WO2010140194A1

WO2010140194A1 - Information processing system management method, information processing system, recording medium, and management program

Info

Publication number: WO2010140194A1
Application number: PCT/JP2009/002551
Authority: WO
Inventors: 山島弘之; 土屋哲
Original assignee: 富士通株式会社
Priority date: 2009-06-05
Filing date: 2009-06-05
Publication date: 2010-12-09

Abstract

Instances are arranged on multiple first physical machines (73) provided in a sequential arrangement area (7) in response to an assignment request. Arrangement information indicating how the multiple instances are arranged on the multiple first physical machines (73) is obtained. Based on the arrangement information, the instances arranged on the multiple first physical machines (73) are changed to an arrangement different from the arrangement on the multiple first physical machines (73) and are moved onto multiple second physical machines (83) provided in an optimal arrangement area (8).

Description

Information processing system management method, information processing system, recording medium, and management program

The present invention relates to an information processing system management method for managing a virtual server, an information processing system, a recording medium, and a management program.

There are cases where a company installs a server and provides various services to users online. In this case, the company must install and maintain the server by itself. Therefore, a data center including a plurality of servers is provided, and an operator of the data center causes the user to use the server, for example, in units of time. In this case, the operator of the data center manages the server resources, and the company does not need to install or maintain the server.

The computer is provided with storage device management means for managing a storage device for storing data used by the computer. The storage device management means periodically monitors the time characteristic information, and the function characteristic corresponding to the time characteristic information is provided. It has been proposed to perform data movement when the storage area is different from the storage area to which data is currently allocated.

JP 2005-196625 A

When a data center including a plurality of servers is provided, the data center operator manages server resources. At this time, from the viewpoint of the data center operator, if the server resources can be managed more effectively, the resource utilization efficiency can be improved.

An object of the present invention is to provide an information processing system management method capable of managing server resources more effectively.

The disclosed information processing system management method includes the following processes. That is, in response to the allocation request, virtual servers are arranged on a plurality of first physical machines provided in the first area. Request information required for allocation is extracted from the allocation request and recorded in the request information storage unit. Request information of a plurality of virtual servers arranged on a plurality of first physical machines is obtained. Based on the request information, the virtual server arranged on the plurality of first physical machines is moved to the plurality of second physical machines provided in the second area.

According to the disclosed information processing system management method, the placement of virtual servers placed on physical machines can be changed, server resources can be managed more effectively, and resource utilization efficiency can be improved. .

It is a figure which shows an example of a structure of an information processing system. It is a figure which shows an example of a structure of an information processing system. It is a figure which shows an example of a structure of an information processing system. It is explanatory drawing of the resource management in an information processing system. It is explanatory drawing of the resource management in an information processing system. It is explanatory drawing of the resource management in an information processing system. It is a resource management processing flow in an information processing system. It is a resource management processing flow in an information processing system. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of an example of resource management. It is explanatory drawing of the resource management in the information processing system which this inventor examined. It is explanatory drawing of the resource management in the information processing system which this inventor examined.

The present inventor examined a method for more effectively managing server resources and improving resource utilization efficiency. According to the study of the present inventor, it is effective to virtualize the server instead of allowing the user to use the server in units of physical machines. Furthermore, it is effective to start and operate a plurality of virtualized servers on one physical machine.

In this specification, a virtualized server is referred to as a “virtual server” or “instance” and is distinguished from a “physical machine”. The physical machine is a physical server, that is, a real server. The server is a computer having a communication function.

For example, 1000 physical machines are provided in the data center. A plurality of instances can be arranged on one physical machine. Therefore, for example, 5000 to 10,000 instances are provided in the data center as a whole. The inventor examined the arrangement of instances in such a data center.

18 and 19 are explanatory diagrams of management of instances, that is, virtual servers in the information processing system examined by the present inventors. In particular, FIG. 18 is an explanatory diagram of the information processing system examined by the present inventors, and FIG. 19 is an instance management processing flow in the information processing system of FIG.

The request information input unit 121 transmits an allocation request to the request information extraction unit 122 when an allocation request for an instance of one tenant is input (step S102) while waiting for an allocation request to be input (step S101). In response to this, the request information extraction unit 122 extracts request information such as the size and number of instances and configuration information from the allocation request (step S103), and stores the request information stored in the data management unit 151 via the access unit 152. The data is written in the unit 153 (step S104). The size and number of instances will be described later. Further, the request information extraction unit 122 transmits the request information to the arrangement unit 123.

Thereafter, the information reading unit 124 reads the machine information about the physical machine from the machine information storage unit 154 of the data management unit 151 and reads the arrangement information about the instance from the arrangement information storage unit 155 via the access unit 152 ( In step S105, the machine information and the arrangement information are transmitted to the arrangement unit 123.

Thereafter, the placement unit 123 creates new placement information about the instance by sequentially executing the placement design based on the request information, the machine information, and the placement information (step S106). The data is written into the arrangement information storage unit 155 of the data management unit 151 (step S107).

Thereafter, the deployment unit 157 reads the placement information from the placement information storage unit 155 of the data management unit 151 via the access unit 152, and based on the placement information, the tenant instance request information input in step S102 The instance, that is, the virtual server is deployed on the physical machine so as to correspond to (Step S108).

Here, for such a data center, the user requests an unpredictable number of instances with an unpredictable size at an unpredictable timing. In response to this request, the data center needs to immediately allocate an instance on one of the physical machines and start a service in response to a user request.

Therefore, the data center sequentially arranges and deploys the requested instances on the physical machine when the instance request arrives in order to respond to the user's request. For this reason, the instances are arranged and deployed in the order of arrival on a physical machine having a space that can be arranged when the request arrives. As a result, a resource that is not used as an instance is generated in each physical machine. Since the number of physical machines is large, extremely large unused resources are generated in the entire data center. Therefore, the accommodation rate of instances in the physical machine is reduced. In other words, the resource utilization rate decreases.

In the information processing system disclosed below, a management method capable of improving the accommodation rate of instances on a physical machine is realized.

FIG. 1 is a diagram illustrating an example of a configuration of an information processing system.

The information processing system includes a data center 100, a user terminal 300, a tenant request terminal 500, a network 200 that connects the data center 100 and the user terminal 300, and a data center 100 and the tenant request terminal 500. And a network 400 to be connected. The data center 100 includes one or a plurality of virtual servers or instances constituting a tenant requested from the tenant request terminal 500 on a plurality of physical machines, and causes the user terminal 300 to use the tenant or virtual server. The

networks

200 and 400 are, for example, the Internet or dedicated lines. The user terminal 300 is a computer that uses a tenant of the data center 100 or a virtual server. The tenant request terminal 500 is a computer that requests a tenant to be used to the data center 100, that is, one or more virtual servers. A service provided by a set of a plurality of virtual servers requested from the tenant request terminal 500 or a service provided by one virtual server is referred to as a “tenant”.

The tenant request terminal 500 transmits an instance allocation request to the data center 100 via the network 400. An allocation request is issued when, for example, a new instance is used, an instance is already used, but the instance quantity is to be increased further, or an instance is already used but the instance quantity is to be reduced. Is done. When the data center 100 places / deploys an instance on a physical machine in response to an allocation request, the user terminal 300 can connect to the physical machine on which the instance is placed via the network 200 and use a tenant. it can.

Note that the user terminal 300 and the tenant request terminal 500 may be the same physical machine, and the

networks

200 and 400 may be the same network.

The data center 100 includes a management device 1 and a machine group 6. The management apparatus 1 manages the machine group 6 and arranges and deploys the instance on the physical machine 73 or 83 (see FIG. 2) of the machine group 6. The machine group 6 includes a plurality of

physical machines

73 and 83. Instances are arranged and deployed on the

physical machines

73 and 83.

Instance placement is to allocate a virtual server (instance) on the activated physical machine, and instance deployment is to start the virtual server (instance) on the activated physical machine. . The tenant request terminal 500 possesses the right to use and partly manage the instance, that is, the virtual server, when the instance is arranged and deployed on the physical machine according to the request information.

The management apparatus 1 includes a sequential arrangement processing unit 2, a data management deployment processing unit 3, a determination processing unit 4, and an optimal arrangement processing unit 5. The management device 1 will be described later with reference to FIG. The machine group 6 includes a sequential placement area 7 and an optimal placement area 8. The machine group 6 will be described later with reference to FIG.

FIG. 2 is a diagram showing an example of the configuration of the information processing system, and shows the machine group 6.

In the machine group 6, the sequential arrangement area 7 is a first area including a plurality of first physical machines 73. In the sequential placement area 7, instances are placed and placed on the plurality of first physical machines 73 in response to the assignment request at the timing when the assignment request is input. The instances are sequentially arranged and arranged in the arrangement area 7 in the order in which the management device 1 receives the allocation request. Therefore, the first area is the sequential arrangement area 7. The sequential arrangement area 7 includes a rack 71 for mounting the physical machine 73.

For example, as shown in FIG. 2, it is assumed that allocation requests for instances # 1 to # 9 arrive at the management apparatus 1 in this order. In this case, the instances # 1 to # 9 are arranged and deployed in the order of arrival on the physical machine 73 having a space that can be arranged when the allocation request arrives. As a result, in each physical machine 73, a useless resource in which no instance is arranged is generated. However, since the number of physical machines 73 is small, when viewed from the entire data center 100, useless resources generated in the physical machines 73 can be ignored.

In the machine group 6, the optimum arrangement area 8 is an area physically different from the first area, and is a second area including a plurality of second physical machines. In the optimal placement area 8, instances are placed and deployed on the plurality of second physical machines at the timing when the conditions for instance movement are satisfied. Specifically, at the timing, the instances deployed on the plurality of first physical machines 73 are moved onto the plurality of second physical machines 83 provided in the second area. As a result, instances are arranged and deployed on the plurality of second physical machines 83.

At this time, as will be described later, the instance arranged on the first physical machine 73 is moved to the second physical machine 83 after being changed to the optimum arrangement based on the request information. Therefore, the second area is the optimum arrangement area 8. The optimal arrangement area 8 includes a rack 81 for mounting the physical machine 83.

For example, as shown in FIG. 2, the instances # 1 to # 9 are arranged and deployed on the second physical machine 83 after being changed to an optimal arrangement regardless of the arrival order. Thus, in order to arrange the instances # 1 to # 9, five physical machines 73 are sequentially used in the arrangement area 7 before the movement, but after the movement, four physical machines 83 in the optimum arrangement area 8 are used. Only used. As a result, since the number of physical machines 83 is extremely large, the utilization efficiency of the physical machines 83 can be greatly improved when viewed from the entire data center 100.

The number of racks in the sequential arrangement area is, for example, one. The number of racks in the optimal arrangement area is, for example, 99. For example, one rack includes ten

physical machines

73 or 83. Therefore, 1000

physical machines

73 or 83 are provided in the machine group 6. The number of racks, the number of physical machines 73, and the number of physical machines 83 may be various values other than those described above.

For example, 10 instances can be arranged on one physical machine 73. The number of instances may be a value other than ten. The same applies to the physical machine 83. The number of instances that can be arranged on the physical machine 73 may be different from the number of instances that can be arranged on the physical machine 83. For example, the number of instances that can be arranged on the physical machine 73 may be ten, and the number of instances that can be arranged on the physical machine 83 may be twenty.

For example, in the physical machine 73, it is assumed that the number of clocks of its CPU (Central Processing Unit) is 1 GHz and the capacity of the main memory is 1 Gbyte. The instance size “1” is assumed that the number of clocks of the CPU is 100 MHz and the capacity of the main memory is 100 Mbytes. In this case, ten instances of size “1” can be arranged on the physical machine 73. The instance size is the performance value of the instance. The performance value of the instance may be a value other than the above. The performance value of the instance may be determined using a parameter that determines performance other than the number of clocks of the CPU. For example, the number of CPUs (cores), communication bandwidth, hard disk capacity, access speed to the hard disk, memory amount, and the like may be used, or a combination of the above-described parameters.

The instance size “1” is the smallest unit of the virtual server that can be specified in the allocation request. The instance size is specified along with the number of instances in the allocation request. For example, it is requested to allocate five instances of size “4”. Accordingly, the instances are arranged on the plurality of physical machines 73 or the plurality of physical machines 83 based on the size and quantity of the instances specified in the allocation request.

FIG. 3 is a diagram showing an example of the configuration of the information processing system, and shows the management apparatus 1.

In the management apparatus 1, the sequential arrangement processing unit 2 receives the instance allocation request input from the tenant request terminal 500, and creates request information based on the request. Then, based on the request information and machine information, the sequential placement processing unit 2 sequentially places the allocation requested instances on the physical machines 73 in the placement area 7, and creates placement information for the placement.

The actual instance deployment is executed by the data management deployment processing unit 3 as described later. Therefore, the sequential placement processing unit 2 actually only determines the sequential placement of instances in the sequential placement area 7. In other words, the sequential arrangement processing unit 2 creates arrangement information for the arrangement, and notifies the determination processing unit 4 of the creation.

The sequential arrangement processing unit 2 includes a request information input unit 21, a request information extraction unit 22, a sequential arrangement unit 23, and a sequential arrangement information reading unit 24.

The request information input unit 21 receives the instance allocation request input from the tenant request terminal 500 and transmits it to the request information extraction unit 22. The request information extraction unit 22 extracts request information from the allocation request, writes the request information to the request information storage unit 33 via the access unit 32, and sequentially arranges the request information into the arrangement unit 23 and the request information analysis unit 43. And send to. The sequential arrangement information reading unit 24 reads the machine information from the machine information storage unit 34, reads the arrangement information from the arrangement information storage unit 35, and transmits the machine information and the arrangement information to the arrangement unit 23 sequentially via the access unit 32. To do. The request information and machine information will be described later.

The sequential placement unit 23 sequentially places the allocation requested instances on the physical machines 73 in the placement area 7 based on the request information, the machine information, and the placement information. In addition, the sequential placement unit 23 creates new placement information based on the placement and writes the new placement information in the placement information storage unit 35. The arrangement information will be described later. Further, the sequential arrangement unit 23 updates the machine information in the machine information storage unit 34 based on the new arrangement information.

The data management deployment processing unit 3 stores request information created by the sequential placement processing unit 2, machine information prepared in advance, placement information created by the sequential placement processing unit 2, and movement information created by the optimal placement processing unit 5. .

Also, the data management deployment processing unit 3 actually deploys the instances on the physical machines 73 in the placement area 7 based on the placement information created from the request information and the machine information. Further, the data management / deployment processing unit 3 actually moves the instances to be moved sequentially from the physical machine 73 in the arrangement area 7 to the physical machine 83 in the optimum arrangement area 8 based on the arrangement information and the movement information. .

At this time, when the power state of the second physical machine to which the plurality of instances arranged on the plurality of physical machines 73 are moved in the plurality of physical machines 83 is OFF. Prior to the movement, the power is turned on to start up. In addition, the data management deployment processing unit 3 turns on the first physical machine that is the migration source of the plurality of instances arranged on the plurality of physical machines 73 in the optimum arrangement area 8 after the migration. Let it be in the activated state. However, if the number of physical machines 73 that have been activated is greater than or equal to a preset threshold value, the power may be turned off to the threshold value.

The data management deployment processing unit 3 includes a data management unit 31, an access unit 32, a request information storage unit 33, a machine information storage unit 34, an arrangement information storage unit 35, a movement information storage unit 36, and a deployment unit 37. And a mobile deployment unit 38.

The data management unit 31 includes an access unit 32, a request information storage unit 33, a machine information storage unit 34, an arrangement information storage unit 35, and a movement information storage unit 36. The access unit 32 sequentially reads or writes the information instructed from the arrangement information reading unit 24, the deployment unit 37, and the mobile deployment unit 38 in the storage units 33 to 36 corresponding to the instruction.

The request information storage unit 33 stores the request information as will be described later with reference to FIG. The machine information storage unit 34 stores machine information as will be described later with reference to FIG. The arrangement information storage unit 35 stores arrangement information as will be described later with reference to FIG. The movement information storage unit 36 stores movement information as will be described later with reference to FIG.

The deployment unit 37 sequentially deploys the instances on the physical machines 73 in the placement area 7 based on the placement information in the placement information storage unit 35. The movement deployment unit 38 deploys the instance on the physical machine 83 in the optimal arrangement area 8 based on the arrangement information in the arrangement information storage unit 35 and the movement information in the movement information storage unit 36.

The determination processing unit 4 examines, for example, whether or not resources allocated to a plurality of instances sequentially arranged on the placement area 7 have exceeded a preset threshold value. Alternatively, for example, the determination processing unit 4 checks whether or not the total number of request instances of a plurality of request information exceeds a preset threshold value. In addition, the determination processing unit 4 checks, for example, whether a preset time has elapsed since the previous movement. As described above, the determination processing unit 4 determines whether or not the instance movement condition is satisfied. The determination processing unit 4 notifies the optimal placement processing unit 5 of the determination result.

The determination processing unit 4 includes a timer 41, an instance number determination unit 42, a request information analysis unit 43, and an additional adjustment unit 44.

The instance number determination unit 42 checks whether or not resources allocated to a plurality of instances sequentially arranged on the arrangement area 7 exceed a preset threshold value. In addition, the instance number determination unit 42 checks, for example, whether or not the total number of request instances of a plurality of request information exceeds a preset threshold value. When the resource allocated to the instance exceeds the threshold, the instance number determination unit 42 notifies the additional adjustment unit 44 of this.

The threshold is, for example, 10 instances. A value other than this may be used as the threshold value. For example, the minimum number of virtual server units, the number of physical machines 73 in which instances are arranged, and the like may be used.

The timer 41 checks whether a preset time has elapsed since the previous movement. When the preset time has elapsed, the timer 41 generates a timer interrupt and activates the additional adjustment unit 44.

The preset time is, for example, 30 minutes. The preset time determines the cycle of instance movement. The determination processing unit 4 uses both a threshold value and a preset time. Therefore, the movement of the instance is executed in the cycle of the movement even if the number of instances does not exceed the threshold value. Other values may be used as the preset time. For example, assuming that the preset time is one day, the instance may be moved before or after the start of the day's work.

The request information analysis unit 43 accumulates and analyzes the request information from the request information extraction unit 22. For example, the request information analysis unit 43 accumulates all the request information for all the instances arranged on the physical machine 73 in the sequential arrangement area 7 and rearranges them in the order of the size of the request instance size. And how many are there. That is, all sequential arrangement request information is created. The request information analysis unit 43 transmits all sequential arrangement request information as an analysis result to the instance number determination unit 42 and the additional adjustment unit 44.

The additional adjustment unit 44 is optimized based on all sequential arrangement request information sent from the request information analysis unit 43, optimum arrangement machine information and optimum arrangement information acquired from the optimum arrangement information reading unit 53 via the access unit 32. In the arrangement area 8, whether or not the instances sequentially arranged in the arrangement area 7 can be allocated to surplus resources in the physical machine 83 in which the instances are already arranged, that is, resources in which no instances are arranged. Investigate. When the instances sequentially arranged in the placement area 7 can be assigned to the surplus resources, the additional adjustment unit 44 determines to assign the instances to the physical machine 83, and these are allocated to the optimum placement unit 51, The optimum arrangement information creating unit 52 is notified.

The optimum placement processing unit 5 sequentially moves the instances in the placement area 7 to the optimum placement area 8 when the condition for moving the instance is satisfied. Alternatively, request instances of request information that have not been sequentially placed in the placement area 7 are placed in the optimum placement area 8. In other words, the optimum placement processing unit 5 has a plurality of resources arranged on the plurality of physical machines 73 when resources allocated to the plurality of instances sequentially arranged on the placement area 7 exceed a preset threshold. The instance is moved onto a plurality of physical machines 83. Alternatively, the optimum placement processing unit 5 may replace the plurality of instances placed on the plurality of physical machines 73 with the plurality of physical machines when the total number of request instances of the plurality of request information exceeds a preset threshold. The request instances of the request information that have not yet been sequentially placed in the placement area 7 are placed in the optimum placement area 8. Further, the optimum placement processing unit 5 moves a plurality of instances arranged on the plurality of physical machines 73 onto the plurality of physical machines 83 when a preset time has elapsed since the previous movement.

At this time, the optimum arrangement processing unit 5 arranges the instances arranged on the plurality of physical machines 73 to the optimum physical machine 83 in the optimum arrangement area 8 based on the request information.

In addition, the movement of the actual instance is executed by the data management deployment processing unit 3 as described above. Therefore, the optimum arrangement processing unit 5 actually only determines the optimum arrangement of instances in the optimum arrangement area 8. In other words, the optimum arrangement processing unit 5 creates movement information for the movement, creates new arrangement information after the movement, and notifies the data management deployment processing unit 3 of the creation. The optimum arrangement processing unit 5 includes an optimum arrangement unit 51, an optimum arrangement information creation unit 52, an optimum arrangement information reading unit 53, and a movement information creation unit 54.

When receiving the notification from the additional adjustment unit 44, the optimal arrangement unit 51 receives the result of the additional adjustment of the instance in the additional adjustment unit 44, the fully sequential arrangement request information, the optimum arrangement information read by the optimum arrangement information reading unit 53, and the optimum arrangement. Based on the machine information, the additional adjustment unit 44 places an instance whose destination has not been determined. Also, the optimum placement unit 51 creates the optimum placement information of the placed machines based on the placement and transmits it to the movement information creation unit 54. The optimal placement unit 51 uses an optimal placement algorithm to optimally place instances in the optimal placement area 8. As the optimal placement algorithm, for example, a bin packing algorithm, a knapsack problem solving algorithm, or the like can be used.

The optimum placement information creation unit 52 creates the optimum placement information of the physical machines that have been placed based on the additional adjustment of the instance in the addition adjustment unit 44, and transmits the optimum placement information to the movement information creation unit 54.

The optimum arrangement information reading unit 53 reads the optimum arrangement information and the optimum arrangement machine information from the arrangement information storage unit 35 and transmits them to the optimum arrangement unit 51 and the additional adjustment unit 44.

In other words, the movement information creation unit 54 is based on the optimal arrangement of the physical machines that have already been arranged as a result of the additional adjustment of the instance in the additional adjustment unit 44 and the optimal arrangement of the instance in the optimal arrangement unit 51. Based on the above, the movement information is created and written in the movement information storage unit 36. Further, the movement information creation unit 54 writes the optimal arrangement information into the arrangement information storage unit 35.

4 to 6 are explanatory diagrams of resource management in the information processing system, and show the movement of instances.

FIG. 4 shows the movement of the instance from the physical machine 73 in the sequential placement area 7 to the physical machine 83 in the optimal placement area 8. In FIG. 4, some physical machines 83 in the optimum arrangement area 8 are shown.

For example, an instance of the management ID “R # 1-1” is arranged on the physical machine 73 with the machine ID “M # 1”. The size of the instance of the management ID “R # 1-1” is “4”. Instances with management IDs “R # 1-1” to “R # 1-5” correspond to the same request ID “R # 1”. Accordingly, the instances with the management IDs “R # 1-1” to “R # 1-5” are arranged in this order on the physical machines 73 with the machine IDs “M # 1” to “M # 3”. The other instances are similarly arranged on the physical machine 73. In the physical machine 73 where the instance is arranged, a resource (excess resource) where the instance is not arranged is generated.

In this state, the instances in the placement area 7 are sequentially moved to the optimum placement area 8. At this time, the instance arrangement is optimized. As a result, the instances of the sequential placement area 7 are placed in the optimal placement area 8 with a different placement. As a result, in the optimum arrangement area 8, the physical machines 83 with the machine IDs “M # 11” to “M # 14” do not have surplus resources. In other words, the utilization efficiency of the physical machines 83 with the machine IDs “M # 11” to “M # 14” is 100%. Therefore, the utilization efficiency of the physical machine 83 can be greatly improved.

FIG. 5A shows an example of the request information storage unit 33 and request information stored therein.

The request information storage unit 33 stores a plurality of request information, for example, in the input order. For example, request IDs are given to the request information in the order of input. The request ID is an identification number uniquely determined in the information processing system.

1 piece of request information includes tenant ID, quantity, size, and management ID for each request ID. The tenant ID may be the user name that entered the request information. Actually, a tenant is a service provided by a plurality of virtual server sets or one virtual server. Alternatively, the virtual server capability itself may be a tenant. The quantity is the number of instances requested in the request information. The size is the size of the instance requested in the request information. The management ID is given to each instance requested in the request information. Each instance is managed by a management ID. The management ID is an identification number uniquely determined in the information processing system.

In the request information storage unit 33, information other than the above may be stored as request information. For example, the time when the request information arrives at the management apparatus 1 may be stored.

For example, when request information of a tenant ID: A is input from the tenant request terminal 500, the request ID “R # 1” is given to the request information, and the content of the request information is extracted. As the request information, it is extracted that the tenant ID is “A”, the quantity is “5”, and the size is “4”. Based on the extracted quantity “5”, management IDs “R # 1-1” to “R # 1-5” are given to each instance.

FIG. 5B shows an example of the machine information storage unit 34 and machine information stored in the machine information storage unit 34.

The machine information storage unit 34 stores a plurality of machine information. For example, machine IDs are assigned to the

physical machines

73 and 83 in advance in the order of installation. The machine ID is an identification number uniquely determined in the information processing system. In this example, the machine IDs M # 1 to M # 10 are assigned to the ten physical machines 73 sequentially arranged in the arrangement area 7 in the order of arrangement. Machine IDs of M # 11 to M # 1000 are assigned to the 990 physical machines 83 arranged in the optimum arrangement area 8 in the order of arrangement.

1 machine information includes area, state, and size for each machine ID. The area is an area in which the

physical machine

73 or 83 is arranged, and is either “sequential arrangement area” or “optimum arrangement area”. The status indicates the status of the

physical machine

73 or 83, for example, the status of its power supply. The size is the minimum unit unit number of the virtual server that the

physical machine

73 or 83 has, in other words, can be arranged on the

physical machine

73 or 83.

Note that various states other than the power supply state may be stored as the state. For example, information indicating that the physical machine is out of order, the physical machine is in a standby state, or the like may be stored. In this case, the instance is not arranged on the failed physical machine. Further, the physical machines 73 in the sequential arrangement area 7 are normally in a standby state. However, when the number of machines equal to or greater than the preset threshold is turned off, the number of machines equal to the threshold is in a standby state.

As a result, when request information for instances that can be arranged with a number equal to or less than the threshold number is input, an instance is immediately allocated on any of the physical machines 73 sequentially arranged in the arrangement area 7 according to the request information. Service can be started.

The state of the physical machine 83 arranged in the optimum arrangement area 8 is normally “OFF” and is changed from “OFF” to “ON” at a preset timing. This timing is when the instances in the sequential placement area 7 are moved to the optimal placement area 8 sequentially. For example, when the movement occurs, the state of the necessary number of physical machines 83 is changed from “OFF” to “ON” prior to the movement.

FIG. 6A shows an example of the arrangement information storage unit 35 and the arrangement information stored therein.

The arrangement information storage unit 35 stores the latest arrangement information. The arrangement information includes usage information, difference information, and vacancy for each machine ID. The usage information indicates a management ID assigned to the

physical machine

73 or 83 having the latest machine ID. The difference information indicates information obtained by updating the usage information with the latest request information. The free space is the number of the minimum unit unit of the virtual server that is not assigned to the management ID in the

physical machine

73 or 83 of the machine ID.

For example, an instance of the management ID “R # 1-1” and the size “4” of “R # 1-2” is arranged on the physical machine 73 with the machine ID “M # 1”. At this time, the empty or surplus resource of the physical machine 73 with the machine ID “M # 1” is “2”.

After this, when the instances are sequentially moved from the placement area 7 to the optimum placement area 8, the usage information and vacancy are changed for each machine ID. For example, the instances of the management IDs “R # 1-1” and “R # 1-2” are moved from the physical machine 73 with the machine ID “M # 1”. As a result, as shown in FIG. 6A, in the physical machine 73 with the machine ID “M # 1”, the usage information is “vacant”, and the vacant or surplus resource is “10”. The difference information is “R # 1-1, R # 1-2 → (empty)”. On the other hand, for example, the instances of the management IDs “R # 2-1” and “R # 2-2” are moved to the physical machine 83 with the machine ID “M # 11”. As a result, in the physical machine 83 with the machine ID “M # 11”, the usage information is in the state where the instances of the management IDs “R # 2-1” and “R # 2-2” are arranged from “free”. The empty or surplus resource is “0”. The difference information is “(empty) → R # 2-1, R # 2-2”.

FIG. 6B shows an example of the movement information storage unit 36 and movement information stored therein.

The movement information storage unit 36 stores a plurality of movement information. The movement information includes a request ID, a movement source, and a movement destination for each management ID. The migration source indicates the physical machine 73 where the instance of the management ID is currently arranged. The migration destination indicates the physical machine 83 to which the instance of the management ID is to be migrated. For example, the instance with the management ID “R # 1-1” is moved from the physical machine 73 with the machine ID “M # 1” to the physical machine 83 with the machine ID “M # 13”. The optimal arrangement is realized by moving instances having the same request ID to different physical machines 83. Here, for example, there may be an instance that does not have a migration source, such as an instance with a management ID “R # 3-1” or the like. This will be described later. The movement information may be overwritten in an empty state after the movement of the instance is completed, may be deleted, or may be stored by being overwritten when the next movement information is stored. You may add movement information at any time.

Note that the movement information storage unit 36 may store information other than the movement source as movement information. For example, the time when the request information about the instance to be moved arrives at the management apparatus 1, the movement time, the size of the instance to be moved, and the like may be stored.

7 and 8 show a machine arrangement process flow executed by the information processing system of FIG.

The request information input unit 21 is waiting for input (step S1). In this state, when an instance allocation request is input from one tenant, that is, the tenant request terminal 500 (step S2), the request information input unit 21 receives the allocation request.

Thereafter, the request information extraction unit 22 extracts request information such as the instance size, number, and configuration information from the input allocation request (step S3). Then, the sequential arrangement information reading unit 24 writes the extracted request information for the one tenant in the request information storage unit 33 of the data management unit 31 (step S4).

Thereafter, the request information analysis unit 43 receives the request information from the request information extraction unit 22, for example, and sequentially receives the request information storage unit 33 from the request information storage unit 33 through the sequential information reading unit 24, and the physical machines in the sequential arrangement area 7. Request information (all sequential placement request information) of all instances placed on 73 is collected and analyzed (step S5). Then, the instance number determination unit 42 determines the number of instances corresponding to the request information included in the collected all sequential arrangement request information. That is, the instance number determination unit 42 checks whether or not the number of instances in the sequential arrangement area 7 is larger than a threshold value (step S6).

When the number of instances in the sequential arrangement area 7 is not larger than the threshold (No in step S6), the sequential arrangement unit 23 that has received the notification from the instance number determination unit 42 sequentially transmits the arrangement area via the sequential arrangement information reading unit 24. 7 reads machine information (sequential placement area 7 machine information) of the physical machine 73 where the instance is placed from the machine information storage unit 34, and stores placement information (sequential placement information) about the instances in the sequential placement area 7 placement information. Reading from the unit 35 (step S7).

Thereafter, the sequential placement unit 23 performs sequential placement design based on the request information of one tenant, the machine information in the sequential placement area 7, and the sequential placement information. Then, the sequential layout unit 23 creates new layout information (new sequential layout information) for the instances of the sequential layout area 7 based on the result of the sequential layout design (step S8), and the sequential layout information reading unit 24 Then, the new sequential arrangement information is written into the arrangement information storage unit 35 of the data management unit 31 (step S9).

Thereafter, the deployment unit 37 sequentially deploys instances corresponding to the request information for one tenant on the physical machines 73 in the placement area 7 in accordance with the new sequential placement information (step S10). Thereafter, step S1 is repeated.

When there is a timer interrupt from the timer 41 in step S1, the additional adjustment unit 44 that has received the timer interrupt receives the optimum placement machine information from the machine information storage unit 34 of the data management unit 31 via the optimum placement information reading unit 53. And the optimum arrangement information is read from the arrangement information storage unit 35 (step S11).

In addition, when the number of instances in the sequential arrangement area 7 is larger than the threshold value in step S6 (Yes in step S6), the additional adjustment unit 44 that has received the notification from the instance number determination unit 42 executes step S11.

Prior to step S11, optimum placement machine information is created and stored in the machine information storage unit 34, and optimum placement information is created and stored in the placement information storage unit 35.

Thereafter, the additional adjustment unit performs adjustment to add the instance to the physical machine 83 (arranged machine) in which the instance has already been arranged based on the optimum arrangement machine information, the optimum arrangement information, and the all-sequential arrangement request information. (Step S12). Then, the optimum arrangement information creating unit 52 creates the optimum arrangement information of the arranged machines based on the additional adjustment of the instance (Step S13).

Further, the optimum placement unit 51 transfers the remaining instances that cannot be additionally adjusted to the placed machines (that is, the instances included in the all sequential placement request information) to the unused physical machines 83 in the optimum placement area 8. The optimum placement design for placement is performed, and the optimum placement information for the physical machine 83 (additional machine) to which the instance is additionally placed is created (step S14).

Thereafter, the movement information creation unit 54 creates movement information based on the sequential arrangement information, the optimum arrangement information of the arranged machines, and the optimum arrangement information to the additional machines (step S15), and the movement of the data management unit 31 is performed. The movement information is written into the information storage unit 36 (step S16).

Thereafter, the movement deployment unit 38 sequentially moves and deploys the instances from the placement area 7 to the optimum placement area 8 based on the movement information (step S17). Based on the actual movement, Together with the optimal placement information of the additional machine, the optimal placement information is overwritten, the sequential placement information is overwritten in an empty state, and the movement information is overwritten in an empty state (step S18). Thereafter, step S1 is repeated.

9 to 17 are explanatory diagrams of an example of resource management. In particular, FIG. 9 shows the input of request information, and FIGS. 10 to 17 show the assignment of instances.

In FIG. 10, FIG. 10 (A) shows instance allocation in the information processing system of FIG. FIG. 10B is shown for comparison with FIG. 10A, and shows the allocation of instances in the information processing system of FIG. The same applies to FIGS. 11 to 17.

As shown in FIG. 9, an assignment request is input from tenant A at time t1. This allocation request is a request for “5” instances having a size of “4”. When the instance request information is input, the data center 100 needs to start the service immediately. Therefore, when an allocation request is input at time t1, the management apparatus 1 assigns a request ID “R # 1” to the request and extracts the request information, as shown in FIG. And stored in the request information storage unit 33. Thereafter, the management apparatus 1 sequentially arranges instances corresponding to the allocation request for R # 1 on the physical machines 73 in the arrangement area 7.

As described above, in the example of FIG. 1, one physical machine 73 can be arranged with instances corresponding to a minimum instance unit “10”. On the other hand, the size of the instance corresponding to the allocation request for R # 1 is “4”. Further, at this time, since no instance is arranged in all the physical machines 73, the instances are arranged in order from the physical machine 73 having the head machine ID “M # 1”. For the machine ID, the machine ID shown in FIG. 4 is used (the same applies hereinafter).

Therefore, as shown in FIG. 10A, two instances with management IDs “R # 1-1” and “R # 1-2” are arranged on the physical machine 73 with the machine ID “M # 1”. Is done. Further, two instances of management IDs “R # 1-3” and “R # 1-4” are arranged on the physical machine 73 with the machine ID “M # 2”. Also, one instance of the management ID “R # 1-5” is arranged on the physical machine 73 with the machine ID “M # 3”. As a result, the arrangement information is stored in the arrangement information storage unit 35 as shown in FIG.

In FIG. 10B, as in FIG. 10A, “5” instances each having a size of “4” are arranged on the physical machine.

Next, as shown in FIG. 9, an allocation request for requesting “4 instances” of the size “5” is input from the tenant B at time t2. The management apparatus 1 assigns a request ID “R # 2” to this allocation request, and sequentially arranges instances corresponding to the allocation request for R # 2 on the physical machines 73 in the allocation area 7. The instances are arranged in order from the physical machine 73 with the machine ID “M # 3”. As a result, as shown in FIG. 11A, the management IDs “R # 2-1” to “R # 2-4” are placed on the physical machines 73 with the machine IDs “M # 3” to “M # 5”. Are arranged.

In FIG. 11B, as in FIG. 11A, “4” instances each having a size of “5” are arranged on the physical machine.

Next, as shown in FIG. 9, an assignment request (request ID = R # 3) for requesting “4 instances” of the size “3” is input from the tenant C at time t3.

In this case, in FIG. 12B, “4” instances each having a size of “3” are arranged on the physical machine. The management ID of this instance is “R # 3-1” to “R # 3-4”.

On the other hand, in FIG. 12A, the number of instances of the all-sequential arrangement request information is 13, which exceeds the threshold “10”. Therefore, the management apparatus 1 creates the movement information and stores it in the movement information storage unit 36 as shown in FIG. 6B, and then sequentially arranges the arrangement area up to that point as shown in FIG. 7 are moved onto the physical machine 83 in the optimum arrangement area 8, and the instance of the allocation request that triggered the threshold is arranged on the physical machine 83 in the optimum arrangement area 8. As a result of this movement, no instance exists in the sequential arrangement area 7 as indicated by the dotted line in FIG. On the other hand, the power of the physical machines 73 in the sequential arrangement area 7 remains on.

Note that the number of instances in the all sequential arrangement request information is the sum of the number of instances of the new allocation request # R3 and the number of instances actually arranged in the sequential arrangement area 7. The check of the number of instances in the all-sequential placement request information (step S6 in FIG. 7) is executed before the new allocation request # R3 instance is placed in the sequential placement area 7. For example, the instances of the management IDs “R # 3-1” to “R # 3-4” are actually arranged in the optimal arrangement area 8 without being sequentially arranged in the arrangement area 7. As a result, the movement information of the instances having the management IDs “R # 3-1” to “R # 3-4” does not have “movement source” as shown in FIG.

As described above, the state in which the instances are sequentially moved from the physical machine 73 in the arrangement area 7 to the physical machine 83 in the optimum arrangement area 8 corresponds to FIG. 4 described above. Also, the request information, machine information, arrangement information, and movement information at this time correspond to FIGS. 5A, 5B, 6A, and 6B, respectively.

There are nine instances of management IDs “R # 1-1” to “R # 1-5”, “R # 2-1” to “R # 2-4”. There are four R # 3-1 "to" R # 3-4 ". The migration or placement destination is the physical machine 83 with the machine IDs “M # 11” to “M # 16” in the optimum placement area 8. Prior to this movement or placement, only the physical machines 83 with the machine IDs “M # 11” to “M # 16” are powered on, and the other physical machines 83 of the optimal placement area 8 are not powered on. As a result, the updated machine information is stored in the machine information storage unit 34 as shown in FIG.

At this time, the management apparatus 1 optimally arranges the instances in the optimal arrangement area 8. As a result, as shown in FIG. 13A, the instances are arranged irrespective of the arrival times t1 to t3 of the corresponding allocation request, and the updated arrangement information is arranged as shown in FIG. It is stored in the information storage unit 35.

On the other hand, in FIG. 13B, the instance is not optimally arranged in the optimal arrangement area 8, and is in the same state as FIG. 12B. Therefore, there are many resources on which no instance is arranged on the physical machine on which the instance is arranged.

At this point, the number of

physical machines

73 and 83 that are powered on is larger in FIG. 13A than in FIG. 13B. However, as the number of instances to be arranged increases, it is not a problem that the physical machines 73 in the arrangement area 7 are sequentially turned on.

Next, as shown in FIG. 9, an allocation request for requesting “three instances” of size “5” is input from tenant D at time t4. Furthermore, an allocation request for requesting “four instances” of the size “4” is input from the tenant E at time t5. As a result, as shown in FIG. 14A, the management IDs “R # 4-1” to “R # 4-3” are placed on the physical machines 73 with the machine IDs “M # 1” to “M # 4”. Seven instances of “R # 5-1” to “R # 5-4” are arranged.

In FIG. 14B, the above-mentioned “20” instances are all arranged on 10 physical machines.

Suppose that a timer interrupt occurs after a preset time has elapsed. In response to this, the management apparatus 1 first performs additional adjustment of the instance. For example, on the physical machine 83 with the machine ID “M # 16”, an instance having a size of “6” or less can be additionally arranged. On the other hand, among the instances in the sequential placement area 7, when placed on the physical machine 83 with the machine ID “M # 16”, it is the instance of the size “5” that can reduce the waste of resources most. Therefore, in the sequential arrangement area 7, the instance of the management ID “R # 4-1” at the head of the instance of the size “5” is changed to the machine ID “M #” in the optimum arrangement area 8, as shown in FIG. 16 ”on the physical machine 83.

In FIG. 15B, no additional adjustment is performed, and therefore, the state is the same as FIG. 14B.

Next, as shown in FIG. 16A, the management apparatus 1 places all the instances sequentially arranged in the arrangement area 7 on the physical machine 83 in the optimum arrangement area 8 after the additional adjustment. Moving. As a result of this movement, no instance exists in the sequential arrangement area 7 as shown by the dotted line in FIG. The power of the physical machines 73 in the sequential arrangement area 7 remains on. On the other hand, prior to this movement, the physical machines 83 with the machine IDs “M # 17” to “M # 19” are newly powered on, but the other physical machines 83 of the optimal placement area 8 are powered on. Not.

At this time, the management apparatus 1 optimally arranges the instances in the optimal arrangement area 8. As a result, as shown in FIG. 16A, it can be seen that only nine physical machines 83 are used to arrange all the above-described instances, and there is little waste of resources. For example, paying attention to the physical machines 83 with the machine IDs “M # 11” to “M # 19”, the surplus resource is “7”. Therefore, the resource utilization efficiency is 92% = (90−7) / 90.

On the other hand, in FIG. 16B, ten physical machines 83 are used to arrange all the above-described instances. Therefore, it can be seen that there is a lot of waste of resources. For example, paying attention to the 10 physical machines, the surplus resource is “17”. Therefore, the resource utilization efficiency is 83% = (100−17) / 100.

Further, focusing on power consumption, the power amount for nine physical machines is shown in FIG. 16A, whereas the power amount for ten physical machines is shown in FIG. 16A. Therefore, in the optimum arrangement area 8, the power consumption is reduced by about 10%. As described above, as the number of instances to be arranged increases, it is not a problem that the physical machines 73 in the arrangement area 7 are sequentially powered on.

DESCRIPTION OF SYMBOLS 1 Management apparatus 2 Sequential arrangement | positioning process part 3 Optimal arrangement | positioning process part 4 Judgment process part 5 Data management arrangement | positioning process part 6 Machine group 7 Sequential arrangement area 8 Optimal arrangement area 21 Request input part 22 Request extraction part 23 Sequential arrangement part 24 Sequential arrangement information Reading unit 31 Optimal arrangement unit 32 Optimal arrangement information creation unit 33 Optimal arrangement reading unit 34 Movement information creation unit 41 Timer 42 Instance number determination unit 43 Request analysis unit 44 Additional adjustment unit 51 Data management unit 52 Access unit 53 Request information storage unit 54 Machine information storage unit 55 Arrangement information storage unit 56 Movement information storage unit 57 Deployment unit 58

Movement deployment unit

73, 83 Physical machine

Claims

In response to the allocation request, a process of arranging a virtual server on the plurality of first physical machines provided in the first area;
Processing for extracting request information necessary for allocation from the allocation request and recording it in the request information storage unit;
Processing for obtaining request information of a plurality of the virtual servers arranged on the plurality of first physical machines;
And a process of moving the virtual server arranged on the plurality of first physical machines to the plurality of second physical machines provided in the second area based on the request information. A method for managing an information processing system.
When resources allocated to the plurality of virtual servers arranged on the plurality of first physical machines exceed a preset threshold, a plurality of pieces arranged on the plurality of first physical machines The information processing system management method according to claim 1, wherein the virtual server is moved onto the plurality of second physical machines.
When a preset time has elapsed from the previous movement, the plurality of virtual servers arranged on the plurality of first physical machines are moved onto the plurality of second physical machines. The information processing system management method according to claim 1.
The allocation request specifies the size and quantity of the virtual server, and based on the size and quantity of the virtual server, the virtual server uses the plurality of first physical machines or the plurality of second servers. The information processing system management method according to claim 1, wherein the information processing system is disposed on a physical machine.
In the plurality of second physical machines, only the second physical machine that is the migration destination of the plurality of virtual servers arranged on the plurality of first physical machines is powered on prior to the migration. The information processing system management method according to claim 1, wherein:
A first area comprising a plurality of first physical machines and a virtual server being arranged on the plurality of first physical machines in response to an allocation request;
A second area comprising a plurality of second physical machines, wherein a virtual server is arranged on the plurality of second physical machines;
Processing for extracting request information necessary for allocation from the allocation request and recording it in the request information storage unit;
Based on the request information of the plurality of virtual servers arranged on the plurality of first physical machines, the virtual server arranged on the plurality of first physical machines is provided in a second area. An information processing system comprising: a management device that moves onto the plurality of second physical machines.
A recording medium recording a management program for an information processing system,
The management program is stored in a computer.
In response to the allocation request, a process of arranging a virtual server on the plurality of first physical machines provided in the first area;
Processing for extracting request information necessary for allocation from the allocation request;
Processing for obtaining request information of a plurality of the virtual servers arranged on the plurality of first physical machines;
A process of moving the virtual server arranged on the plurality of first physical machines to the plurality of second physical machines provided in the second area based on the request information. A recording medium characterized by the above.
An information processing system management program,
On the computer,
In response to the allocation request, a process of arranging a virtual server on the plurality of first physical machines provided in the first area;
Processing for extracting request information necessary for allocation from the allocation request;
Processing for obtaining request information of a plurality of the virtual servers arranged on the plurality of first physical machines;
Executing the processing of moving the virtual server arranged on the plurality of first physical machines to the plurality of second physical machines provided in the second area based on the request information. An information processing system management program characterized by