WO2013029487A1 - 一种资源分配方法及资源管理平台 - Google Patents

一种资源分配方法及资源管理平台 Download PDF

Info

Publication number
WO2013029487A1
WO2013029487A1 PCT/CN2012/080467 CN2012080467W WO2013029487A1 WO 2013029487 A1 WO2013029487 A1 WO 2013029487A1 CN 2012080467 W CN2012080467 W CN 2012080467W WO 2013029487 A1 WO2013029487 A1 WO 2013029487A1
Authority
WO
WIPO (PCT)
Prior art keywords
resource
host
application
allocation
resources
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2012/080467
Other languages
English (en)
French (fr)
Inventor
柴晓前
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huawei Technologies Co Ltd
Original Assignee
Huawei Technologies Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huawei Technologies Co Ltd filed Critical Huawei Technologies Co Ltd
Priority to EP12826692.1A priority Critical patent/EP2725862A4/en
Publication of WO2013029487A1 publication Critical patent/WO2013029487A1/zh
Priority to US14/184,342 priority patent/US9442763B2/en
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/50Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/50Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals

Definitions

  • the invention belongs to the technical field of computers, and in particular relates to a resource allocation method and a resource management platform. Background technique
  • Cloud computing can be allocated on demand according to the requirements of user resources, that is, when the service needs resources, and when resources are not needed, when there are multiple peak-peak services, the utilization efficiency of each part of the system can be improved as a whole.
  • This feature of cloud computing has effectively reduced the hardware and management costs of enterprise operations, improved the utilization of hardware resources, and greatly improved cloud computing technology.
  • VM Virtual Machine
  • the resource may be applied to the resource management platform through the interface exposed by the resource management platform.
  • the resource provides a resource capability requirement parameter, where the requirement parameters include: VM CPU, memory, hard disk, and network capability. , VM location requirements, VM image (Image), etc.
  • the demand parameters mainly refer to some technical performance parameter requirements that the resource management platform needs to meet, and then the resource management platform provides resources to the user according to the user's demand parameters.
  • the isolation performance is different depending on the virtualization technology.
  • This isolation is mainly the interaction between VMs, such as a VM.
  • the running application enters an infinite loop state, which may cause the VM to crash, but the VM's crash does not affect the operation of other VMs.
  • the host in the present invention refers to a machine running a VM, which may be a physical machine or a VM.
  • the VM shares hardware resources, the characteristics of the hardware resources themselves may cause competition of hardware resources between the VMs, for example: read and write of the hard disk, due to the speed of the physical hard disk, read and write speed, etc., multiple virtual Frequently reading and writing hard disks together will affect each other.
  • the applications running on each VM may have resource contention, the performance of each VM cannot be guaranteed, and the physical machine cannot perform optimally.
  • a physical machine has 2 VMs, and the applications run by these two VMs are applications that frequently read and write hard disks, such as data processed by hard disk cache or copying and changing of hard disk data. At this time, applications on these two VMs will cause VMs.
  • VMs with homogeneous requirements for hardware resources have the same or similar requirements for hardware resources, for example, both VMs frequently read and write hard disks
  • long-term operation will be The hardware of the physical machine is in a high load state for a long time, while other hardware is in a low load state for a long time, so that the overall performance of the physical machine is unbalanced, and energy consumption and hardware loss are increased.
  • the purpose of the embodiments of the present invention is to provide a resource allocation method, which aims to solve the problem that the overall performance of the host is degraded due to the competition of hardware resources generated by multiple application programs sharing hardware resources.
  • the embodiment of the present invention is implemented as a resource allocation method, where the method includes the following steps: receiving a resource request sent by a resource requester, where the resource request includes a resource requirement and an application characteristic of the resource;
  • Another object of the present invention is to provide a resource management platform, where the resource management platform includes:
  • a resource request receiving unit configured to receive a resource request sent by a resource requester, where the resource request includes a resource requirement and an application characteristic of the resource;
  • a host determining unit configured to determine, according to the resource request and the resource application attribute allocation policy and the resource application characteristic combination policy, a host that allocates resources to the resource requester;
  • an allocation information returning unit configured to control the host to allocate resources to the resource requester, and return resource allocation information to the resource requester.
  • the embodiment of the present invention receives a resource request including a resource requirement and an application characteristic of a resource requester, determines a host that allocates a resource to the resource requester according to the resource request, and a preset resource application characteristic allocation policy, and controls the host machine. Allocating resources to the resource requester, returning the host to the resource requester to determine the resource allocation, and solving the competition in the prior art that the hardware resources are generated by the shared hardware resources of multiple applications, causing the performance of the host and the allocated resources themselves.
  • the problem of the decline is to allocate resources according to the resource request and the resource application characteristic allocation strategy, optimize the resource allocation, effectively improve the resource utilization efficiency, reduce the hardware loss, and improve the user experience.
  • FIG. 1 is a flowchart of implementing a resource allocation method according to Embodiment 1 of the present invention
  • FIG. 2 is a flowchart of implementing a resource allocation method according to Embodiment 2 of the present invention.
  • FIG. 3 is a flowchart of implementing a resource allocation method according to Embodiment 3 of the present invention.
  • FIG. 4 is a schematic structural diagram of a resource management platform according to Embodiment 5 of the present invention.
  • FIG. 5 is a schematic structural diagram of a resource management platform according to Embodiment 6 of the present invention.
  • FIG. 6 is a schematic structural diagram of a resource management platform according to Embodiment 7 of the present invention. detailed description
  • Embodiment 1 is a diagrammatic representation of Embodiment 1:
  • FIG. 1 is a flowchart showing an implementation process of a resource allocation method according to Embodiment 1 of the present invention. Details are as follows: In step S101, a resource request sent by a resource requester is received, where the resource request includes a resource requirement and an application characteristic of the resource.
  • the resource request may be received through a preset interface, so that the interface may receive an application feature of the resource, where the preset interface may be a RESTful interface, a web service (WebService) interface, or
  • the resource mainly refers to a virtual machine (VM) with a central processing unit (CPU), memory, hard disk, network, etc., or a computing power resource, such as a Java virtual machine running on an operating system (JVM) ) or computing agents.
  • the resource request is initiated by the resource requester.
  • the resource requester may be any user requesting resources from the resource management platform.
  • the resource requester also includes a program, such as: a scheduling module of the distributed computing system. .
  • the resource requirement included in the resource request refers to a resource performance parameter requirement requested by the user, that is, some capability requirements that the provided resource needs to meet, which is referred to as a resource specification in the present invention, for example, when requested
  • the resource specification may be one of the host location of the resource, the CPU of the virtual machine, the memory, the disk, the network, the operating system, or an application, or any combination thereof, where the location of the host where the resource is located may be a designated sink. Which location should the host be, or what requirements are required for the location, such as multiple resources being applied to the same host, as shown in Table 1.
  • the application characteristic of the resource refers to the type of the application running on the resource
  • the type of the application running on the resource may be defined in advance by the resource provider.
  • the resource requester selects, from the types of applications running on the resource defined by the resource provider, the type of the program that is most suitable for the application to which the application is to be applied, or the application characteristic of the resource refers to the resource requester.
  • the usage of the resources to be applied including but not limited to one or any combination of the following: CPU usage, memory usage, disk read and write, network throughput (up and down), high real-time, graphics processor (Graphic Processing Unit, GPU) usage, etc., which defines the usage load of the capabilities of the resource, which is a synthesis of the characteristics of all applications running on the requested resource. For example, if a virtual machine runs an application and the application is frequently read and written by the hard disk, the hard disk is frequently read and written by the hard disk. In the present invention, the application feature of the resource is that the hard disk is frequently read and written.
  • the overall performance of the two applications is a high network load, which is referred to in the present invention.
  • the application feature is high network load. Specific examples are shown in Table 2.
  • Table 2 describes the application characteristics of resources in a resource request.
  • the CPU usage is high (usage >75%), memory, disk read and write, network throughput, real-time performance, etc., which can be applied according to system requirements.
  • the definition of a feature, such as memory usage ⁇ 10%, defines the application characteristics of the memory as low.
  • step S102 a host that allocates resources to the resource requester is determined according to the resource request and the resource application property allocation policy.
  • the resource application attribute allocation policy may be configured in advance on the resource management platform.
  • the resource management platform determines the host to allocate resources to the user according to the resource requirements, the application characteristics of the resources, and the resource application characteristics allocation policy.
  • the resource application characteristic allocation policy may specifically include a resource application characteristic allocation policy based on the regional allocation, and a resource application characteristic allocation policy based on the host allocation, wherein the resource allocation characteristic allocation policy based on the regional allocation refers to an application characteristic.
  • the resource application attribute allocation policy based on the host allocation refers to which host or resources are allocated by the resource of an application characteristic, and the area refers to a logical area or a physical area, a logical area.
  • the resource application feature combination strategy is used to define the application characteristics of the resources that can share the host machine, or the application characteristics of the resources of the host machine, or the mutual influence of the resources of the two application characteristics when sharing a host. That is, the coexistence degree, wherein the application characteristics of the resources that can share the host machine refer to which application characteristics resources can share the host machine, and the application characteristics of the resources that cannot share the host machine refer to which application feature resources cannot share the host machine.
  • a rule file can be used to save the resource application feature combination strategy. As shown in Table 3, an example of a resource application feature combination strategy is described.
  • Table 3 Type 1 resource type 2 resource coexistence
  • High CPU usage High network access 2 Describes the impact of different types of resources 1 and resource 2 on the performance of resource sharing on the resource performance, using coexistence, and dividing the impact factor into three levels.
  • the coexistence degree of the high disk read/write class resource 1 and the high disk read/write class resource 2 shared host is 0, the coexistence degree is low, and coexistence is not recommended.
  • the high network access class resource 1 The coexistence degree of the shared host with the high network access class resource 2 is 1, indicating that the impact is medium, the coexistence degree is the second highest, and can coexist to a certain extent.
  • the high CPU uses the class resource 1 and the high network access class resource. 2
  • the shared host has a coexistence of 2, which means less impact, higher coexistence, and coexistence.
  • step S103 the host that controls the allocation of resources allocates resources to the resource requester, and returns resource allocation information to the resource requester.
  • the control host after determining the host that allocates the resource to the resource requester, allocates the resource to the resource requester, and returns the resource allocation information to the resource requester, where the resource allocation information may be the identifier of the resource and/or Or the address of the resource and/or any other information that can locate or identify the resource.
  • the related information of the allocated resources is further saved as a basis for subsequent resource allocation or resource adjustment, wherein the resource adjustment may be resource migration.
  • the related information of the saved resource includes the application characteristics of the resource, and may also include: a location of the resource, a user of the resource, a resource allocation time, a resource validity period, a resource capability, a resource security policy, and a resource migration policy. one or more.
  • the embodiment of the present invention receives a resource request including an application feature of a resource requirement and a resource through a RESTful interface, a WebService interface, or a remote call interface, and determines a resource allocation request to the resource requester according to a preset resource application characteristic allocation policy preset by the resource management platform.
  • the host machine controls the host to allocate resources to the resource requester, thereby optimizing the resource allocation according to the application characteristics of the requested resource and the configured resource application characteristic combination strategy before the resource allocation, thereby effectively improving the resource utilization efficiency and improving the resource utilization efficiency. user experience.
  • Embodiment 2 is a diagrammatic representation of Embodiment 1:
  • FIG. 2 is a flowchart showing an implementation process of a resource allocation method according to Embodiment 2 of the present invention. Details are as follows:
  • a resource requester receives a resource request, where the resource request includes a resource requirement and an application characteristic of the resource.
  • step S202 an idle resource that satisfies the resource requirement is determined in the resource pool according to the resource requirement of the resource request, the application characteristic of the resource, and the configured resource application characteristic allocation policy.
  • the resource management platform determines, according to the resource requirement carried in the resource request and the idle resource in the resource pool, whether there is an idle resource that satisfies the request, and if so, Then, step 203 is performed on the idle resource.
  • the satisfaction specifically means that if the resource requirement is less than the number of idle resources in the current resource pool, and the maximum single resource available in the resource pool satisfies the maximum single resource requirement in the resource requirement, the resource requester is considered to be satisfied. The demand is otherwise considered to be unsatisfied.
  • a VM Since a VM must be located on a single physical machine, if the VM's specification requirements are 16-core CPU and 16G memory, the specifications of the largest single physical resource in the resource pool are configured as 8-core CPU and 8G memory. If the specification requirement of the VM is greater than the specification configuration of all the physical machines in the resource pool, the resource is considered to be unable to meet the resource requirement of the resource request, where a single physical machine refers to a physical machine.
  • a single resource is a resource that cannot be subdivided, such as a single server.
  • a resource requirement in a resource request has an area requirement, or a resource providing area is set in a resource authority control policy preset in the resource pool, or an application characteristic of the allocated area requested requirement and resource
  • the area corresponding to (specified) is located according to a certain search algorithm to determine an idle resource that satisfies the resource requirement. Otherwise, all the resource pools are located to determine an idle resource that satisfies the resource requirement. For example, if the resource requirement in the resource request has a regional requirement, or the resource authority control policy set by the resource provider defines that the resource requester is provided with resources in a specific area, or according to the The area is located to determine the idle resource that meets the resource requirement.
  • the resource application characteristic allocation policy of the resource provider herein defines a resource application characteristic allocation policy based on the area allocation, and/or a resource application characteristic allocation policy based on the host allocation.
  • the foregoing area is a logical area or a physical location, where the resource authority control policy set by the resource provider may be one or any combination of the following policies: a policy based on a resource requester level, and a resource category based on a resource requester subscription The policy, the allocation policy of the group in which the resource requester account is located, the regional policy of the resource provider (such as the location of the requester determined according to the requester IP), the resource failure response policy, the national legal restrictions, or other policies of the resource provider.
  • step S203 when there is an idle resource that satisfies the resource requirement, it is determined whether the host where the idle resource is located is a host that allocates resources.
  • the sex combination policy directly determines that the host where the idle resource is located is a host host that allocates resources to the resource requester. If the resource application feature combination policy is defined in the resource application feature allocation policy, the application feature of the allocated resource on the host where the idle resource is located, the application feature of the resource in the resource request, and the resource application feature combination policy are determined. Whether the host where the free resource is located is the host that allocates the resource.
  • step 202 is performed again. Next free resource.
  • the resource application feature combination policy defines an application feature that can share the resource of the host device or an application feature that cannot share the resource of the host device, directly determine the allocated resource according to the resource application feature combination policy.
  • the application characteristics and the application characteristics of the resources in the resource request can coexist.
  • the application attribute combination policy of the resource defines the coexistence degree of the application characteristics of the resource, Then, based on the degree of resource coexistence, it is judged whether or not it can coexist.
  • selecting a resource with a high degree of coexistence to allocate to the user according to the order of resource coexistence may include the following steps:
  • the resource search algorithm of the resource management platform finds an available host as an alternative host, where the resource search algorithm may be a search algorithm for balancing the load of each host, or centralized use. Search algorithms for resources (to save power for non-use resources), etc.;
  • step 3 After finding an alternate host that can allocate the requested resource, determine whether some resources have been allocated on the candidate host, and if so, perform step 3), otherwise perform step 4);
  • step 3 Re-execute steps 1) ⁇ 4), until the resource allocation is completed, or all available hosts are found, the resource allocation cannot be completed, then step 3) is performed on the candidate host found in step 2),
  • the coexistence requirement that is, the application feature of the allocated resource and the application characteristic of the resource in the resource request have the second highest degree of coexistence. If the suitable resource is still not found, the candidate host found in step 2) is further executed. Step 3), reduce the coexistence requirement, and so on, search according to the coexistence degree from high to low, until the resources satisfying the resource request are found and allocated, thereby avoiding resource competition and realizing the optimal allocation of resources. Improve the efficiency of resource allocation and efficiency of use.
  • step S204 the host that controls the allocation of resources allocates resources to the resource requester, and returns resource allocation information to the resource requester.
  • the related information of the allocated resources is saved as a basis for subsequent resource allocation or resource adjustment (such as migration).
  • the related information of the saved resource includes the application characteristics of the resource, and may also include: a location of the resource, a user of the resource, a resource allocation time, a resource validity period, a resource capability, a resource security policy, and a resource migration policy. one or more.
  • Embodiment 3 In the embodiment of the present invention, since the resource requirements input by the user and the application characteristics of the resource may have errors or are only input in the system default manner, or when the resource is initially allocated to the user, there is no more preferable alternative host, which results in allocation to the user.
  • the host where the resource is located is not optimal.
  • the system default mode input means that the user has no input.
  • the resource management platform increases the collection function of the resource application characteristics to determine whether the resource needs to be adjusted.
  • the collection function of the resource application feature refers to the resource management platform running. Application characteristics derived from monitoring information.
  • FIG. 3 is a flowchart showing an implementation process of a resource allocation method according to Embodiment 3 of the present invention. Details are as follows: In step S301, a resource application feature allocation policy is pre-configured.
  • the resource application feature allocation policy specifically includes a resource allocation feature allocation policy based on the region allocation, a resource application feature allocation policy based on the host device allocation, and a resource application feature allocation policy based on the resource application feature combination policy. Or any combination thereof.
  • the resource application characteristic allocation policy based on the area allocation refers to which area or areas of resources of an application characteristic are allocated
  • the resource application attribute allocation policy based on the host allocation refers to which host or resources of the resource of an application characteristic are located.
  • An allocation, an area refers to a logical area or a physical area.
  • a logical area is a logical grouping of one or more physical resources into a group, which has the same management rules or security policies.
  • step S302 a resource request sent by a resource requester is received, where the resource request includes a resource requirement and an application characteristic of the resource.
  • step S303 a host that allocates resources to the resource requester is determined according to the resource request and the resource application attribute allocation policy.
  • step S304 the host that controls the allocation of resources allocates resources to the resource requester, and returns resource allocation information to the resource requester.
  • step S305 the usage information of the resource is collected according to a preset policy, and the application characteristic of the resource is calculated according to the usage information of the resource.
  • the resource requirement input by the user and the application characteristic of the resource may exist.
  • the error is either entered in the system default mode, or there is no more preferred alternate host when resources are initially allocated to the user, resulting in a host that is not optimal for the resources allocated to the user. Therefore, the usage information of the resource is collected according to the preset policy, and the application characteristic of the resource is calculated according to the usage information of the resource, and the current actual application characteristic of the resource is obtained, where the resource usage information may be hardware usage information.
  • the calculation method of the application characteristics of the resource may be: If the CPU usage information of the collected resource is analyzed, and the load is greater than 75% in 80% of the time, the application characteristic of the resource is high CPU usage.
  • the example given here is the CPU.
  • the application characteristics of the resource are high hard disk read and write, and network throughput is For example, if the average network uplink or downlink data volume exceeds the threshold value, the application characteristics of the resource may be similarly obtained for the high network throughput other index values. Obtained, this is not limited in the present invention.
  • step S306 when the calculated application characteristics are different from the application characteristics of the resources in the resource allocation request, or when there is a better idle resource in the current resource pool, the resource adjustment is performed according to the preset resource adjustment policy.
  • the resource adjustment when the resource adjustment trigger condition of the resource application attribute allocation policy is satisfied, the resource adjustment is performed according to the preset resource adjustment policy.
  • the adjustment of resources includes the migration of resources and the adjustment of resource specifications. Among them, the adjustment of resource specifications may be the increase or decrease of memory.
  • the resource adjustment policy should be preset, and the resource adjustment policy includes a combination of one or more of whether the migration determination policy, the migrated resource determination policy, the target host selection policy, and the like are allowed.
  • the determining policy of the migrated resource may be a policy for determining which resource to migrate when multiple resources conflict.
  • further information should be further combined with other information of the resource management platform to determine whether conditions are available for resource migration. If migration is possible, then Steps determine the resources to be migrated and the target host, and perform resource migration after determining the above information, wherein other information of the resource management platform, such as whether there is a destination host available or the like.
  • the migration method uses the corresponding method. For example, when the resource is a virtual machine, the virtual machine migration technology is used. When the resource is a computing agent or other resource, the corresponding migration technology is captured, where the virtual machine migration is adopted. Technology can be offline migration, online migration, and intrinsic migration technologies.
  • a corresponding heuristic mechanism is established after allocating resources to the user, and when the migration trigger information is received, whether the resource is migrated according to the application characteristics of the current host device resources is determined. Therefore, the resource allocation of the host machine is adjusted in time, the utilization efficiency of resources on the host machine is improved, and the loss of hardware is reduced.
  • Embodiment 4 is a diagrammatic representation of Embodiment 4:
  • each Java virtual machine or computing agent may provide a computing service for one or more jobs, when receiving a computing request of the user (in the computing request) Carrying resource requirements and application characteristics of the job), the distributed computing platform (resource management platform) determines the host that provides the resource according to the configured resource application characteristic allocation policy, and then allocates the computing task of the job to the Java virtual machine or computing of the host node. Agent to complete the user submitted job.
  • the specific resource allocation manner is the same as the steps of Embodiments 1 and 2.
  • Embodiment 5 Only parts related to the embodiment of the present invention are shown.
  • the resource request receiving unit 41 receives a resource request sent by a resource requester, wherein the resource request includes a resource requirement and an application characteristic of the resource.
  • the resource request may be received through a preset interface, and the receiving interface of the resource request may be preset, so that the interface can receive the application characteristic of the resource, and the preset interface It can be a RESTful interface, a WebService interface, or a remote call interface.
  • Resources mainly refer to virtual machines (VMs) with a central processing unit (CPU), memory, hard disks, networks, etc., or resources with computing power, such as running on an operation. Java Virtual Machine (JVM) or computing agent on the system.
  • the resource request is initiated by the resource requester.
  • the resource requester may be any user requesting resources from the resource management platform, including a program, such as a scheduling module of the distributed computing system.
  • the host determining unit 42 determines a host to allocate resources to the resource requester based on the resource request and the resource application property allocation policy.
  • the resource application feature allocation policy should be configured in advance on the resource management platform.
  • the resource management platform determines the host that allocates resources to the user according to the resource requirements, the application characteristics of the resources, and the pre-configured resource application feature allocation policies.
  • the resource application characteristic allocation policy specifically includes a resource application characteristic allocation policy based on the area allocation (that is, in which area or regions the resource of an application characteristic is allocated), and a resource application characteristic allocation policy based on the resource allocated by the host machine (that is, a resource of an application characteristic) One or any combination of which host distribution and resource application characteristic allocation strategy based on the resource application characteristic combination policy.
  • the area refers to a logical area or a physical area
  • the logical area is one or more
  • the physical resources are logically grouped into a group for management, which has the same management rules or security policies.
  • the resource application feature combination strategy is used to define application characteristics of resources that can share the host machine, or application characteristics of resources that cannot share the host machine. Or, when the resources of the two application characteristics share a host, the degree of mutual influence between them, that is, the degree of coexistence, wherein the application characteristics of the resources that can share the host machine refer to which application characteristics of the resources can share the host. Cannot share the funds of the host
  • the application characteristics of the source refer to which application feature resources cannot share the host.
  • the allocation information returning unit 43 controls the host to allocate resources to the resource requester and returns the resource allocation information to the resource requester.
  • the host after determining the host that allocates the resource to the resource requester, the host is controlled to allocate resources to the resource requester, and returns resource allocation information to the resource requester, where the resource allocation information may be the identifier of the resource. And/or the address of the resource and/or any other information that can locate or identify the resource.
  • Example 6
  • FIG. 5 shows the structure of a resource allocation platform according to Embodiment 6 of the present invention. For the convenience of description, only parts related to the embodiment of the present invention are shown.
  • the resource request receiving unit 51 receives a resource request sent by a resource requester, wherein the resource request includes a resource requirement and an application characteristic of the resource.
  • the host determining unit 52 determines a host to allocate resources to the resource requester based on the resource request and the resource application property allocation policy.
  • the allocation information returning unit 53 controls the host to allocate resources to the resource requester and returns the resource allocation information to the resource requester.
  • the host determining unit 52 may specifically include an idle resource determining subunit 521 and a host determining subunit 522, where:
  • the idle resource determining sub-unit 521 is configured to determine, in the resource pool, an idle resource that satisfies the resource requirement according to the resource requirement in the resource request, the application characteristic of the resource, and the configured resource application characteristic allocation policy.
  • the resource requirement in the resource request has an area requirement
  • the resource providing area is set in a resource authority control policy preset in the resource pool
  • the application characteristics of the requested resource and resource in the allocation area are Locate the corresponding area and determine the idle resources that meet the resource requirements. Otherwise, locate all the resource pools to determine the idle resources that meet the resource requirements.
  • the host determining subunit 522 is configured to determine, according to the resource application characteristic combination policy, whether the host where the idle resource is located is a host that allocates resources when there is an idle resource that satisfies the user resource requirement.
  • the host where the idle resource is located is directly determined to be a host that allocates resources to the resource requester. If the resource application feature combination policy is defined in the resource application feature allocation policy, the application feature of the allocated resource on the host where the idle resource is located, the application feature of the resource in the resource request, and the resource application feature combination policy are determined. The free resource Whether the host is the host that allocates resources.
  • the idle resource determination sub-unit 521 is triggered to find the next idle resource.
  • the resource application feature combination policy defines an application feature that can share the resource of the host device or an application feature that cannot share the resource of the host device, directly determine the allocated resource according to the resource application feature combination policy.
  • the application characteristics and the application characteristics of the resources in the resource request can coexist.
  • the application characteristic combination policy of the resource defines the coexistence degree of the application characteristics of the resource, the coexistence can be determined according to the level of the resource coexistence. Specifically, the resource with a higher degree of coexistence is allocated to the user according to the order of the resource coexistence from high to low, and the following steps may be included:
  • the resource search algorithm of the resource management platform finds an available host as an alternative host, where the resource search algorithm may be a search algorithm for balancing the load of each host, or centralized use. Search algorithms for resources (to save power for non-use resources), etc.;
  • step 3 After finding an alternate host that can allocate the requested resource, determine whether some resources have been allocated on the candidate host, and if so, perform step 3), otherwise perform step 4);
  • step 3) is performed on the candidate host found in step 2), Coexistence requirements, namely: the application characteristics of the allocated resource and the application of the resource in the resource request Sexual coexistence is second highest. If you still can't find the right resources, go to step 3) for the alternative host found in step 2), reduce the coexistence requirement, and so on, and find the coexistence from high to low. Until you find a resource that meets the resource request and assign it.
  • Embodiment 7 Only parts related to the embodiment of the present invention are shown.
  • the policy configuration unit 61 pre-configures a resource application feature allocation policy, which includes a resource allocation feature allocation policy based on the region allocation, and any combination of resource applications based on the host allocation.
  • the resource request receiving unit 62 receives a resource request sent by a resource requester, wherein the resource request includes a resource requirement and an application characteristic of the resource.
  • the host determining unit 63 determines a host to allocate resources to the resource requester based on the resource request and the resource application property allocation policy.
  • the resource requirement in the resource request has an area requirement
  • the resource providing area is set in a preset resource authority control policy in the resource pool, or the allocation area is requested to be applied and the resource is applied.
  • the area corresponding to (specified) is located according to a certain search algorithm to determine an idle resource that meets the resource requirement. Otherwise, all resource pools are located to determine an idle resource that satisfies the resource requirement.
  • the host where the idle resource is located is directly determined to be a host that allocates resources to the resource requester. If the resource application feature combination policy is defined in the resource application feature allocation policy, the application feature of the allocated resource on the host where the idle resource is located, the application feature of the resource in the resource request, and the resource application feature combination policy are determined. Whether the host where the free resource is located is the host that allocates the resource.
  • the allocation information returning unit 64 controls the host to allocate resources to the resource requester, and returns resource allocation information to the resource requester.
  • the information holding unit 65 holds information related to the allocated resources, which includes the application characteristics of the resources.
  • the application characteristic calculation unit 66 collects the usage information of the resource according to the preset policy, and calculates the application characteristic of the resource according to the usage information of the resource.
  • the resource requirements input by the user and the application characteristics of the resource may have errors or are only input in the system default manner, or when the resource is initially allocated to the user, there is no more preferable alternative host, which results in allocation to the user.
  • the host where the resource is located is not optimal. Therefore, the usage information of the resource is collected according to the preset policy, and the application characteristic of the resource is calculated according to the usage information of the resource, and the current actual application characteristic of the resource is obtained, wherein the resource usage information may be hardware usage information.
  • the calculation method of the application characteristics of the resource may be: If the CPU usage information of the collected resource is analyzed, and the load is greater than 75% in 80% of the time, the application characteristic of the resource is high CPU usage. The example given here is the CPU. Other indicator values can also be obtained similarly.
  • the resource adjustment unit 67 adjusts the resource according to the preset resource adjustment policy when the calculated application characteristics are different from the application characteristics of the resources in the resource allocation request, or when the current resource pool has a better idle resource.
  • the resource adjustment when the resource adjustment trigger condition of the resource application attribute allocation policy is satisfied, the resource adjustment is performed according to the preset resource adjustment policy.
  • the adjustment of resources includes the migration of resources and the adjustment of resource specifications. Among them, the adjustment of resource specifications may be the increase or decrease of memory.
  • the resource adjustment policy should be preset, and the resource adjustment policy includes a combination of one or more of whether the migration determination policy, the migrated resource determination policy, the target host selection policy, and the like are allowed.
  • the device of the present invention is used to implement the method embodiment of the present invention. For a specific implementation process, reference may be made to the description of the method embodiment of the present invention, and details are not described herein again.
  • the client receives the resource allocation request input by the user, determines the resource specification of the resource, and the application feature of the resource, and sends the resource specification including the resource and the application request of the resource to the server, and the server is configured according to the pre-configured
  • the allocation policy of the resource and the application characteristics of the resource in the resource allocation request control the allocation of the resource by the host where the idle resource is located, and save the application characteristics of the allocated resource, the usage information of the resource collected after the resource is allocated, and according to the collected resource usage information.
  • the application characteristics of the computing resources adjust the resources in time, and solve the problem that the resources of the multiple applications share the resources and cause the overall performance of the host to decrease, thereby effectively improving the utilization efficiency of the server resources and reducing the hardware.
  • the loss has improved the user experience.

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Debugging And Monitoring (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Computer And Data Communications (AREA)

Abstract

本发明适用于计算机技术领域,提供了一种资源分配方法及资源管理平台,所述方法包括下述步骤:接收资源请求者发送的资源请求,其中所述资源请求包括资源需求和资源的应用特性;根据所述资源请求和资源应用特性分配策略,确定向所述资源请求者分配资源的宿主机;控制所述宿主机向所述资源请求者分配资源,向资源请求者返回资源分配信息,从而解决了现有技术中由于多个应用程序共用硬件资源产生硬件资源的竞争,导致宿主机整体性能下降的问题,有效提高了服务器资源的利用效率、降低了硬件的损耗,提高了用户体验。

Description

说 明 书 一种资源分配方法及资源管理平台
本申请要求 2011年 08月 29日递交的申请号为 201110251036.3、发明名称 为 "一种资源分配方法及资源管理平台" 的中国专利申请的优先权, 其全部内 容通过引用结合在本申请中。 技术领域
本发明属于计算机技术领域,尤其涉及一种资源分配方法及资源管理平台。 背景技术
云计算可以根据用户资源的需求进行按需分配, 即业务需要资源时进行分 配, 不需要时释放资源, 当有多个错峰业务时, 可以从整体上提高系统各个部 分的利用效率。 云计算的这一特点使得企业运营的硬件成本和管理成本得到了 有效地降低, 提高了硬件资源利用率, 也使得云计算技术得到了极大的提高。
作为一种最为常见的资源分配方式, 虚拟机 ( Virtual Machine , 缩写 为 VM )通过将硬件资源虚拟化, 在同一个计算机硬件上向多个用户提供服务, 提高了硬件资源的利用率。 当用户请求一定能力的资源时, 可以通过资源管理 平台暴露的接口向资源管理平台申请资源, 申请资源时, 用户提供资源能力需 求参数, 该需求参数包括: VM的 CPU、 内存、 硬盘、 网络能力, VM的位置 要求, VM的镜像(Image )等, 需求参数主要是指需要资源管理平台满足的一 些技术性能参数要求, 进而资源管理平台根据用户的需求参数向用户提供资 源。 在现有技术中, 同一宿主机上的 VM之间具有一定隔离性, 而且随着虚拟 化技术的不同, 隔离性能也不一样, 这个隔离性主要是 VM之间的相互影响, 如一个 VM上运行的应用进入死循环状态, 可能会导致该 VM死机, 但该 VM 的死机不会影响其他 VM的运行。 本发明中的宿主机指运行 VM的机器, 其可 以是物理机, 也可以是一个 VM。 然而, 由于 VM共用硬件资源, 而硬件资源本身的特性限制会使得 VM之 间仍然会存在硬件资源的竟争, 例如: 硬盘的读写, 由于物理硬盘的转速、 读 写速度等, 多个虚拟机共同频繁读写硬盘会相互影响。 当同一物理机上有两个 或多个 VM, 而各 VM所运行的应用可能会有资源争夺, 就会使得各个 VM性 能无法保证, 且物理机不能发挥最优性能, 例如: 一个物理机有 2个 VM, 而 这 2个 VM所运行的应用都是频繁读写硬盘的应用, 如使用硬盘緩存所处理的 数据或者硬盘数据的拷贝、 更改等, 此时这 2个 VM上的应用会导致 VM之间 有硬件资源的使用竟争, 会使得 VM所在的物理机的整体性能下降。 由于目前 的基础设施即服务(IaaS ) 的物理资源是多用户共享, 用户无法知道其所申请 到的资源和哪些其他的应用共享物理资源, 使得用户无法避免 VM间硬件资源 竟争情况的出现, 同时, 由于资源管理模块无法获知所分配的 VM上将要运行 的应用的特征, 使得资源管理模块同样无法避免上述资源竟争情况的出现, 也 就无法避免竟争导致 VM性能的下降。 另外, 对于硬件资源有同质需求的多个 VM (同质指 VM对硬件资源的需求相同或相似, 例如两个 VM都是频繁读写 硬盘) , 如果部署在同一物理机上, 长期运行则会使得该物理机的某个硬件长 期处于高负荷状态, 而其他的硬件则长期处于低负荷状态, 使得该物理机的整 体性能发挥不均衡, 增加了能耗和硬件损耗。 发明内容
本发明实施例的目的在于提供一种资源分配方法, 旨在解决由于多个应用 程序共用硬件资源产生硬件资源的竟争, 导致宿主机整体性能下降的问题。
本发明实施例是这样实现的,一种资源分配方法, 所述方法包括下述步骤: 接收资源请求者发送的资源请求, 其中所述资源请求包括资源需求和资源 的应用特性;
根据所述资源请求和资源应用特性分配策略, 确定向所述资源请求者分配 资源的宿主机; 控制所述宿主机向所述资源请求者分配资源, 向所述资源请求者返回资源 分配信息。
本发明实施例的另一目的在于提供一种资源管理平台, 所述资源管理平台 包括:
资源请求接收单元, 用于接收资源请求者发送的资源请求, 其中所述资源 请求包括资源需求和资源的应用特性;
宿主机确定单元, 用于根据资源请求和资源应用特性分配策略以及资源应 用特性组合策略, 确定向资源请求者分配资源的宿主机; 以及
分配信息返回单元, 用于控制所述宿主机向资源请求者分配资源, 向所述 资源请求者返回资源分配信息。
本发明实施例通过接收资源请求者的包括资源需求和资源的应用特性的资 源请求, 根据该资源请求、 预设的资源应用特性分配策略, 确定向资源请求者 分配资源的宿主机, 控制宿主机向资源请求者分配资源, 向资源请求者返回资 确定分配资源的宿主机, 解决了现有技术中由于多个应用程序共用硬件资源产 生硬件资源的竟争、 导致宿主机以及分配的资源本身性能下降的问题, 从而根 据资源请求和资源应用特性分配策略进行资源的分配, 优化了资源的分配, 有 效提高了资源的利用效率, 降低了硬件损耗, 提高了用户体验。 附图说明
图 1是本发明实施例一提供的资源分配方法的实现流程图;
图 2是本发明实施例二提供的资源分配方法的实现流程图;
图 3是本发明实施例三提供的资源分配方法的实现流程图;
图 4是本发明实施例五提供的资源管理平台的结构示意图;
图 5是本发明实施例六提供的资源管理平台的结构示意图;
图 6是本发明实施例七提供的资源管理平台的结构示意图。 具体实施方式
为了使本发明的目的、 技术方案及优点更加清楚明白, 以下结合附图及实 施例, 对本发明进行进一步详细说明。 应当理解, 此处所描述的具体实施例仅 仅用以解释本发明, 并不用于限定本发明。
以下结合
实施例一:
图 1示出了本发明实施例一提供的资源分配方法的实现流程, 详述如下: 在步骤 S101 中, 接收资源请求者发送的资源请求, 资源请求包括资源需 求和资源的应用特性。
在本发明实施例中, 可以通过预设的接口接收资源请求, 从而使该接口可 以接收资源的应用特性, 该预设的接口可以为表征状态转移 (RESTful )接口、 网络服务(WebService )接口或远程调用接口, 资源主要是指具有中央处理器 ( CPU ), 内存、 硬盘、 网络等的虚拟机(VM ), 或者具有计算能力的资源, 如 运行于某一操作系统上的 Java虚拟机( JVM )或计算代理。 该资源请求是由资 源请求者发起, 在具体实施过程中, 该资源请求者可以是向资源管理平台请求 资源的任何用户, 当然该资源请求者也包括程序, 如: 分布式计算系统的调度 模块。
在本发明实施例中, 资源请求中包括的资源需求是指用户请求的资源性能 参数要求, 即: 所提供资源需要满足的一些能力要求, 本发明中称之为资源规 格, 例如, 当请求的资源为虚拟机时, 资源规格可以是资源所在宿主机位置、 虚拟机的 CPU、 内存、 磁盘、 网络、 操作系统或应用程序之一或其任意组合, 其中, 资源所在宿主机位置可以是指定宿主机应该在哪个位置, 或者对位置有 哪些要求, 如所申请的多个资源位于同一个宿主机等, 具体示例如表 1所示。
表 1 :
资源需求规格 CPU 2核 3.0 GHz
内存 4G
硬盘 160G
网络 1000Mbps
操作系统 Linux 在表 1中列出了一资源请求中资源的需求: 2核的 CPU, 频率为 3.0GHz、 内存为 4G以及硬盘为 160G等。
在本发明实施例中, 资源的应用特性是指资源上所运行的应用的类型, 资 源上所运行的应用的类型可以由资源提供者预先进行定义。 资源请求者从资源 提供者定义的资源上所运行的应用的类型中选择最符合其所要运行的程序的类 型作为其所申请资源的应用特性, 或者, 资源的应用特性是指资源请求者将对 所要申请的资源的使用情况, 该使用情况包括但不限于如下项目中的一个或任 意组合: CPU使用率、 内存使用率、 磁盘读写、 网络吞吐(上下行)、 高实时 性、 图形处理器 (Graphic Processing Unit, GPU )使用率等, 该使用情况定义 了对该资源各能力的使用负载情况, 其是运行于所申请的资源上的所有应用程 序的特性的综合。 例如, 某虚拟机运行一个应用, 而该应用是硬盘频繁读写的, 则在虚拟机层面表现是硬盘频繁读写, 在本发明中称之为该资源的应用特性是 硬盘频繁读写。 再例如, 某虚拟机运行两个应用, 其中一个是网络高负载, 另 外一个各硬件资源占用均不高, 则该两个应用的综合表现是网络高负载, 在本 发明中称之为该资源的应用特性是网络高负载。 具体示例如表 2所示。
表 2:
资源的应用特性 特性
CPU 高 ( >75% )
内存 低
磁盘读写 低 网络吞吐 低
实时性 低
GPU 高
在表 2 描述了一资源请求中资源的应用特性, CPU 的使用率为高 (使用 率>75% ) , 内存、 磁盘读写、 网络吞吐、 实时性等较低, 具体可以根据系统需 要进行应用特性的定义, 例如内存的使用率 <10%, 则将内存的应用特性定义为 低。
在步骤 S102中,根据资源请求和资源应用特性分配策略,确定向资源请求 者分配资源的宿主机。
在本发明中, 可以预先在资源管理平台配置资源应用特性分配策略。 资源 管理平台根据资源需求、 资源的应用特性以及资源应用特性分配策略, 确定向 用户分配资源的宿主机。 其中, 资源应用特性分配策略可以具体包括基于区域 分配的资源应用特性分配策略, 基于宿主机分配的资源应用特性分配策略, 以 中, 基于区域分配的资源应用特性分配策略指的是某应用特性的资源在哪个或 哪些区域分配, 基于宿主机分配的资源应用特性分配策略指的是某应用特性的 资源在哪个或哪些宿主机分配, 区域指的是逻辑上的区域或物理上的区域, 逻 辑区域是将一个或多个物理资源逻辑上归入一个组进行管理, 其有相同的管理 规则或安全策略。 资源应用特性组合策略用于定义可以共享宿主机的资源的应 用特性, 或者不能共享宿主机的资源的应用特性, 或者两种应用特性的资源共 享一个宿主机时, 它们之间的相互影响程度, 即共存度, 其中, 可以共享宿主 机的资源的应用特性指的是哪些应用特性的资源可以共享宿主机, 不能共享宿 主机的资源的应用特性指的是哪些应用特性的资源不能共享宿主机。 在具体实 施过程中, 可以釆用一个规则文件保存资源应用特性组合策略。 如表 3所示, 举例描述了一个配置的资源应用特性组合策略。
表 3: 资源 1的类型 资源 2的类型 共存度
高磁盘读写 高磁盘读写 0
高网络访问 高网络访问 1
高 CPU使用 高网络访问 2 在表 3中描述了不同类型的资源 1与资源 2共享宿主机对资源性能的影响 的大小, 使用共存度进行表示, 并将影响因子分为三级。 在第二行中, 高磁盘 读写类资源 1与高磁盘读写类资源 2共享宿主机的共存度为 0 , 共存度低, 不 推荐共存, 在第三行中, 高网络访问类资源 1与高网络访问类资源 2共享宿主 机的共存度为 1 , 表示影响中等, 共存度次高, 在一定程度上可以共存, 在第 四行中,高 CPU使用类资源 1与高网络访问类资源 2共享宿主机的共存度为 2, 表示影响较少, 共存度较高, 可以共存。
在步骤 S103 中, 控制分配资源的宿主机向资源请求者分配资源, 向资源 请求者返回资源分配信息。
在本发明实施例中, 确定向资源请求者分配资源的宿主机后, 控制宿主机 向资源请求者分配资源, 并向资源请求者返回资源分配信息, 该资源分配信息 可以是资源的标识和 /或资源的地址和 /或其他任何可以定位或标识资源的信息。
向资源请求者成功分配资源后, 进一步保存所分配的资源的相关信息, 作 为后续资源分配或资源调整的基础, 其中资源调整可以是资源的迁移。 所保存 的资源的相关信息包括资源的应用特性, 还可以包括: 资源的位置、 资源的用 户、 资源的分配时间、 资源有效期、 资源的能力、 资源的安全策略、 资源的迁 移策略等信息中的一个或多个。
本发明实施例通过 RESTful接口、 WebService接口或远程调用接口接收包 括资源需求和资源的应用特性的资源请求, 根据资源管理平台预设的预设的资 源应用特性分配策略, 确定向资源请求者分配资源的宿主机, 控制宿主机向资 源请求者分配资源, 从而在资源分配前根据请求资源的应用特征以及配置的资 源应用特性组合策略, 优化资源的分配, 有效提高了资源的利用效率, 提高了 用户体验。
实施例二:
图 2示出了本发明实施例二提供的资源分配方法的实现流程, 详述如下: 在步骤 S201中,接收资源请求者发送资源请求,其中资源请求包括资源需 求和资源的应用特性。
在步骤 S202中,根据资源请求的资源需求、资源的应用特性以及配置的资 源应用特性分配策略, 在资源池中确定一满足资源需求的空闲资源。
在本发明实施例中, 资源管理平台接收到资源请求者提交的资源请求后, 根据该资源请求中携带的资源需求和资源池中的空闲资源确定是否存在满足该 请求的空闲资源, 若有, 则对该空闲资源执行步骤 203。 这里的满足具体指, 如果该资源需求小于当前资源池中空闲资源数, 且该资源池中可提供的最大单 体资源满足该资源需求中的最大单体资源需求, 则认为可以满足资源请求者的 需求, 否则认为是不满足。 例如: 由于一个 VM必须位于一个单一的物理机上, 所以, 如果 VM的规格需求是 16核 CPU、 16G内存, 而该资源池中最大的单 体物理资源的规格配置为 8核 CPU、 8G内存, 则该 VM的规格需求大于该资 源池中所有物理机的规格配置,则认为该资源无法满足该资源请求的资源需求, 其中, 单一的物理机上指的是不能跨物理机。 单体资源指不可再分割的资源, 如单个服务器。
在具体实施过程中, 当资源请求中的资源需求存在区域要求, 或资源提供 区域在所述资源池中预设的资源权限控制策略中被设定, 或者分配区域被请求 需求和资源的应用特性, 到 (指定)对应的区域按一定的搜索算法定位以确定 满足资源需求的空闲资源, 否则, 定位到所有资源池确定满足资源需求的空闲 资源。 例如, 若资源请求中的资源需求有区域要求, 或者资源提供者设置的资 源权限控制策略限定了在某个特定区域为该资源请求者提供资源, 或者根据该 区域, 则定位到该区域确定满足所述资源需求的空闲资源, 如果无区域要求, 则定位到全部区域确定满足所述资源需求的空闲资源。 其中, 这里的资源提供 者的资源应用特性分配策略定义了基于区域分配的资源应用特性分配策略,和 / 或基于宿主机分配的资源应用特性分配策略。 其中, 上述区域为逻辑区域或物 理位置, 其中, 资源提供者设置的资源权限控制策略可以为如下策略中的一个 或任意组合: 基于资源请求者级别的策略、 基于资源请求者订购的资源类别的 策略、 资源请求者账号所在群组的分配策略、 资源提供者的区域策略(如根据 请求者 IP确定的请求者所在位置)、 资源故障应对策略、 国家法律限制或者资 源提供者的其他策略等。
在步骤 S203中, 当存在满足所述资源需求的空闲资源时,确定该空闲资源 所在的宿主机是否为分配资源的宿主机。 性组合策略, 则直接确定该空闲资源所在宿主机为向资源请求者分配资源的宿 主机。 若预先在资源应用特性分配策略中定义了资源应用特性组合策略, 则根 据该空闲资源所在宿主机上的已分配资源的应用特性、 资源请求中的资源的应 用特性、 以及资源应用特性组合策略确定该空闲资源所在的宿主机是否为分配 资源的宿主机。
作为一种优选的实施方式, 在确定一空闲资源后, 确定该空闲资源所在的 宿主机上是否有其他的已分配资源, 如果有, 则根据配置的资源应用特性组合 策略, 判断该已分配的资源的应用特性和该资源请求中的资源的应用特性可否 共存, 如果可以共存, 则确定该空闲资源所在宿主机为向资源请求者分配资源 的宿主机, 执行步骤 204, 否则重新执行步骤 202查找下一空闲资源。
在具体实施过程中, 如果资源应用特性组合策略定义了可以共享宿主机的 资源的应用特性或者不能共享宿主机的资源的应用特性, 则直接根据该资源应 用特性组合策略判断该已分配的资源的应用特性和该资源请求中的资源的应用 特性可否共存,如果资源的应用特性组合策略定义了资源的应用特性的共存度, 则根据资源共存度的高低判断可否共存。 具体的, 按照资源共存度从高到低的 顺序选择共存度较高的资源分配给用户, 可以包括以下步骤:
1 )在所定位的区域中的资源中,根据资源管理平台的资源搜索算法查找可 用的宿主机作为备选宿主机, 这里的资源搜索算法可以是均衡各宿主机负载的 搜索算法、 或集中使用资源 (以便对非使用资源断电节能) 的搜索算法等;
2 ) 查找到一个可分配请求资源的备选宿主机后,确定该备选宿主机上是否 已经有部分资源已经分配出去了, 如果有, 则执行步骤 3 ), 否则执行步骤 4 );
3 ) 获取已经分配的资源的应用特性,根据配置的资源应用特性组合策略和 资源请求中资源的应用特性, 确定该备选宿主机上已经分配的资源和欲分配的 资源可否共存(根据配置的资源应用特性组合策略的共存度进行判断, 首先选 择当前可用的共存度最高的), 如果可以, 则执行步骤 4 ), 否则执行步骤 5 );
4 ) 在该备选宿主机上为该资源请求分配所需资源;
5 )重新执行步骤 1 ) ~4 ), 直到完成资源分配, 或者查找完所有可用的宿 主机, 均无法完成该资源分配, 则对步骤 2 )中找到的备选宿主机执行步骤 3 ), 降低共存度要求, 即: 已分配的资源的应用特性和该资源请求中资源的应用特 性的共存度次高, 如果仍然无法找到合适的资源, 则进一步对步骤 2 ) 中找到 的备选宿主机执行步骤 3 ), 降低共存度要求, 依此类推, 按照共存度从高到低 进行查找, 直到找到满足资源请求的资源并进行分配,从而避免了资源的竟争, 实现了资源的最优分配, 提高了资源的分配效率和使用效率。
在步骤 S204中,控制分配资源的宿主机向资源请求者分配资源,向资源请 求者返回资源分配信息。
在本发明实施例中, 进一步地, 保存所分配的资源的相关信息, 作为后续 资源分配或资源调整(如迁移) 的基础。 所保存的资源的相关信息包括资源的 应用特性, 还可以包括: 资源的位置、 资源的用户、 资源的分配时间、 资源有 效期、 资源的能力、 资源的安全策略、 资源的迁移策略等信息中的一个或多个。
实施例三: 在本发明实施例中, 由于用户输入的资源需求和资源的应用特性可能存在 误差或者只是以系统默认方式输入, 或者当初向用户分配资源时没有更优选的 备选宿主机, 导致为用户分配的资源所在的宿主机并非最优, 其中, 系统默认 方式输入指的是用户本身没有输入。 为了分配最大可能地给用户分配更优先的 资源, 资源管理平台增加资源应用特性的收集功能, 用以确定是否需要调整资 源, 其中, 资源应用特性的收集功能指的是资源管理平台在运行过程中根据监 控信息推算的应用特性。
图 3示出了本发明实施例三提供的资源分配方法的实现流程, 详述如下: 在步骤 S301中, 预先配置资源应用特性分配策略。
在本发明实施例中, 资源应用特性分配策略具体包括基于区域分配的资源 应用特性分配策略, 基于宿主机分配的资源应用特性分配策略, 以及基于资源 应用特性组合策略的资源应用特性分配策略之一或其任意组合。 其中, 基于区 域分配的资源应用特性分配策略指的是某应用特性的资源在哪个或哪些区域分 配, 基于宿主机分配的资源应用特性分配策略指的是某应用特性的资源在哪个 或哪些宿主机分配, 区域指的是逻辑上的区域或物理上的区域, 逻辑区域是将 一个或多个物理资源逻辑上归入一个组进行管理, 其有相同的管理规则或安全 策略。
在步骤 S302 中, 接收资源请求者发送的资源请求, 其中资源请求包括资 源需求和资源的应用特性。
在步骤 S303 中, 根据资源请求和资源应用特性分配策略, 确定向资源请 求者分配资源的宿主机。
在步骤 S304 中, 控制分配资源的宿主机向资源请求者分配资源, 向资源 请求者返回资源分配信息。
在步骤 S305 中, 根据预设的策略收集资源的使用信息, 根据资源的使用 信息计算该资源的应用特性。
在本发明实施例中, 由于用户输入的资源需求和资源的应用特性可能存在 误差或者只是以系统默认方式输入, 或者当初向用户分配资源时没有更优选的 备选宿主机, 导致为用户分配的资源所在的宿主机并非最优。 因此, 根据预设 的策略收集资源的使用信息, 根据资源的使用信息计算该资源的应用特性, 获 取资源的当前实际的应用特性, 其中, 资源的使用信息可以是硬件使用信息。 在具体实施过程中, 资源的应用特性的计算方法可以是: 如果分析收集到的资 源的 CPU使用信息, 得出在 80%的时间里负荷大于 75%, 则该资源的应用特 性为高 CPU使用, 这里的举的例子是以 CPU为例。 以磁盘读写为例, 80%的 时间内硬盘读写时间超过 60%, 或者平均每小时硬盘读写字节数超过阔值, 则 该资源的应用特性为高硬盘读写, 以网络吞吐为例, 平均每小时网络上行或下 行数据量超过阔值, 则该资源的应用特性为高网络吞吐其他指标值也可以类似 地获得。 获取, 本发明中对此不作限制。
在步骤 S306中,当计算得到的应用特性与资源分配请求中资源的应用特性 不同, 或当前的资源池存在更优的空闲资源时, 根据预设的资源调整策略进行 资源的调整。
在本发明实施例中, 当基于资源应用特性分配策略的资源调整触发条件满 足时, 则根据预设的资源调整策略进行资源的调整。 其中, 当基于资源应用特 性分配策略的资源调整触发条件满足可以是计算得到的应用特性与资源分配请 求中资源的应用特性不同, 或当前的资源池存在更优的空闲资源时。 资源的调 整包括资源的迁移和资源规格的调整等, 其中, 资源规格的调整可以是内存的 增减等。 在本发明实施例中, 应预先设置资源调整策略, 资源调整策略包括是 否允许迁移的判断策略、 被迁移的资源的确定策略、 目标宿主机选择策略等中 的一个或多个的组合。其中,被迁移的资源的确定策略可以是多个资源冲突时, 确定迁移其中哪个资源的策略。 具体实施过程中, 以迁移为例, 应进一步结合 资源管理平台其他信息确定是否有条件进行资源迁移, 如果可以迁移, 则进一 步确定要迁移的资源以及目标宿主机, 并在确定上述信息后进行资源的迁移, 其中, 资源管理平台其他信息如是否有可用的目的宿主机等。 资源的迁移釆用 对应的方法, 例如, 当资源为虚拟机时, 则釆用虚拟机迁移技术, 当资源为计 算代理或其它资源时, 釆取对应的迁移技术, 其中, 釆用虚拟机迁移技术可以 是离线迁移、 在线迁移和内在迁移技术等。
在本发明实施例中, 为了有效提高资源的利用效率, 在向用户分配资源后 建立相应的启发式机制, 在接收到迁移触发信息时根据当前宿主机上资源的应 用特性判断是否对资源进行迁移, 从而及时地调整宿主机的资源分配, 提高宿 主机上资源的利用效率, 降低了硬件的损耗。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分步骤是 可以通过程序来指令相关的硬件来完成, 所述的程序可以存储于一计算机可读 取存储介质中, 所述的存储介质, 如 ROM/RAM、 磁盘、 光盘等。
实施例四:
在本发明实施例中, 当资源是 Java虚拟机或计算代理时,每个 Java虚拟机 或者计算代理可以为一个或多个作业提供计算服务, 在接收到用户的计算请求 时(该计算请求中携带资源需求和作业的应用特性) , 分布式计算平台 (资源 管理平台)根据配置的资源应用特性分配策略确定提供资源的宿主机, 进而分 配该作业的计算任务到宿主节点的 Java虚拟机或计算代理,从而完成用户提交 的作业。 具体的资源分配方式同实施例一、 二的步骤。
实施例五: 仅示出了与本发明实施例相关的部分。
资源请求接收单元 41接收资源请求者发送的资源请求,其中资源请求包括 资源需求和资源的应用特性。
在本发明实施例中, 可以通过预设的接口接收资源请求, 该资源请求的接 收接口可以是预设的, 从而使该接口可以接收资源的应用特性, 该预设的接口 可以为 RESTful接口、 WebService接口或远程调用接口, 资源主要是指具有中 央处理器(CPU )、 内存、 硬盘、 网络等的虚拟机(VM ), 或者具有计算能力的 资源, 如运行于某一操作系统上的 Java虚拟机( JVM )或计算代理。 该资源请 求是由资源请求者发起, 在具体实施过程中, 该资源请求者可以是向资源管理 平台请求资源的任何用户, 包括程序, 如: 分布式计算系统的调度模块。
宿主机确定单元 42根据资源请求和资源应用特性分配策略,确定向资源请 求者分配资源的宿主机。
在本发明中, 应预先在资源管理平台配置资源应用特性分配策略。 资源管 理平台根据资源需求、资源的应用特性以及预先配置的资源应用特性分配策略, 确定向用户分配资源的宿主机。 其中, 资源应用特性分配策略具体包括基于区 域分配的资源应用特性分配策略(即某应用特性的资源在哪个或哪些区域分 配),基于宿主机分配的资源应用特性分配策略(即某应用特性的资源在哪个或 哪些宿主机分配 以及基于资源应用特性组合策略的资源应用特性分配策略之 一或其任意组合。 其中, 区域指的是逻辑上的区域或物理上的区域, 逻辑区域 是将一个或多个物理资源逻辑上归入一个组进行管理, 其有相同的管理规则或 安全策略。资源应用特性组合策略用于定义可以共享宿主机的资源的应用特性, 或者不能共享宿主机的资源的应用特性, 或者两种应用特性的资源共享一个宿 主机时, 它们之间的相互影响程度, 即共存度, 其中, 可以共享宿主机的资源 的应用特性指的是哪些应用特性的资源可以共享宿主机, 不能共享宿主机的资 源的应用特性指的是哪些应用特性的资源不能共享宿主机。
分配信息返回单元 43控制宿主机向资源请求者分配资源,向资源请求者返 回资源分配信息。
在本发明实施例中, 在确定向资源请求者分配资源的宿主机后, 控制该宿 主机向资源请求者分配资源, 并向资源请求者返回资源分配信息, 该资源分配 信息可以是资源的标识和 /或资源的地址和 /或其他任何可以定位或标识资源的 信息。 实施例六:
图 5示出了本发明实施例六提供的资源分配平台的结构, 为了便于说明, 仅示出了与本发明实施例相关的部分。
资源请求接收单元 51接收资源请求者发送的资源请求,其中资源请求包括 资源需求和资源的应用特性。
宿主机确定单元 52根据资源请求和资源应用特性分配策略,确定向资源请 求者分配资源的宿主机。
分配信息返回单元 53控制宿主机向资源请求者分配资源,向资源请求者返 回资源分配信息。
在具体实施过程中,宿主机确定单元 52具体可以包括空闲资源确定子单元 521和宿主机确定子单元 522, 其中:
空闲资源确定子单元 521 , 用于根据资源请求中的资源需求、 资源的应用 特性以及配置的资源应用特性分配策略, 在资源池中确定一满足资源需求的空 闲资源。
在本发明实施例中, 当资源请求中的资源需求存在区域要求, 或资源提供 区域在资源池中预设的资源权限控制策略中被设定, 或者分配区域被请求的资 和资源的应用特性定位到对应的区域并确定满足资源需求的空闲资源, 否则, 定位到所有资源池确定满足资源需求的空闲资源。
宿主机确定子单元 522, 用于当存在满足用户资源需求的空闲资源时, 根 据资源应用特性组合策略,确定空闲资源所在宿主机是否为分配资源的宿主机。
在本发明实施例中, 当确定资源分配区域后, 若预先在资源应用特性分配 策略中未定义资源应用特性组合策略, 则直接确定该空闲资源所在宿主机为向 资源请求者分配资源的宿主机。 若预先在资源应用特性分配策略中定义了资源 应用特性组合策略,则根据该空闲资源所在宿主机上的已分配资源的应用特性、 资源请求中的资源的应用特性、 以及资源应用特性组合策略确定该空闲资源所 在的宿主机是否为分配资源的宿主机。
在本发明实施例, 作为一种优选的实施方式, 在确定一空闲资源后, 确定 该空闲资源所在的宿主机上是否有其他的已分配资源, 如果有, 则根据配置的 资源应用特性组合策略, 判断该已分配的资源的应用特性和该资源请求中的资 源的应用特性可否共存, 如果可以共存, 则确定该空闲资源所在宿主机为向资 源请求者分配资源的宿主机, 同时触发分配信息返回单元 53 , 否则触发空闲资 源确定子单元 521查找下一空闲资源。
在具体实施过程中, 如果资源应用特性组合策略定义了可以共享宿主机的 资源的应用特性或者不能共享宿主机的资源的应用特性, 则直接根据该资源应 用特性组合策略判断该已分配的资源的应用特性和该资源请求中的资源的应用 特性可否共存,如果资源的应用特性组合策略定义了资源的应用特性的共存度, 则根据资源共存度的高低判断可否共存。 具体的, 按照资源共存度从高到低的 顺序选择共存度较高的资源分配给用户, 可以包括以下步骤:
1 )在所定位的区域中的资源中,根据资源管理平台的资源搜索算法查找可 用的宿主机作为备选宿主机, 这里的资源搜索算法可以是均衡各宿主机负载的 搜索算法、 或集中使用资源 (以便对非使用资源断电节能) 的搜索算法等;
2 ) 查找到一个可分配请求资源的备选宿主机后,确定该备选宿主机上是否 已经有部分资源已经分配出去了, 如果有, 则执行步骤 3 ), 否则执行步骤 4 );
3 ) 获取已经分配的资源的应用特性,根据配置的资源应用特性组合策略和 资源请求中资源的应用特性, 确定该备选宿主机上已经分配的资源和欲分配的 资源可否共存(根据配置的资源应用特性组合策略的共存度进行判断, 首先选 择当前可用的共存度最高的), 如果可以, 则执行步骤 4 ), 否则执行步骤 5 );
4 ) 在该备选宿主机上为该资源请求分配所需资源;
5 )重新执行步骤 1 ) ~4 ), 直到完成资源分配, 或者查找完所有可用的宿 主机, 均无法完成该资源分配, 则对步骤 2 )中找到的备选宿主机执行步骤 3 ), 降低共存度要求, 即: 已分配的资源的应用特性和该资源请求中资源的应用特 性的共存度次高, 如果仍然无法找到合适的资源, 则进一步对步骤 2 ) 中找到 的备选宿主机执行步骤 3 ), 降低共存度要求,依此类推,共存度从高到低查找, 直到找到满足资源请求的资源并进行分配。
实施例七: 仅示出了与本发明实施例相关的部分。
策略配置单元 61预先配置资源应用特性分配策略,该资源应用特性分配策 略包括基于区域分配的资源应用特性分配策略, 基于宿主机分配的资源应用特 其任意组合。
资源请求接收单元 62接收资源请求者发送的资源请求,其中资源请求包括 资源需求和资源的应用特性。
宿主机确定单元 63根据资源请求和资源应用特性分配策略,确定向资源请 求者分配资源的宿主机。
在本发明实施例中, 当资源请求中的资源需求存在区域要求, 或资源提供 区域在所述资源池中预设的资源权限控制策略中被设定, 或者分配区域被请求 需求和资源的应用特性, 到 (指定)对应的区域按一定的搜索算法定位以确定 满足资源需求的空闲资源, 否则, 定位到所有资源池确定满足资源需求的空闲 资源。
当确定资源分配区域后, 若预先在资源应用特性分配策略中未定义资源应 用特性组合策略, 则直接确定该空闲资源所在宿主机为向资源请求者分配资源 的宿主机。 若预先在资源应用特性分配策略中定义了资源应用特性组合策略, 则根据该空闲资源所在宿主机上的已分配资源的应用特性、 资源请求中的资源 的应用特性、 以及资源应用特性组合策略确定该空闲资源所在的宿主机是否为 分配资源的宿主机。 分配信息返回单元 64控制宿主机向资源请求者分配资源,向资源请求者返 回资源分配信息。
信息保存单元 65保存分配的资源的相关信息,该相关信息包括资源的应用 特性。
应用特性计算单元 66根据预设的策略收集资源的使用信息,根据资源的使 用信息计算资源的应用特性。
在本发明实施例中, 由于用户输入的资源需求和资源的应用特性可能存在 误差或者只是以系统默认方式输入, 或者当初向用户分配资源时没有更优选的 备选宿主机, 导致为用户分配的资源所在的宿主机并非最优。 因此, 根据预设 的策略收集资源的使用信息, 根据资源的使用信息计算该资源的应用特性, 获 取资源的当前实际的应用特性, 其中, 资源的使用信息可以是硬件使用信息。 在具体实施过程中, 资源的应用特性的计算方法可以是: 如果分析收集到的资 源的 CPU使用信息, 得出在 80%的时间里负荷大于 75%, 则该资源的应用特 性为高 CPU使用, 这里的举的例子是以 CPU为例。 其他指标值也可以类似地 获得。
资源调整单元 67 当计算得到的应用特性与资源分配请求中资源的应用特 性不同, 或当前的资源池存在更优的空闲资源时, 根据预设的资源调整策略进 行资源的调整。
在本发明实施例中, 当基于资源应用特性分配策略的资源调整触发条件满 足时, 则根据预设的资源调整策略进行资源的调整。 其中, 当基于资源应用特 性分配策略的资源调整触发条件满足可以是计算得到的应用特性与资源分配请 求中资源的应用特性不同, 或当前的资源池存在更优的空闲资源时。 资源的调 整包括资源的迁移和资源规格的调整等, 其中, 资源规格的调整可以是内存的 增减等。 在本发明实施例中, 应预先设置资源调整策略, 资源调整策略包括是 否允许迁移的判断策略、 被迁移的资源的确定策略、 目标宿主机选择策略等中 的一个或多个的组合。 本发明的装置是用于实现本发明所述的方法实施例的, 其具体的实现过程 可以参照本发明方法实施例部分的描述, 在此不再赘述。
本发明实施例通过客户端接收用户输入的资源的分配请求, 确定资源的资 源规格以及资源的应用特性, 将包含资源的资源规格以及资源的应用特性的分 配请求发送给服务器, 服务器根据预先配置的资源的分配策略以及资源分配请 求中资源的应用特性, 控制空闲资源所在宿主机分配资源, 并保存分配的资源 的应用特性, 在分配资源后收集的资源的使用信息, 并根据收集的资源使用信 息计算资源的应用特征及时地进行资源的调整, 解决了由于多个应用程序共用 资源时对资源的竟争, 导致宿主机整体性能下降的问题, 从而有效提高了服务 器资源的利用效率, 降低了硬件的损耗, 提高了用户的体验。 以上所述仅为本 发明的较佳实施例而已, 并不用以限制本发明, 凡在本发明的精神和原则之内 所作的任何修改、 等同替换和改进等, 均应包含在本发明的保护范围之内。

Claims

权 利 要 求 书
1、 一种资源分配方法, 其特征在于, 所述方法包括下述步骤:
接收资源请求者发送的资源请求, 其中所述资源请求包括资源需求和资源 的应用特性;
根据所述资源请求和资源应用特性分配策略, 确定向所述资源请求者分配 资源的宿主机;
控制所述分配资源的宿主机向所述资源请求者分配资源, 向所述资源请求 者返回资源分配信息。
2、如权利要求 1所述的方法, 其特征在于, 在根据所述资源请求和资源应 用特性分配策略, 确定向所述资源请求者分配资源的宿主机的步骤之前, 所述 方法还包括步骤:
预先配置所述资源应用特性分配策略, 所述资源应用特性分配策略包括基 于区域分配的资源应用特性分配策略, 基于宿主机分配的资源应用特性分配策 合。
3、如权利要求 2所述的方法, 其特征在于, 根据所述资源请求和资源应用 特性分配策略, 确定向所述资源请求者分配资源的宿主机的步骤包括:
根据所述资源请求中的资源需求、 资源的应用特性以及配置的资源应用特 性分配策略, 在资源池中确定一满足所述资源需求的空闲资源;
当存在满足所述资源需求的空闲资源时, 确定所述空闲资源所在宿主机是 否为分配资源的宿主机。
4、如权利要求 3所述的方法, 其特征在于, 所述根据所述资源请求中的资 源需求、 资源的应用特性以及配置的资源应用特性分配策略, 在资源池中确定 一满足所述资源需求的空闲资源的步骤包括:
当所述资源请求中的资源需求存在区域要求, 或资源提供区域在所述资源 池中的资源权限控制策略中被设定, 或者分配区域被请求的资源的应用特性以 及所述资源应用特性分配策略限定时, 根据所述资源请求中的资源需求和资源 的应用特性定位到对应的区域并确定满足所述资源需求的空闲资源, 否则, 定 位到所有资源池确定满足所述资源需求的空闲资源。
5、如权利要求 3所述的方法, 其特征在于, 当存在满足所述资源需求的空 闲资源时, 确定所述空闲资源所在宿主机是否为分配资源的宿主机的步骤包括 步骤: 定所述空闲资源所在宿主机为分配资源的宿主机;
若所述资源应用特性分配策略中定义了资源应用特性组合策略, 则根据所 述空闲资源所在宿主机上的已分配资源的应用特性、 所述资源请求中的资源的 和所述资源请求中的资源的应用特性可否共存, 当可共存时则确定所述空闲资 源所在宿主机为分配资源的宿主机。
6、如权利要求 5所述的方法, 其特征在于, 所述当可共存时则确定所述空 闲资源所在宿主机为分配资源的宿主机的步骤包括:
根据所述空闲资源所在宿主机上的已分配资源的应用特性、 资源请求中的 资源的应用特性以及所述资源应用特性组合策略, 按照资源共存度从高到低的 顺序选择宿主机上的资源分配给资源请求者。
7、如权利要求 1所述的方法, 其特征在于, 所述资源的资源需求包括资源 所在宿主机位置、 资源的 CPU、 内存、 磁盘、 网络、 操作系统以及应用程序之 一或其任意组合的资源规格;
所述资源的应用特性为资源上所运行的应用的类型, 或者指资源请求者将 对所要申请的资源的使用情况, 所述使用情况包括资源的 CPU使用率、 内存使 用率、 磁盘读写、 网络上下行、 GPU使用率之一或其任意组合。
8、 如权利要求 1所述的方法, 其特征在于, 控制所述宿主机分配资源, 并 返回资源分配信息的步骤之后, 所述方法还包括步骤: 保存所述分配的资源的相关信息, 所述相关信息包括资源的应用特性。
9、如权利要求 1所述的方法, 其特征在于, 控制所述分配资源的宿主机向 资源请求者分配资源, 向资源请求者返回资源分配信息的步骤之后, 所述方法 还包括:
根据预设的策略收集资源的使用信息, 根据资源的使用信息计算所述资源 的应用特性。
10、 如权利要求 9所述的方法, 其特征在于, 所述方法还包括步骤: 资源池存在更优的空闲资源时, 根据预设的资源调整策略进行资源的调整。
11、 一种资源管理平台, 其特征在于, 所述资源管理平台包括: 资源请求接收单元, 用于接收资源请求者发送的资源请求, 其中所述资源 请求包括资源需求和资源的应用特性;
宿主机确定单元, 用于根据资源请求和资源应用特性分配策略, 确定向资 源请求者分配资源的宿主机; 以及
分配信息返回单元, 用于控制所述分配资源的宿主机向资源请求者分配资 源, 向所述资源请求者返回资源分配信息。
12、 如权利要求 11所述的资源管理平台, 其特征在于, 所述资源管理平台 还包括:
策略配置单元, 用于预先配置资源应用特性分配策略, 所述资源应用特性 分配策略包括基于区域分配的资源应用特性分配策略, 基于宿主机分配的资源 应用特性分配策略, 以及基于资源应用特性组合策略的资源应用特性分配策略 之一或其任意组合。
13如权利要求 11所述的资源管理平台, 其特征在于, 所述宿主机确定单 元包括:
空闲资源确定子单元, 用于根据所述资源请求中的资源需求、 资源的应用 特性以及配置的资源应用特性分配策略, 在资源池中确定一满足所述资源需求 的空闲资源; 以及
宿主机确定子单元, 用于当存在满足用户资源需求的空闲资源时, 根据资 源应用特性组合策略,确定所述空闲资源所在宿主机是否为分配资源的宿主机。
14、 如权利要求 11所述的资源管理平台, 其特征在于, 所述资源的资源需 求包括资源所在宿主机位置、 资源的 CPU、 内存、 磁盘、 网络、 操作系统以及 应用程序之一或其任意组合的资源规格;
所述资源的应用特性为资源上所运行的应用的类型, 或者指资源请求者将 对所要申请的资源的使用情况, 所述使用情况包括资源的 CPU使用率、 内存使 用率、 磁盘读写、 网络上下行、 GPU使用率之一或其任意组合。
15、 如权利要求 11所述的资源管理平台, 其特征在于, 所述资源管理平台 还包括:
信息保存单元, 用于保存所述分配的资源的相关信息, 所述相关信息包括 资源的应用特性。
16、 如权利要求 11所述的资源管理平台, 其特征在于, 所述资源管理平台 还包括:
应用特性计算单元, 用于根据预设的策略收集资源的使用信息, 根据资源 的使用信息计算所述资源的应用特性。
17、 如权利要求 11所述的资源管理平台, 其特征在于, 所述资源调整单元 包括:
资源调整单元, 用于当计算得到的应用特性与资源分配请求中资源的应用 特性不同, 或当前的资源池存在更优的空闲资源时, 根据预设的资源调整策略 进行资源的调整。
PCT/CN2012/080467 2011-08-29 2012-08-22 一种资源分配方法及资源管理平台 Ceased WO2013029487A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP12826692.1A EP2725862A4 (en) 2011-08-29 2012-08-22 RESOURCE ALLOCATION METHOD AND RESOURCE MANAGEMENT PLATFORM
US14/184,342 US9442763B2 (en) 2011-08-29 2014-02-19 Resource allocation method and resource management platform

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201110251036.3 2011-08-29
CN201110251036.3A CN102958166B (zh) 2011-08-29 2011-08-29 一种资源分配方法及资源管理平台

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US14/184,342 Continuation US9442763B2 (en) 2011-08-29 2014-02-19 Resource allocation method and resource management platform

Publications (1)

Publication Number Publication Date
WO2013029487A1 true WO2013029487A1 (zh) 2013-03-07

Family

ID=47755313

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2012/080467 Ceased WO2013029487A1 (zh) 2011-08-29 2012-08-22 一种资源分配方法及资源管理平台

Country Status (4)

Country Link
US (1) US9442763B2 (zh)
EP (1) EP2725862A4 (zh)
CN (1) CN102958166B (zh)
WO (1) WO2013029487A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103605574A (zh) * 2013-11-18 2014-02-26 浪潮(北京)电子信息产业有限公司 一种服务器集群中的虚拟机资源调度方法及系统
CN114661448A (zh) * 2022-05-17 2022-06-24 中电云数智科技有限公司 Kubernetes的资源调度方法及调度组件

Families Citing this family (79)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9507612B1 (en) 2012-08-31 2016-11-29 United Services Automobile Association (Usaa) Managing dedicated and floating pool of virtual machines based on demand
JP6114829B2 (ja) * 2012-09-28 2017-04-12 サイクルコンピューティング エルエルシー 仮想環境における演算インフラストラクチャのリアルタイム最適化
CN103167021A (zh) * 2013-02-01 2013-06-19 浪潮(北京)电子信息产业有限公司 资源配置方法和装置
CN103455378A (zh) * 2013-09-05 2013-12-18 华为技术有限公司 一种虚拟化集群的资源分配方法、装置和系统
US20150163223A1 (en) * 2013-12-09 2015-06-11 International Business Machines Corporation Managing Resources In A Distributed Computing Environment
US20150163157A1 (en) * 2013-12-09 2015-06-11 Alcatel-Lucent Usa Inc. Allocation and migration of cloud resources in a distributed cloud system
CN104812002B (zh) * 2014-01-23 2019-02-01 中国科学院计算技术研究所 集中式ran架构下的处理资源动态调整方法、设备及系统
CN104978234A (zh) * 2014-04-14 2015-10-14 北京千橡网景科技发展有限公司 用于动态配置vps资源的方法和设备
CN111666148A (zh) * 2014-04-30 2020-09-15 华为技术有限公司 计算机,控制设备和数据处理方法
CN105307130A (zh) * 2014-06-30 2016-02-03 中兴通讯股份有限公司 一种资源分配方法及系统
CN104158841B (zh) * 2014-07-09 2017-08-15 中电科华云信息技术有限公司 计算资源分配方法
CN104281496B (zh) * 2014-10-27 2018-02-02 深圳中兴网信科技有限公司 虚拟资源的分配方法及分配装置
CN104320481A (zh) * 2014-11-04 2015-01-28 浪潮电子信息产业股份有限公司 一种基于用户体验的虚拟资源动态算法
CN104537045A (zh) * 2014-12-24 2015-04-22 华为技术有限公司 一种基于分布式系统的业务分配方法及装置
CN105808341B (zh) * 2014-12-29 2019-05-28 中国移动通信集团公司 一种资源调度的方法、装置和系统
CN105988872B (zh) * 2015-02-03 2020-02-18 阿里巴巴集团控股有限公司 一种cpu资源分配的方法、装置及电子设备
US9940150B2 (en) * 2015-02-27 2018-04-10 International Business Machines Corporation Policy based virtual resource allocation and allocation adjustment
CN106209746B (zh) * 2015-05-07 2019-12-27 阿里巴巴集团控股有限公司 一种安全业务提供方法及服务器
US9781163B2 (en) * 2015-06-23 2017-10-03 EMC IP Holding Company LLC Trust overlays for disaggregated infrastructures
CN105049257A (zh) * 2015-08-12 2015-11-11 北京因特信安软件科技有限公司 基于可信地理位置信息的云平台调度方法
CN105183561B (zh) * 2015-09-02 2018-09-14 浪潮(北京)电子信息产业有限公司 一种资源分配方法和系统
CN106571935B (zh) * 2015-10-08 2020-01-17 阿里巴巴集团控股有限公司 一种资源调度的方法与设备
CN105263053B (zh) * 2015-10-15 2019-04-30 深圳创维数字技术有限公司 多业务的硬件资源调度方法、系统及数字电视
WO2017072933A1 (ja) * 2015-10-30 2017-05-04 株式会社日立製作所 計算機システムの管理システム及び管理方法
CN105589751B (zh) * 2015-11-27 2019-03-15 新华三技术有限公司 一种物理资源调度方法及装置
CN105630604A (zh) * 2015-12-18 2016-06-01 国云科技股份有限公司 一种基于sla的多租户虚拟机资源分配方法
CN107168777B (zh) * 2016-03-07 2021-04-30 阿里巴巴集团控股有限公司 分布式系统中资源的调度方法以及装置
CN107168788B (zh) * 2016-03-07 2021-01-08 阿里巴巴集团控股有限公司 分布式系统中资源的调度方法以及装置
WO2017166136A1 (zh) 2016-03-30 2017-10-05 华为技术有限公司 一种vnf的资源分配方法及装置
US10303488B2 (en) * 2016-03-30 2019-05-28 Sony Interactive Entertainment Inc. Real-time adjustment of application-specific operating parameters for backwards compatibility
CN106027420A (zh) * 2016-04-27 2016-10-12 乐视控股(北京)有限公司 一种资源模型创建方法和装置
CN105827455A (zh) * 2016-04-27 2016-08-03 乐视控股(北京)有限公司 一种资源模型修正方法和装置
CN105955826A (zh) * 2016-05-10 2016-09-21 广东睿江云计算股份有限公司 一种云主机系统中服务质量的控制方法及装置
KR102199275B1 (ko) * 2016-07-13 2021-01-06 후아웨이 테크놀러지 컴퍼니 리미티드 분산 컴퓨팅 시스템에서의 적응적 리소스 관리
US10733003B2 (en) * 2016-11-03 2020-08-04 Salesforce.Com, Inc. Cost efficient and on-demand pool of running instances in a virtual machine environment
CN106790006B (zh) * 2016-12-13 2020-08-04 北京元心科技有限公司 设备管理方法和系统
CN106686081B (zh) * 2016-12-29 2020-08-28 北京奇虎科技有限公司 一种数据库服务系统的资源分配方法和装置
US10397347B2 (en) 2017-01-03 2019-08-27 International Business Machines Corporation Geolocation-based activation and de-activation of hardware and software functionalities in the cloud
CN107688495B (zh) * 2017-06-22 2020-11-03 平安科技(深圳)有限公司 调度处理器的方法及设备
US10686728B2 (en) * 2017-07-06 2020-06-16 Huawei Technologies Co., Ltd. Systems and methods for allocating computing resources in distributed computing
CN107357661B (zh) * 2017-07-12 2020-07-10 北京航空航天大学 一种针对混合负载的细粒度gpu资源管理方法
CN108121601B (zh) * 2017-11-08 2021-11-05 上海格蒂电力科技有限公司 一种基于权重的应用资源调度装置及方法
CN107729154A (zh) * 2017-11-09 2018-02-23 北京小米移动软件有限公司 资源分配方法及装置
CN107943586A (zh) * 2017-11-21 2018-04-20 深信服科技股份有限公司 一种虚拟化资源优先级控制方法、装置、设备及系统
WO2019135704A1 (en) * 2018-01-08 2019-07-11 Telefonaktiebolaget Lm Ericsson (Publ) Adaptive application assignment to distributed cloud resources
CN108170520A (zh) * 2018-01-29 2018-06-15 北京搜狐新媒体信息技术有限公司 一种云计算资源管理方法及装置
CN108519917B (zh) * 2018-02-24 2023-04-07 国家计算机网络与信息安全管理中心 一种资源池分配方法和装置
CN108388476A (zh) * 2018-03-01 2018-08-10 郑州云海信息技术有限公司 一种服务器计算加速资源池化方法及系统
CN108810107B (zh) * 2018-05-24 2021-07-13 南方电网深圳数字电网研究院有限公司 一种双态it业务的云管平台以及服务方法
CN110647394B (zh) * 2018-06-27 2022-03-11 阿里巴巴集团控股有限公司 一种资源分配方法、装置及设备
CN108924221B (zh) * 2018-06-29 2020-08-25 华为技术有限公司 分配资源的方法和装置
CN108984125B (zh) * 2018-07-17 2022-02-18 郑州云海信息技术有限公司 一种资源分配的方法、装置及存储介质
CN109510877B (zh) * 2018-12-21 2022-03-15 中国航空工业集团公司西安航空计算技术研究所 一种动态资源群的维护方法、装置及存储介质
CN109684065B (zh) * 2018-12-26 2020-11-03 北京云联万维技术有限公司 一种资源调度方法、装置及系统
CN111435317B (zh) * 2019-01-14 2023-04-11 阿里巴巴集团控股有限公司 数据处理方法、计算设备及存储介质
CN111506414B (zh) * 2019-01-30 2023-04-25 阿里巴巴集团控股有限公司 资源调度方法、装置、设备、系统及可读存储介质
CN112395080B (zh) * 2019-08-19 2025-09-05 阿里巴巴集团控股有限公司 基于fpga的资源分配方法、装置、存储介质及计算机设备
CN111338786A (zh) * 2019-08-19 2020-06-26 海通证券股份有限公司 云平台资源的配额管理方法、装置和计算机设备
CN112631766B (zh) * 2019-10-08 2023-09-19 中国移动通信集团浙江有限公司 项目环境资源的动态调整方法及装置
CN112685168B (zh) * 2019-10-17 2023-12-26 中国移动通信集团浙江有限公司 资源管理方法、装置及设备
CN110750330A (zh) * 2019-10-18 2020-02-04 北京浪潮数据技术有限公司 一种虚拟机创建方法、系统、电子设备及存储介质
CN110995863B (zh) * 2019-12-19 2021-07-30 上海交通大学 基于负载需求特征的数据中心负载分配方法及系统
US11301305B2 (en) 2020-01-07 2022-04-12 Bank Of America Corporation Dynamic resource clustering architecture
US11334393B2 (en) 2020-01-07 2022-05-17 Bank Of America Corporation Resource cluster chaining architecture
US10938742B1 (en) 2020-01-31 2021-03-02 Bank Of America Corporation Multiplexed resource allocation architecture
US10860381B1 (en) * 2020-05-14 2020-12-08 Snowflake Inc. Flexible computing
CN111800506B (zh) * 2020-07-06 2023-09-19 深圳市网心科技有限公司 一种边缘计算节点部署方法及相关装置
CN112085419B (zh) * 2020-09-25 2025-01-03 中国建设银行股份有限公司 一种资源获取方法、装置及设备
CN113010309B (zh) * 2021-03-02 2022-10-25 北京达佳互联信息技术有限公司 集群资源调度方法、装置、存储介质、设备和程序产品
US11954524B2 (en) * 2021-05-26 2024-04-09 International Business Machines Corporation Compliance aware application scheduling
CN113760798B (zh) * 2021-08-05 2025-02-18 阿里巴巴创新公司 Rdma设备的分配方法、计算设备及存储介质
CN113656046B (zh) * 2021-08-31 2025-02-21 北京京东乾石科技有限公司 一种应用部署方法和装置
CN113886086B (zh) * 2021-10-15 2025-12-16 济南浪潮数据技术有限公司 云平台计算资源分配方法、系统、终端及存储介质
US12353915B2 (en) * 2022-01-14 2025-07-08 Dell Products L.P. System and method for multi-scheme allocation of virtual resources in an information handling system
CN114064251A (zh) * 2022-01-19 2022-02-18 浪潮云信息技术股份公司 一种函数启动加速方法、装置、设备及可读存储介质
US20230236890A1 (en) * 2022-01-25 2023-07-27 Poplar Technologies, Inc. Apparatus for generating a resource probability model
CN115048222A (zh) * 2022-07-08 2022-09-13 鼎道智联(北京)科技有限公司 硬件逻辑重组方法、装置、系统、电子设备及存储介质
US20260050477A1 (en) * 2022-08-08 2026-02-19 Esh Os Ltd. Anonymous centralized transfer or allocation of resource
WO2025178925A1 (en) * 2024-02-19 2025-08-28 Vaybid Llc System and method for managing a resource

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1713561A (zh) * 2004-06-25 2005-12-28 三星电子株式会社 多载波通信系统中的资源分配方法
CN101969475A (zh) * 2010-11-15 2011-02-09 张军 基于云计算的商业数据可控分发与融合应用系统
CN102004671A (zh) * 2010-11-15 2011-04-06 北京航空航天大学 一种云计算环境下数据中心基于统计模型的资源管理方法
CN102137014A (zh) * 2011-03-11 2011-07-27 华为技术有限公司 资源管理方法、系统和资源管理器

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060294238A1 (en) * 2002-12-16 2006-12-28 Naik Vijay K Policy-based hierarchical management of shared resources in a grid environment
US20050005018A1 (en) * 2003-05-02 2005-01-06 Anindya Datta Method and apparatus for performing application virtualization
US20050149940A1 (en) * 2003-12-31 2005-07-07 Sychron Inc. System Providing Methodology for Policy-Based Resource Allocation
US7620953B1 (en) * 2004-10-05 2009-11-17 Azul Systems, Inc. System and method for allocating resources of a core space among a plurality of core virtual machines
US20060143617A1 (en) * 2004-12-29 2006-06-29 Knauerhase Robert C Method, apparatus and system for dynamic allocation of virtual platform resources
US8112756B2 (en) * 2006-07-20 2012-02-07 Hewlett-Packard Development Company, L.P. System and method for evaluating a workload and its impact on performance of a workload manager
JP5110315B2 (ja) * 2006-05-24 2012-12-26 日本電気株式会社 仮想マシン管理装置、仮想マシン管理方法およびプログラム
US7912955B1 (en) * 2007-04-24 2011-03-22 Hewlett-Packard Development Company, L.P. Model-based provisioning of resources
JP5256744B2 (ja) * 2008-01-16 2013-08-07 日本電気株式会社 資源割当てシステム、資源割当て方法及びプログラム
US20100319004A1 (en) 2009-06-16 2010-12-16 Microsoft Corporation Policy Management for the Cloud
US8694638B2 (en) * 2009-06-29 2014-04-08 Red Hat Israel Selecting a host from a host cluster to run a virtual machine

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1713561A (zh) * 2004-06-25 2005-12-28 三星电子株式会社 多载波通信系统中的资源分配方法
CN101969475A (zh) * 2010-11-15 2011-02-09 张军 基于云计算的商业数据可控分发与融合应用系统
CN102004671A (zh) * 2010-11-15 2011-04-06 北京航空航天大学 一种云计算环境下数据中心基于统计模型的资源管理方法
CN102137014A (zh) * 2011-03-11 2011-07-27 华为技术有限公司 资源管理方法、系统和资源管理器

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP2725862A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103605574A (zh) * 2013-11-18 2014-02-26 浪潮(北京)电子信息产业有限公司 一种服务器集群中的虚拟机资源调度方法及系统
CN114661448A (zh) * 2022-05-17 2022-06-24 中电云数智科技有限公司 Kubernetes的资源调度方法及调度组件

Also Published As

Publication number Publication date
CN102958166B (zh) 2017-07-21
EP2725862A1 (en) 2014-04-30
EP2725862A4 (en) 2014-05-07
CN102958166A (zh) 2013-03-06
US20140173620A1 (en) 2014-06-19
US9442763B2 (en) 2016-09-13

Similar Documents

Publication Publication Date Title
WO2013029487A1 (zh) 一种资源分配方法及资源管理平台
US9183016B2 (en) Adaptive task scheduling of Hadoop in a virtualized environment
US10810045B2 (en) Method and apparatus for allocating central processing unit (CPU) resources in a default resource pool
CN102317917B (zh) 热点域虚拟机cpu调度方法及虚拟机系统
KR101957006B1 (ko) 네트워크 액세스가능 블록 저장을 위한 조정된 승인 제어
US9600337B2 (en) Congestion avoidance in network storage device using dynamic weights
US9304803B2 (en) Cooperative application workload scheduling for a consolidated virtual environment
US9807014B2 (en) Reactive throttling of heterogeneous migration sessions in a virtualized cloud environment
US10489207B2 (en) System for resource management using performance specification and description information of the process
WO2011124077A1 (zh) 虚拟机管理方法、系统及虚拟机管理服务器
WO2016078178A1 (zh) 一种虚拟cpu调度方法
CN103516623A (zh) 资源分配方法及系统
CN109522090B (zh) 资源调度方法及装置
CN107870877B (zh) 用于在存储系统中管理数据访问的方法和系统
CN103440113A (zh) 一种磁盘io资源分配方法和装置
CN106126319B (zh) 中央处理器资源分配方法和计算节点
CN115311123B (zh) 一种像素流gpu资源调度方法及装置
KR102214782B1 (ko) 가상 머신 성능 최적화를 위한 하드웨어 리소스 등급 분류 및 자원 할당방법
KR20150070930A (ko) 최적 비용을 고려한 마이그레이션 자원 할당 시스템 및 할당 방법
CN116320031A (zh) 服务器响应方法、设备及介质
CN120596256A (zh) 应用于存算一体化系统的资源调度系统、方法及电子设备
HK40042985B (zh) 一种资源分配的方法、装置及存储介质
Zheng et al. CLIBE: Precise Cluster-Level I/O Bandwidth Enforcement in Distributed File System

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 12826692

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE