JP6511005B2 - Compute resource management system and compute resource management method - Google Patents

Description

  The present invention relates to a compute resource management system and a compute resource management method for managing compute resources.

  In recent career services and financial services, etc., a high quality that can continue the service even in severe disasters is required. Therefore, it has been practiced to distribute and arrange a plurality of compute resources in a plurality of areas. Here, a compute resource is a physical device having an information processing capability composed of a CPU (Central Processing Unit), a memory, etc., a virtual device built on these physical devices, etc., and can operate various services. . Hereinafter, compute resources are simply referred to as "resources". This resource is roughly classified into "vacant resource", "strongly available corresponding resource in use", and "strongly non-correspondence non-used resource" according to service operation existence and characteristics.

An available resource is a resource in which a service is not operating. Intended response In-use resources are resources that the service is operating and need to be rescued in the event of a catastrophic disaster. Unresponsiveness In-use resources are resources that the service is operating and there is no need to rescue this service in the event of a catastrophe. The service provider decides in advance whether it is necessary to rescue a service operating on each in-use resource in the event of a catastrophe.
If all resources in one area become unavailable due to a catastrophe etc., move the service that was operating on the busy resource of this area to a vacant resource of another area, and this busy resource It has been done to rescue the service that was operating. This makes it possible to continue the service even in severe disasters. In addition, the area of the movement destination is one area due to operational restrictions and the like. This is because if services that were operating on multiple busy resources in one area are distributed to free resources in multiple areas and the network is set, recovery time will become unacceptably large. is there.

  There is OpenStack as a technology to link such multiple resources. OpenStack is software that can operate a large number of server computers in an integrated manner to construct an information infrastructure. According to OpenStack, you can build a private cloud for operating your company system, or build a public cloud to provide cloud services to customers.

Non-Patent Document 1 describes a filter of OpenStack. The filter scheduler supports filtering and weighting to make informed decisions at where new instances are created.
Non-Patent Document 2 describes an invention of reserving a computation host in OpenStack.

“Filter Scheduler”, [online], OpenStack Foundation, [search April 5, 2016], <URL: http://docs.openstack.org/developer/nova/filter_scheduler.html> “Blazer”, [online], Wikipedia, [search on April 5, 2016], <URL: https://wiki.openstack.org/wiki/Blazar>

When newly available resources are used to expand services, etc., existing technologies such as OpenStack do not take into consideration the resources required in the event of a catastrophe. Therefore, there is a risk that resources prepared at the time of severe outbreak can not be secured.
Moreover, when using a free resource newly for service expansion etc., the availability of a resource changes with the presence or absence of the response to severe disaster. Therefore, in order to minimize the "resource securing at the time of a severe storm", it is necessary to make the availability judgment taking into consideration the presence or absence of the severe response which is the service characteristic.
Therefore, the present invention is to provide a compute resource management system and a compute resource management method capable of appropriately securing resources necessary for responding to a severe disaster occurrence and having service continuity in a severe disaster occurrence. It will be an issue.

  In order to solve the above-mentioned subject, in the invention according to claim 1, it is a compute resource management system which manages a plurality of areas each provided with a resource which can operate a service, and the resource is the stop of the service by severe disaster. In the case where a service is added to a vacant resource possessed by one area among the plurality of areas, including a rescue resource required to move the service to another area at the same time, and a vacant resource not operating the service. Among the plurality of areas, an area including the largest number of salvage resources and the number of the salvage resources included in the area are specified, and the number of other areas having vacant resources more than the number of the salvage resources is two or more Then, the judging means for judging that it is possible to add a service to the free resource possessed by the one area, It was compute resource management system according to claim comprising.

  By doing this, it is possible to appropriately secure the resources necessary for responding to a catastrophe, and to provide service continuity in the case of a catastrophe.

  In the second aspect of the present invention, the determination means includes, among the plurality of areas, an area including the largest number of required repair resources among the plurality of areas when a service is added to the vacant resources of the one area. Identify the number of the required rescue resources included in the area, and determine that the number of other areas having free resources greater than or equal to the number of the required rescue resources is 1 and the number of available rescue resources included in the other areas The computing resource management system according to claim 1, wherein when there is an area having a resource, it is determined that a service can be added to the free resource of the one area.

  By doing this, it is possible to appropriately secure resources necessary for responding to a catastrophe, and to achieve both service continuity and effective use of resources in a catastrophe.

  In the invention according to claim 3, when there is an area not including vacant resources among the plurality of areas, the judging means judges that the service can not be added to the vacant resources possessed by the one area. It was set as the compute resource management system of Claim 1 or 2 characterized by the above-mentioned.

  By doing this, it is possible to reserve free resources in each area and make a reserve at the time of failure.

  The invention according to claim 4 is characterized in that it comprises storage means for storing the number of the necessary rescue resources which each of the plurality of areas has, and the number of the free resources which each of the plurality of areas have. It is set as the compute resource management system of any one of claim | item 1 thru | or 3.

  By doing this, it is possible to determine whether or not to add a service in a short time while suppressing communication between systems.

  In the invention according to claim 5, when the determination means determines that the service can be added to the free resource possessed by the one area, the additional means adds the service to the free resource possessed by the one area. The compute resource management system according to any one of claims 1 to 4, characterized in that:

  By doing this, it is possible to actually add services.

In the invention according to claim 6 , the computer is a free resource in which the service is not operating and a resource in which the service is operating, and when the service is stopped due to a severe disaster, it is necessary to move the service to another area. A compute resource management method for managing a plurality of areas each provided with a rescue resource, wherein the computer receives an instruction to add a service to a vacant resource possessed by one of the plurality of areas, and the one area In the case where the service is added to the vacant resources possessed by: a step of specifying an area including the largest number of salvage resources among the plurality of areas and the number of the salvage resources included in the area; If the number of other areas with more free resources is 2 or more And a compute resource management method characterized by comprising the steps of: determining a possible additional services to the free resource that the one area has.

  By doing this, it is possible to appropriately secure the resources necessary for responding to a catastrophe, and to provide service continuity in the case of a catastrophe.

In the invention according to claim 7 , the computer has one or more free resources whose number of other areas having free resources equal to or more than the number of required rescue resources is one and the number of required relief resources included in the other areas. The computer resource management method according to claim 6 , further comprising the step of determining that a service can be added to the free resource of the one area when the area exists.

  By doing this, it is possible to appropriately secure resources necessary for responding to a catastrophe, and to achieve both service continuity and effective use of resources in a catastrophe.

  According to the present invention, it is possible to appropriately secure resources necessary for responding to a severe disaster and to provide service continuity in a severe disaster.

It is a block diagram of the compute resource management system in this embodiment. It is a flowchart which shows the resource expansion process in this embodiment. It is a figure explaining each area before updating and after updating. It is a block diagram of the compute resource management system of a comparative example. It is a flowchart which shows the resource expansion process of a comparative example. It is a flowchart which shows the resource movement process at the time of severe disaster. It is a figure explaining the state of each area before disaster occurrence, and the state of each area which moved the resource after disaster occurrence. It is a figure explaining the state of each area before and behind disaster occurrence.

  Hereinafter, after a comparative example is demonstrated in detail with reference to each figure, the form for implementing this invention is demonstrated.

Comparative Example
FIG. 4 is a block diagram of the compute resource management system Sa of the comparative example.
The compute resource management system Sa of the comparative example includes the compute resource management apparatus 1A, the maintenance worker terminal 2 instructing this, and areas # 1 (3a), # 2 (3b), and # 3 (areas to be managed). 3c) is included. Hereinafter, the area # 1 (3a), the area # 2 (3b), and the area # 3 (3c) may be simply described as the area 3 when not distinguished from one another. The compute resource management system Sa is, for example, a telephone system, but is not limited thereto.

  The compute resource management device 1A is, for example, a device configured of a CPU, a memory, and the like, and the CPU executes a compute resource management program (not shown) to realize the compute resource operation unit 12. The compute resource operation unit 12 instructs the compute resource of each area 3 to add / drop / move. A part of the processing of the compute resource operation unit 12 will be described with reference to FIG. 5 described later.

  The compute resource information 13 is storage means for storing resource information of each area 3 respectively. The compute resource information 13 will be described in detail with reference to FIGS. 7 and 8 described later.

Area 3 is a set of compute resources in which the service can operate, and a range in which the set of compute resources can not be used simultaneously at the time of a catastrophe. Hereinafter, compute resources are simply referred to as resources. The resources can be divided into free resources 31, severely used and in-use resources 32, and severely non-corresponded and non-used resources 33.
The free resource 31 refers to a resource in which the service is not operating. Severe response available In-use resource 32 is a resource on which a service is operating, and means a resource that needs to be rescued in the event of a catastrophe (required resource required). In-service response In-use resource 33 is a resource in which the service is operating and does not need to be relieved in the event of a catastrophe.

  In FIG. 4, the area # 1 (3a) includes four free resources 31, one severe response corresponding in-use resource 32, and one severe response no-in-use resource 33. The area # 2 (3 b) includes two free resources 31, four intensely supported in-use resources 32, and one intense non-correspondence in-use resource 33. The area # 3 (3 c) includes three free resources 31, one intensely supported in-use resource 32, and two intense non-correspondence in-use resources 33.

FIG. 5 is a flow chart showing resource addition processing of the comparative example.
For example, upon receiving an instruction to add a resource from the maintenance worker terminal 2, the compute resource operation unit 12 starts a series of resource expansion processing.
First, the compute resource operation unit 12 issues an expansion instruction to each area 3 (step S20), reflects the expansion instruction result on the compute resource information 13 (step S21), and ends the processing of FIG.

FIG. 6 is a flowchart showing resource migration processing at the time of a catastrophe disaster.
When all resources in one area 3 become unusable due to a catastrophe or the like, the service is moved and operated on the vacant resources 31 in the other areas 3. Specifically, the compute resource operation unit 12 specifies one area 3 which has received a severe disaster and the like, and the number of in-use response resources 32 which the area 3 has (step S30). Then, the compute resource operation unit 12 determines the number of other areas 3 having the vacant resources 31 equal to or more than the specified number (step S31).

  If the number of other areas 3 is one or more (step S31: 1 or more), the compute resource operation unit 12 selects one of the other areas 3 and severes the area 3 which has suffered a severe disaster etc. There is a response The service used in the in-use resource 32 is moved (step S32). As a result, even in the event of a catastrophe, it is possible to salvage and continue the service operating on the in-use resource 32 with the catastrophic response.

FIGS. 7A to 7C are diagrams for explaining the state of each area 3 before the occurrence of a disaster and the state of each area 3 to which resources have been moved after the occurrence of a disaster. FIGS. 7A to 7C show the compute resource information 13.
FIG. 7A shows the state of each area 3 before the occurrence of a disaster, which is the same as that shown in FIG.

FIG. 7B shows the state of each area 3 after the occurrence of a disaster in area # 2 (3b).
In the area # 2 (3b), the vacant resource 31, the busy resource 32 in use, and the busy resource 33 in use are all changed to 0 due to the occurrence of the catastrophe. The service operating on the four busy / in-use resources 32 in area # 2 (3b) moves to area # 1 (3a).
As a result, the free resources 31 in the area # 1 (3a) change from four to zero and the in-use resources 32 in use from one to five. Other than this, it is the same as the state shown in FIG. 7 (a).
As in the case of FIG. 7A, area # 3 (3c) also has three free resources 31, one severe response available in-use resource 32, and two severe severity no-in-use resources 33 and Contains.

FIG. 7C shows the state of each area 3 after a disaster occurs in area # 1 (3a) from the state before the disaster shown in FIG. 7A.
In the area # 1 (3a), the free resource 31, the busy resource 32 in use, and the busy resource 33 in use are all changed to 0 due to the occurrence of the catastrophe. One intense response available resource 32 in area # 1 (3a) moves to area # 3 (3c).
The free resources 31 in the area # 3 (3c) change from three to two, and the heavy duty support in-use resource 32 changes from one to two. The area # 1 (3a) further has two intense non-correspondence in-use resources 33.
As in FIG. 7A, area # 2 (3 b) includes two free resources 31, four intense support available resources 32, and one intense support non-use resources 33. Contains.

FIGS. 8A and 8B are diagrams for explaining the state of each area 3 before and after the occurrence of a disaster. FIGS. 8A and 8B show the compute resource information 13.
FIG. 8A shows the state of each area 3 before the occurrence of a disaster.
Area # 1 (3a) differs from FIG. 7 (a) and includes two free resources 31, one intense support available resource 32 and three intense support non-use resources 33. It is. That is, in the state shown in FIG. 7A, when two in-use non-correspondence in-use resources 33 are added, the state shown in FIG. 8A is obtained.
Area # 2 (3 b) includes two free resources 31, four intensely compatible busy resources 32, and one intense busy no busy resource 33, as in FIG. 7A. It is. The area # 3 (3 c) includes three free resources 31, one intensely supported in-use resource 32, and two intense non-correspondence in-use resources 33.

FIG. 8B shows the state of each area 3 after the occurrence of a disaster in area # 2 (3b).
In the area # 2 (3b), the vacant resource 31, the busy resource 32 in use, and the busy resource 33 in use are all changed to 0 due to the occurrence of the catastrophe. However, the service that was operating on the 4 hard-correspondence-used resources 32 in area # 2 (3b) can not move to area # 1 (3a) or to any of area # 3 (3c) . Therefore, the service operating on the resource # 2 (3b) and the resource # 2 in use in the area # 2 (3b) is interrupted.
As described above, in the comparative example, as a result of the expansion of the service, an event occurs in which the service is interrupted without being able to be rescued when a severe disaster occurs.

<< this embodiment >>
FIG. 1 is a block diagram of a compute resource management system S in the present embodiment.
The compute resource management system S according to the present embodiment includes the compute resource management apparatus 1, the maintenance worker terminal 2 instructing this, and areas # 1 (3a), # 2 (3b), and # 3 that are management targets. It comprises (3c).

  The compute resource management device 1 is, for example, a device including a CPU and a memory, and the CPU executes a compute resource management program (not shown) to realize the expansion determination unit 11 and the compute resource operation unit 12. Be done. The expansion determining unit 11 determines whether the resource of each area 3 can be expanded. The compute resource operation unit 12 instructs the resources of each area 3 to be added / removed / moved, and the like. A part of the processing of the expansion determination unit 11 and the compute resource operation unit 12 will be described with reference to FIG. 2 described later.

  The compute resource information 13 is storage means for storing resource information of each area 3 respectively. Based on the compute resource information 13, the expansion determination unit 11 can know the state of each resource without inquiring about the resource that each area 3 has. The compute resource information 13 will be described in detail in FIG. 3 to be described later.

Area 3 is, like the comparative example, a set of compute resources in which the service can operate, and a range in which the set of resources can not be used simultaneously at the time of a catastrophe. It is also the same as that of the comparative example that the resources are divided into the vacant resource 31, the busy resource 32 in use, and the busy resource 3 not in use.
The free resource 31 refers to a resource in which the service is not operating. Severe response available In-use resource 32 is a resource on which a service is operating, and means a resource that needs to be rescued in the event of a catastrophe (required resource required). In-service response In-use resource 33 is a resource in which the service is operating and does not need to be relieved in the event of a catastrophe.

FIG. 2 is a flowchart showing resource addition processing in the present embodiment.
For example, upon receiving an instruction to add a resource from the maintainer terminal 2, the addition determination unit 11 and the compute resource operation unit 12 start a series of resource expansion processing.
First, the expansion judgment unit 11 copies the data of the compute resource information 13 to a work area such as a random access memory (RAM) not shown (step S10), and reflects the contents of the expansion instruction in this work area (step S11). ). Next, the expansion determining unit 11 determines the number of areas 3 in which there are no free resources 31 (step S12). If the number of areas 3 having no vacant resource 31 is one or more (step S12: 1 or more), the expansion judgment unit 11 responds an error to the maintenance worker terminal 2 (step S18), and ends the processing of FIG. . Thereby, the addition determination unit 11 can prevent the state in which each area 3 does not have the free resource 31. As a result, even if any one of the severely corresponding in-use resources 32 and non-hardly-in-use resources 33 included in each area 3 is stopped due to a failure or the like, the service operating on this resource is available as empty resources 31 Can be salvaged.

  If the number of areas 3 having no vacant resources 31 is 0 (step S12: 0), the expansion judgment unit 11 specifies the area 3 where the number of in-use resources 32 is the largest and the number (S12: 0) Step S13).

  Further, the expansion determining unit 11 determines the number of other areas 3 having vacant resources 31 equal to or more than the number of the in-use resources 32 specified and determined as the severe defect correspondence (step S14). If the number of other areas 3 is 0, the expansion judgment unit 11 responds an error to the maintenance person terminal 2 (step S18), and ends the process of FIG. If the number of other areas 3 is two or more (step S14: two or more), the expansion judgment unit 11 instructs the compute resource operation unit 12 to perform expansion with respect to one of them (step S17).

If the number of other areas 3 is 1 (step S14: 1), the expansion judgment unit 11 specifies the area 3 where the vacant resource 31 is largest and the number of in-use corresponding resources 32 in the area 3 (Step S15). Further, the expansion determining unit 11 determines the number of other areas 3 having vacant resources 31 equal to or more than the number of in-use resources 32 specified and determined (step S16). If the number of other areas 3 is zero, the expansion judgment unit 11 responds an error to the maintenance worker terminal 2 (step S18) and ends the processing of FIG. 2 and if the number of other areas 3 is one or more , And instructs the compute resource operation unit 12 to add (step S17).
After the expansion instruction in step S17, the expansion determination unit 11 reflects the expansion instruction result reflected in the work area on the compute resource information 13 (step S19), and ends the processing of FIG.

FIGS. 3A and 3B are diagrams for explaining the areas 3 before and after the update. FIGS. 3A and 3B show work areas in which the compute resource information 13 is reflected. In the following description, FIGS. 1 and 2 will be referred to as appropriate.
FIG. 3A shows an example of each area 3 before updating, which is similar to the state of FIG. 1 described above.
Assuming that the work area to which the content of the expansion instruction has been reflected in step S11 of FIG. 2 is in the state shown in FIG. 3A, the expansion determination process of step S12 and subsequent steps of FIG. 2 is performed. Then, in step S13, the area # 2 (3b) in which the number of in-use resources 32 used is the largest is identified, and the number of in-progress corresponding resources 32 is specified as four. Thereafter, in step S14, in order to move the four severe response available in-use resources 32, area 3 having four or more vacant resources 31 is identified as area # 1 (3a), and area 3 conforming thereto is It is one.

  Next, the expansion judgment unit 11 specifies the area 3 where the vacant resource 31 is the largest as the area # 1 (3a), and specifies the number of in-use correspondence resources 32 in the area # 1 (3a) as one. (Step S15). Since the expansion determination unit 11 determines that the number of other areas 3 having one or more vacant resources 31 is two, the compute resource operation unit 12 can issue an expansion instruction (step S17).

  At this time, even if any one of area # 1 (3a), area # 2 (3b), and area # 3 (3c) is stopped due to a severe disaster, there is a severe response corresponding use resource 32 in the stopped area 3 has. The service which has been operating can be moved to the free resource 31 of another area 3.

FIG. 3B shows an example of each area 3 after the update. At step S11, the work area to which the extension instruction content is reflected is in the state shown in FIG. 3 (b).
Here, since two intense non-correspondence in-use resources 33 are added to the area # 1 (3a), the number of in-flight non-correspondence in-use resources 33 is three, and the number of vacant resources 31 is two. When expansion determination processing in step S12 and subsequent steps in FIG. 2 is performed on this work area, area # 2 (3b) in which the number of in-use resources 32 is maximum is identified in step S13 and the in-progress correspondence is determined. The number of in-use resources 32 is specified as four. However, although an attempt is made to specify the area 3 having four or more vacant resources 31 in step S14, there is no area 3 conforming to this, and it is zero. Therefore, the addition determination unit 11 sends an error response to the maintainer terminal 2 (step S18).

Since the process ends after step S18, the actual state of each area 3 remains as it was before the update shown in FIG. 3 (a). At this time, even if any one of area # 1 (3a), area # 2 (3b), and area # 3 (3c) is stopped due to a severe disaster, there is a severe response corresponding use resource 32 in the stopped area 3 has. The service which has been operating can be moved to the free resource 31 of another area 3.
As described above, in the present embodiment, the service expansion is performed after it is determined whether or not the service can be rescued when the severe disaster occurs. Therefore, it is possible to secure minimum necessary resources for responding to the occurrence of a catastrophe, and to achieve both service continuity and effective use of resources in the case of a catastrophe.

(Modification)
The present invention is not limited to the above embodiment, and modifications can be made without departing from the spirit of the present invention, and there are, for example, the following (a) to (d).
(A) The number of resources is not limited to three.
(B) Each flowchart is an example, and another step may be inserted in each step, and is not limited.
(C) Each resource may be a physical device or may be a virtual machine configured on the physical device.
(D) The addition determination unit 11 receives an instruction to add a resource is not limited to the instruction from the maintainer terminal 2, and may be an instruction from a monitoring device on which, for example, Zabbix or Celiometer is executed, and is not limited.

S, Sa Compute Resource Management System 1, 1A Compute Resource Management Device 11 Addition Judgment Unit (Judgment Means)
12 Compute Resource Operation Unit (Additional means)
13 Compute resource information (storage means)
2 maintainer terminal 3a area # 1
3b area # 2
3c area # 3
31 Free resource 32 Hard response available In-use resource (relief resource required)
33 Extremely No Response In Use Resources

Claims (7)

A compute resource management system for managing a plurality of areas each having a resource in which a service can operate, comprising:
The resources include relief resources that require the service to move to another area when the service is stopped due to a severe disaster, and free resources in which the service is not operating,
In the case where a service is added to a vacant resource included in one of the plurality of areas, an area including the largest number of required rescue resources among the plurality of areas and the number of the required rescue resources included in the area are specified. Determining means for determining that it is possible to add a service to the vacant resource possessed by the one area if the number of other areas having vacant resource more than the number of the rescue resources required is two or more,
A compute resource management system comprising: The judging means
Among the plurality of areas, when a service is added to a vacant resource included in the one area, an area including the largest number of necessary rescue resources among the plurality of areas and the number of the required rescue resources included in the area are specified If there is an area having one or more other resources having vacant resources equal to or more than the number of salvage required resources and one or more vacant resources equal to or more than the number of salvaged resources included in the other area, the one area Judging that it is possible to add services to the free resources it has,
The compute resource management system according to claim 1, characterized in that: The judging means
If there is an area that does not include free resources among the plurality of areas, it is determined that the service can not be added to the free resources possessed by the one area,
The compute resource management system according to claim 1 or 2, characterized in that: Storage means for storing the number of the required rescue resources in each of the plurality of areas, and the number of the free resources in each of the plurality of areas;
The compute resource management system according to any one of claims 1 to 3, comprising: Extension means for adding a service to the free resource possessed by the one area if the judging means judges that the service can be added to the free resource possessed by the one area;
The compute resource management system according to any one of claims 1 to 4, comprising: A plurality of areas, each having a free resource where the service is not running, and a resource in which the service is running, and which need to be moved to another area of the service when the service is stopped due to a severe disaster It is a compute resource management method to manage, and
The computer
Accepting an instruction to add a service to a vacant resource included in one of the plurality of areas;
Specifying an area including the largest number of required rescue resources among the plurality of areas when a service is added to a vacant resource included in the one area, and the number of the required rescue resources included in the area;
Determining that it is possible to add a service to the free resource of the one area if the number of other areas having free resources equal to or more than the number of rescue resources required is 2 or more;
A compute resource management method comprising: The computer
When there is an area having one or more other resources having vacant resources more than the number of rescue resources required and the number of vacant resources greater than or equal to the number of rescue resources contained in the other area, the vacant region included in one area Determining that services can be added to resources,
The method of claim 6 , further comprising:

Images