer forms a cluster, and grows it in cooperation with the slave management computer.
- the degree of growth depends on the operating status and connection ratio of each computer 10. The higher the connection ratio is, the easier it is to connect to other computers 10 8. In the following, the state of the above three-stage state change will be described in detail.
- FIG. 7 is a procedure explanatory diagram showing the process of formation and growth of a class evening by the master management computer (the class evening management unit 2 1 6 of the management computer 10 6 operating in such a manner).
- the master management computer checks the operating status of its own computers 10 8. Specifically, the status of the computer 'class setting' recorded in the status table held by each computer 1 0 8 is checked (step S 1 0 Do and the computer 1 0 8 that is not operating, that is, the computer ⁇ Class evening ⁇ Several computers with status “f ree” form a class evening with 1 0 8 (step S 1 0 2). For example, it is determined by prior settings, etc. Usually, some computers 1 0 8 of their own computers will be left in the class, and the remaining computers 1 0 8 will be To facilitate cluster growth If possible, make sure that the class is added to another class as much as possible. This state is shown in Fig. 12 (a).
- one computer class C 1 1 is formed by three computers.
- the master management computer generates one master table and sets the components (values) of this master table (step S104). If the management computer is supposed to manage the status table of each computer 108, all computers included in this class will be changed from “free” to “clustered” status. (Step S104).
- the mass management computer determines whether or not there is a candidate computer to be added to the class class that it has formed among other computers 108 under the randomly selected other management contributor 106.
- a search is made by inquiring whether there is any other computer 108 whose cluster status is “free” (step S 105). The search is performed, for example, by randomly selecting several other management computers 106 having a short logical distance starting from the self and sequentially expanding the range. If a candidate computer exists (step S106: Yes), the connection ratio in the master table is examined, the candidate computer is added to the class based on the connection ratio, and information about the candidate computer is stored in the master table. Record (Step S107). Thereafter, the computer / cluster / status of the added candidate computer is updated from “free” to “clustered” (step S 108).
- the management computer 106 that manages the added candidate computer becomes a slave management computer for the candidate computer 108 and generates a slave table.
- Figure 12 (b) shows the state when the candidate computer exists. In the example of Fig. 12 (b), it has been added as a candidate computer with two computers 108, and has grown into a class C 14 that has included 108 computers with five computers.
- the master management computer adds the information of the newly added computer 108 to the mass table.
- step S106 if there is no candidate computer, Step S 1 0 6: No), the process returns to step S 1 0 5.
- FIG. 9 is a procedure explanatory diagram showing the growth process of the class evening by the management computer 10 6 that is not the master management computer.
- the growth of the class evening is realized by the active behavior of the master management computer, but the management computer that is not the master management computer 1 6 You can also access the computer and do it.
- a management computer 10 6 having a non-classified computer 10 8 under its control makes the computer 1 0 8 available by, for example, contacting another management computer 10 6 selected at random. Search for other clusters that can be added (step S 2 0 1). Then, based on the connection ratio of the destination class evening, it is determined whether or not to add the computer 10 8 belonging to the class evening to the class evening (step S 2 0 2). If it is determined not to be added, the process returns to step S 2 0 1 (step S 2 0 3: No).
- the computer 1 0 8 becomes a slave management computer for the connection destination class, so a slave table is generated for the computer 1 0 8 and each component (value) is added to this slave table. ) (Step S 2 0 4). Then, the computer 'class computer' status of the computer 10 8 to be added is updated from “free” to “clus tered” (step S 2 0 5). In addition, the mass management table is updated on the master management computer of the added class (step S 2 0 6). This update may be performed by accessing the master management computer from the slave management computer.
- the state change in the second stage starts with an action by one of the management computers 106 in which a job is submitted by a request from an application or the like.
- the amount of computation and the time required to execute the job such as the The number of pure evening 1 0 8 is a parameter.
- This parameter may be automatically generated by the processor of the management computer 106 based on the size of the submitted job, or may be appropriately given by a user using an application or the like.
- the parameters of the application may be used.
- FIG. 9 is an explanatory diagram of the cluster usage procedure by the management computer 10 6 to which the job is submitted. Based on the above calculation amount, the management computer 10 6 forms a class size of the required size by referring to the class setting list 20 8 or by querying another management computer.
- the master management combi- nation that has been set is searched (step S 3 0 1). If found, the master management computer is requested to execute the job (step S 3 0 2: Yes, S 3 0 3). For example, a job execution request is sent to the master management computer by sending a bucket containing the job, the programs and programs required to execute the job, and a specified address for sending the execution result to itself. Do.
- the master management computer updates the computer 'cluster' status from “idol” or “wait” to "run” for each computer belonging to the class that executes the job. 0 Start job execution by distributed processing.
- Fig. 1 2 (c) shows a state in which the master management computer that manages the class C 1 4 grown as shown in Fig. 1 2 (b) is executing a job.
- the job execution result is sent from the master management computer to the requesting management computer.
- the management computer 10 6 that has received the execution result transmits the execution result to the job submission source (step S 3 0 4: Yes, S 3 0 5).
- step S 3 0 1 to S 3 0 4 are the processing procedures performed by itself. Also, if it is determined in step S 3 0 2 that it is necessary to execute a job across multiple classes, for example, it is determined that image processing and audio processing need to be executed in separate clusters. In this case, the job management computer that manages each cluster is requested to execute the job.
- Step S 4 0 1 When the master management computer finishes the job (step S 4 0 1: Yes), the computer cluster status for all computers 1 0 8 in the class that has executed the job up to that point is “f ree”. (Step SS 4 0 2) o Also, the component (value) in the master table for that class is cleared. At the same time, the components (values) of the slave table are cleared through the slave management combi- cation that has the computer belonging to the class evening. That is, all computers are returned to the state before the class setting (step S 4 0 3). As a result, all computers 10 8 belonging to the class evening are in an unoperated state and ready for class evening, and one computer 10 08 is immediately clustered and other The computer 1 0 8 will be added to another class. Figure 12 (d) shows this state.
- any one of the size, type, and number of jobs to be submitted varies, and a network system with many uncertain elements that is expected to change from moment to moment. Even so, each of the multiple management computers 10 6 behaves as a mass management computer or a slave management computer for each of its affiliated computers 10 8, and the optimal size class. Since evenings are formed and grown, job execution can be assigned to classes of the appropriate size and jobs can be executed effectively. In addition, since it is possible to deal with large-scale processes to very small processes in a uniform manner, it is possible to realize highly flexible distributed combining. In particular, when the process size distribution is a power distribution, the additional ease of use information can be made variable and the class size can also be set to a power distribution, so that the process can be processed most effectively. And increase the use efficiency of computer resources.
- each computer 1 0 8 has a status table for convenience.
- the management table 2 0 6 (master table slave table) has been described on the assumption that the management computer 10 6 possesses, but these tables are stored in each computer 1 0 8 and management computer 1 0 6. It only needs to exist in a memory area that can access them.
- the management computer 10 6 may have a status table of its own computer 10 8, or the server etc. accessible to the table management unit 2 1 2 of the management computer 10 6.
- the management table 206 may be stored collectively.
- whether or not to add candidate reviews under the slave management computer to another class is decided by the slave management computer, but the candidate computer is added to the cluster. Whether or not to do so may be determined by the master management computer.
- the present invention is not limited to the management computer 1 0 6. It is also possible to implement as a network system in which each of a plurality of computers 1 0 8 can directly join and leave.
- the connection ratio should be recorded in the status table of each computer 108, and the status table should be managed by itself. For example, when adding a self to a class created by other computers 1 0 8, the self computer class class status is updated to “c lus tered”.
- any of the computers 10 8 has the same function as the management computer 10 6 described above. In other words, when you create a class evening, you create a mass evening table that records the class evening contents.
- the computer that formed the class 10 8 searches for other candidate computers that are the latest computers in the status table possessed by other computers, class settings, and other additional candidates whose status is “f ree”. , This is the identified weather Decide whether to add an auxiliary computer based on the connection ratio of the class evening.
- the computer 108 that formed the class evening executes the job in the class evening that has grown in this way, and after the execution ends, the class evening disappears.
- the present invention can also be implemented as a network system with a plurality of processors or multi-core processors.
- the internal bus functions as the computer network 104 described above, and either one of the processors or multi-core processors connected to the internal bus or Some may be operated as the management computer 106 described above, and the remaining processors and the like may be operated as the computer 108 described above.
- the individual multi-core processors themselves illustrated in Fig. 13 (b) can be implemented as a network system.
- the operation of the management computer 106 described above is realized by one or several cores (processor cores) connected to the internal bus, the input / output control unit (IZO), and the cache memory.
- the remaining core may be operated as the computer 108 described above.
- the class evening list that lists the identification information (connection address, etc.) of other clusters associated with the computer belonging to the accessed class evening. Is used.
- the class evening list is provided, for example, on the master table.
- FIG 14 shows an example of the contents of the master table in this case.
- This cell evening table 2 1 6 3 is the same as the master table 2 1 6 1 shown in Fig. 6 (a).
- the mass evening management computer evening lists the new cluster identification information in the class evening list 2 3 6 9 each time a new class evening is associated with the class evening managed by itself.
- “association” means that a computer belonging to one class evening and a computer belonging to another class evening are logically connected (contacted) to perform linkage processing.
- the master management computer reads the identification information of one class evening listed in the class evening list 2 3 6 9 in response to a request from the slave management computer, for example, and notifies the slave management computer of it.
- this class list 2 3 6 9 may be browsed, for example, by a slave management computer.
- the reading order from the class evening list 2 3 6 9 may be random or in a predetermined order.
- the order of listing the class evening identification information to the class evening list 2 3 6 9 may be arbitrary.
- the order of listing is weighted.
- the other clusters having higher relevance to the class contents managed by the mass management computer are read out first.
- the relevance in this case is determined by, for example, the number of accesses between clusters, the number of communications within a predetermined time between classes, the degree of relevance of distributed processing between classes (such as audio processing that should be synchronized with image processing), etc. be able to.
- the operation mode in the duplication mode is as follows.
- a plurality of classes C6 to C11 are already formed.
- Each class C 6 to C 11 has one or more computers (combined evening 10 8, the same shall apply hereinafter).
- the management computer that created this class evening C 20 is not only a computer belonging to the class evening C 20, but also a number of computers that make it easier to grow other class evenings, that is, candidate computers (Fig. (Not shown) are also managed by the company.
- candidate computers Fig. (Not shown) are also managed by the company.
- the management computer that manages the candidate computer—the evening becomes the slave management computer.
- the management computer that manages cluster C 2 0 searches for class C 7 from clusters C 6 to C 11. Then, an inquiry is made to the mass evening managing management of this class evening C7, and information on the relationship between the class evening C7 and other class evenings in the master table 2 1 6 3 is obtained. In the case of the middle row in Fig. 15, cluster C 7 is related only to class evening C 1 1. Therefore, the management computer managing class C 2 0 adds the candidate computer to cluster C 11 and associates the candidate computer with class C 2 0 (the computer belonging to it) (see the lower part of Fig. 15). ). Also, update the contents of the slave table owned by itself.
- the master management computer that manages cluster C 1 1 adds the list of newly added computers (candidate computers before addition) to computer table 2 1 6 3 in master table 2 1 6 3.
- the identification information of class evening C 2 0 is added to the cluster list 2 3 6 9.
- Fig. 16 (a) shows that when a class C 20 to which a computer newly added to the network system belongs is created, the management computer that manages the class C 20 adds a candidate computer under its control. Suppose that class evening C 1 1 is reached when searching for a destination.
- the class evening list of class evening C 11 is associated with a plurality of computers in cluster C 1 1 in addition to class evening C 6 and C7, which are related one-to-one with class evening C 11 computers. If C8 and C10 are listed, the candidate computer is added to one of clusters C6, C7, C8, and C10. The same applies to Fig. 16 (b).
- the recording information in the slave table and the mass evening table is updated in the same manner as in the example shown in FIG.
- a new class C 20 is added to the network system.
- the management computer managing the cluster C 20 searches for the addition destination of its own candidate computers, it reaches the class CI 1 until it reaches class C 1. This is the same as the examples in (a) and (b).
- computers # 1 to # 5 belong, and according to cluster list 2369, computers # 1, # 3 are cluster C10, computers Assume that # 4 is associated with class evening C7, computer # 5 is associated with class evening C6, and combination evening # 2 is not associated with anything.
- computers # 1, # 3 are cluster C10, computers Assume that # 4 is associated with class evening C7, computer # 5 is associated with class evening C6, and combination evening # 2 is not associated with anything.
- the reading order of the computer list 2363 is also random, when computer # 1 is selected from the computer list 2363, the class associated with it is selected. In addition, candidate computers are added. When computer # 2 is selected, there is no class associated with it, so it is reselected.
- the present invention can be applied to a network system for facilitating clustering in which the number of nodes connecting computers, for example, follows a normal distribution or a power distribution. Also, it can be applied to distributed computing in general, which can efficiently execute information processing that varies in size, type, and number of jobs submitted. .
Abstract
Description
Claims
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/659,290 US8775622B2 (en) | 2004-08-02 | 2005-08-01 | Computer-based cluster management system and method |
EP05768590A EP1785865A4 (en) | 2004-08-02 | 2005-08-01 | NETWORK SYSTEM, MANAGEMENT COMPUTER, CLUSTER MANAGEMENT PROCESS AND COMPUTER PROGRAM |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2004-225315 | 2004-08-02 | ||
JP2004225315 | 2004-08-02 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2006013992A1 true WO2006013992A1 (ja) | 2006-02-09 |
Family
ID=35787268
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2005/014467 WO2006013992A1 (ja) | 2004-08-02 | 2005-08-01 | ネットワークシステム、管理コンピュータ、クラスタ管理方法およびコンピュータプログラム |
Country Status (4)
Country | Link |
---|---|
US (1) | US8775622B2 (ja) |
EP (1) | EP1785865A4 (ja) |
CN (1) | CN100465901C (ja) |
WO (1) | WO2006013992A1 (ja) |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9594599B1 (en) * | 2009-10-14 | 2017-03-14 | Nvidia Corporation | Method and system for distributing work batches to processing units based on a number of enabled streaming multiprocessors |
US9454441B2 (en) * | 2010-04-19 | 2016-09-27 | Microsoft Technology Licensing, Llc | Data layout for recovery and durability |
US8892740B2 (en) * | 2010-09-10 | 2014-11-18 | International Business Machines Corporation | Dynamic application provisioning in cloud computing environments |
US9467494B1 (en) * | 2011-12-30 | 2016-10-11 | Rupaka Mahalingaiah | Method and apparatus for enabling mobile cluster computing |
JP6155861B2 (ja) * | 2013-06-06 | 2017-07-05 | 富士通株式会社 | データ管理方法、データ管理プログラム、データ管理システム及びデータ管理装置 |
US11422907B2 (en) | 2013-08-19 | 2022-08-23 | Microsoft Technology Licensing, Llc | Disconnected operation for systems utilizing cloud storage |
US9798631B2 (en) | 2014-02-04 | 2017-10-24 | Microsoft Technology Licensing, Llc | Block storage by decoupling ordering from durability |
CN106371928A (zh) * | 2016-09-18 | 2017-02-01 | 安徽爱她有果电子商务有限公司 | 一种管理计算机的方法 |
CN106357444A (zh) * | 2016-09-18 | 2017-01-25 | 安徽爱她有果电子商务有限公司 | 一种计算机网络管理系统 |
CN106339261A (zh) * | 2016-09-18 | 2017-01-18 | 安徽爱她有果电子商务有限公司 | 一种计算机集群管理方法 |
US10705883B2 (en) * | 2018-06-19 | 2020-07-07 | Microsoft Technology Licensing, Llc | Dynamic hybrid computing environment |
CN113139704B (zh) * | 2020-01-17 | 2024-04-09 | 中国石油化工股份有限公司 | 一种用于钻井仿真的钻井多参数计算系统及方法 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0750256A2 (en) | 1995-06-23 | 1996-12-27 | Data General Corporation | Framework for managing cluster membership in a multiprocessor system |
JP2004341962A (ja) * | 2003-05-16 | 2004-12-02 | Toyota Infotechnology Center Co Ltd | 分散コンピューティング・システム、方法、プログラムおよび記録媒体 |
JP2005251160A (ja) * | 2004-02-06 | 2005-09-15 | Nippon Telegraph & Telephone East Corp | 利用形態指向p2p型グリッドコンピューティングシステム、及び、コンピュータプログラム |
Family Cites Families (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2834210B2 (ja) * | 1988-09-14 | 1998-12-09 | 株式会社日立製作所 | リング状ネットワークにおけるメッセージ制御方法 |
US5822531A (en) * | 1996-07-22 | 1998-10-13 | International Business Machines Corporation | Method and system for dynamically reconfiguring a cluster of computer systems |
US6748429B1 (en) * | 2000-01-10 | 2004-06-08 | Sun Microsystems, Inc. | Method to dynamically change cluster or distributed system configuration |
US6675265B2 (en) * | 2000-06-10 | 2004-01-06 | Hewlett-Packard Development Company, L.P. | Multiprocessor cache coherence system and method in which processor nodes and input/output nodes are equal participants |
US20020049859A1 (en) * | 2000-08-25 | 2002-04-25 | William Bruckert | Clustered computer system and a method of forming and controlling the clustered computer system |
US7120681B2 (en) * | 2001-01-30 | 2006-10-10 | International Business Machines Corporation | Cluster control in network systems |
US6889253B2 (en) * | 2001-04-30 | 2005-05-03 | International Business Machines Corporation | Cluster resource action in clustered computer system incorporation prepare operation |
EP1415377A4 (en) * | 2001-07-06 | 2007-05-30 | Computer Ass Think Inc | SYSTEM AND METHOD FOR MANAGING OBJECT BASED GROUPS |
CN1213567C (zh) * | 2002-02-27 | 2005-08-03 | 华为技术有限公司 | 一种网络设备的集群管理方法 |
US7139925B2 (en) * | 2002-04-29 | 2006-11-21 | Sun Microsystems, Inc. | System and method for dynamic cluster adjustment to node failures in a distributed data system |
CA2489142C (en) * | 2002-06-20 | 2013-11-26 | Paul Francis Mckee | Distributed computer |
-
2005
- 2005-08-01 EP EP05768590A patent/EP1785865A4/en not_active Withdrawn
- 2005-08-01 US US11/659,290 patent/US8775622B2/en active Active
- 2005-08-01 CN CNB2005800330053A patent/CN100465901C/zh active Active
- 2005-08-01 WO PCT/JP2005/014467 patent/WO2006013992A1/ja active Application Filing
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0750256A2 (en) | 1995-06-23 | 1996-12-27 | Data General Corporation | Framework for managing cluster membership in a multiprocessor system |
JP2004341962A (ja) * | 2003-05-16 | 2004-12-02 | Toyota Infotechnology Center Co Ltd | 分散コンピューティング・システム、方法、プログラムおよび記録媒体 |
JP2005251160A (ja) * | 2004-02-06 | 2005-09-15 | Nippon Telegraph & Telephone East Corp | 利用形態指向p2p型グリッドコンピューティングシステム、及び、コンピュータプログラム |
Non-Patent Citations (5)
Title |
---|
DOU W. ET AL: "A P2P approach for global computing", PROCEEDINGS OF INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS'03), 26 April 2003 (2003-04-26), XP010645374 * |
HAN J. AND PARK D.: "A lightweight personal grid using a supernode network", PROCEEDINGS OF THIRD INTERNATIONAL CONFERENCE ON PEER-TO-PEER COMPUTING (P2P'03), 3 September 2003 (2003-09-03), pages 168 - 175, XP010657537 * |
OGSTON E. ET AL: "A method for decentralized clustering in large multi-agent systems", PROCEEDINGS OF THE SECOND INTERNATIONAL JOINT CONFERENCE ON AUTONOMOUS AGENTS AND MULTIAGENT SYSTEMS, 18 July 2003 (2003-07-18), pages 789 - 796, XP002993363 * |
See also references of EP1785865A4 |
VAZQUEZ A.: "Growing network with local rules: Preferential attachment, clustering hierarchy, and degree correlations", PHYSICAL REVIEW, vol. 67, no. 5, 2003, pages 056104.1 - 056104.15, XP002993364 * |
Also Published As
Publication number | Publication date |
---|---|
CN101031886A (zh) | 2007-09-05 |
US8775622B2 (en) | 2014-07-08 |
US20080320138A1 (en) | 2008-12-25 |
CN100465901C (zh) | 2009-03-04 |
EP1785865A1 (en) | 2007-05-16 |
EP1785865A4 (en) | 2008-12-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2006013992A1 (ja) | ネットワークシステム、管理コンピュータ、クラスタ管理方法およびコンピュータプログラム | |
US7107323B2 (en) | System and method of file distribution for a computer system in which partial files are arranged according to various allocation rules | |
US7383286B2 (en) | File management system with parent directory search functions | |
JP4681374B2 (ja) | ストレージ管理システム | |
US8549248B2 (en) | Data migration method and information processing system | |
US7783737B2 (en) | System and method for managing supply of digital content | |
US8683482B2 (en) | Computer system for balancing access load of storage systems and control method therefor | |
US20060218369A1 (en) | Storage control method and system | |
CN108833503A (zh) | 一种基于ZooKeeper的Redis集群方法 | |
JPH09244940A (ja) | 分散計算機資源の管理方法 | |
US20100138540A1 (en) | Method of managing organization of a computer system, computer system, and program for managing organization | |
JP2003248611A (ja) | 記憶管理統合システム、および、その記憶管理制御方法 | |
JP2005228278A (ja) | 記憶領域の管理方法、管理装置及び管理プログラム | |
WO2011158300A1 (ja) | 計算機システムおよびその更改方法 | |
JP2000207370A (ja) | 分散ファイル管理装置及び分散ファイル管理システム | |
JP4485428B2 (ja) | ネットワークシステム、管理コンピュータ、クラスタ管理方法およびコンピュータプログラム | |
CN114816272B (zh) | Kubernetes环境下的磁盘管理系统 | |
JP2002297432A (ja) | 分散処理型データベース管理システム | |
JP3688471B2 (ja) | エージェントシステム、情報処理方法及び情報処理用ソフトウェアを記録した記録媒体 | |
JP2005301880A (ja) | 計算機システムにおけるデータ入出力処理方法、ストレージ装置、ホスト計算機、および計算機システム、 | |
CN111400110A (zh) | 数据库访问管理系统 | |
JP5543653B2 (ja) | 管理計算機 | |
JP2006268588A (ja) | クライアントサーバシステムおよびクライアントサーバシステムのデータ処理方法 | |
US20230266907A1 (en) | Storage system and data transfer method for storage system | |
JP2017207805A (ja) | 情報処理装置、及び情報処理システム |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS KE KG KM KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NG NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SM SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU LV MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2005768590 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 200580033005.3 Country of ref document: CN |
|
WWP | Wipo information: published in national office |
Ref document number: 2005768590 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 11659290 Country of ref document: US |