CN115981868A - Resource scheduling method and computing equipment - Google Patents
Resource scheduling method and computing equipment Download PDFInfo
- Publication number
- CN115981868A CN115981868A CN202310127021.9A CN202310127021A CN115981868A CN 115981868 A CN115981868 A CN 115981868A CN 202310127021 A CN202310127021 A CN 202310127021A CN 115981868 A CN115981868 A CN 115981868A
- Authority
- CN
- China
- Prior art keywords
- resource
- vector
- determining
- host
- parameter
- 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.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the application provides a resource scheduling method and computing equipment. Wherein a resource loss model is determined in response to a resource scheduling request; calculating loss values of different hosts by using the resource loss model based on the allocable resource vectors of the different hosts; the allocable resource vector is composed of resource components of a plurality of resource dimensions; determining a candidate host set with a loss value meeting the scheduling requirement; and determining a target host for allocating resources for the resource scheduling request from the candidate host set. The technical scheme provided by the embodiment of the application improves the resource allocation rate and reduces the resource cost.
Description
Technical Field
The embodiment of the application relates to the technical field of computers, in particular to a resource scheduling method and computing equipment.
Background
In a cloud computing scenario, cluster resource scheduling is involved, so that in a cluster consisting of a large number of physical servers (also referred to as "hosts", "physical machines"), a suitable host is selected to allocate resources for a cloud computing product such as a virtual machine, a container or a database and run the cloud computing product.
The resources provided by the host include a Central Processing Unit (CPU), a memory, and storage hardware resources with multiple dimensions. In the prior art, a host capable of providing required resources is generally randomly selected to allocate resources to a cloud computing product, but the resource scheduling mode is inaccurate, so that the resources cannot be effectively allocated, and the resource cost is high.
Disclosure of Invention
The embodiment of the application provides a resource scheduling method and computing equipment, which are used for solving the technical problems of low resource allocation rate and high resource cost in the prior art.
In a first aspect, an embodiment of the present application provides a resource scheduling method, including:
in response to a resource scheduling request, determining a resource loss model;
calculating loss values of different hosts by using the resource loss model based on the allocable resource vectors of the different hosts; the allocable resource vector is composed of resource components of a plurality of resource dimensions;
determining that a loss value satisfies a schedule a set of required candidate hosts;
and determining a target host for allocating resources for the resource scheduling request from the candidate host set.
In a second aspect, embodiments of the present application provide a computing device comprising a processing component and a storage component;
the storage component stores one or more computer instructions; the one or more computer instructions are for execution by the processing component to implement the resource scheduling method as described in the first aspect above.
In a third aspect, an embodiment of the present application provides a computer storage medium, which stores a computer program, and when the computer program is executed by a computer, the computer program implements the resource scheduling method according to the first aspect.
In the embodiment of the application, a resource loss model is constructed, so that for a resource scheduling request, based on allocable resource vectors of different hosts, loss values of the different hosts are calculated by using the resource loss model; the allocable resource vector is composed of resource components of a plurality of resource dimensions; determining a candidate host set with a loss value meeting the scheduling requirement; so that the target host for allocating resources for the resource scheduling request can be determined from the candidate host set. The resource loss of the host is evaluated by the resource loss model based on the vectorization resource loss model and the allocable resource vector, so that a candidate host set with a loss value meeting the scheduling requirement can be screened, a target host is selected from the candidate host set to allocate resources for the resource scheduling request, the allocable resource vector comprises a plurality of resource dimensions, and the resource scheduling is performed by comprehensively considering the resource dimensions, so that the accuracy of the resource scheduling is improved, the resource allocation rate is improved, and the resource cost is reduced.
These and other aspects of the present application will be more readily apparent from the following description of the embodiments.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a schematic diagram illustrating an embodiment of a cloud computing system provided herein;
FIG. 2 is a flow chart illustrating one embodiment of a resource scheduling method provided herein;
FIG. 3 is a flow chart illustrating a method for scheduling resources according to yet another embodiment of the present application;
FIG. 4 is a schematic diagram of an interaction field in a practical application of the embodiment of the present application;
fig. 5 is a schematic structural diagram illustrating an embodiment of a resource scheduling apparatus provided in the present application;
FIG. 6 illustrates a schematic structural diagram of one embodiment of a computing device provided herein.
Detailed Description
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application.
In some of the flows described in the specification and claims of this application and in the above-described figures, a number of operations are included that occur in a particular order, but it should be clearly understood that these operations may be performed out of order or in parallel as they occur herein, the number of operations, e.g., 101, 102, etc., merely being used to distinguish between various operations, and the number itself does not represent any order of performance. Additionally, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first", "second", etc. in this document are used for distinguishing different messages, devices, modules, etc., and do not represent a sequential order, nor limit the types of "first" and "second" to be different.
The technical solution of the embodiment of the present application may be applied to an application scenario involving cluster resource scheduling, for example, in a cloud computing scenario, the cluster resource scheduling refers to selecting a suitable host from a cluster formed by a plurality of physical servers (also referred to as "hosts" and "physical machines") to perform resource allocation based on a resource required by a request.
Taking a cloud computing scenario as an example, cloud computing is one of the fastest growing trends in computer technology, and relates to providing hosted services over a network. Cloud computing environments provide computing and storage resources as services to end users. The end user may make a request to the offered service for processing. The processing power of a service is typically limited by the configuration resources.
Cloud computing is a service delivery model intended to enable on-demand network access to a shared pool of configurable computing resources (e.g., networks, network bandwidth, servers, processing, memory, storage, applications, virtual machines, and services) that can be deployed and released quickly with minimal administrative effort or interaction with the service provider. The resources referred to in this application may generally refer to hardware resources, although other types of computing resources are not excluded.
It should be understood that although the present application includes detailed descriptions regarding cloud computing, implementations of the teachings described herein are not limited to cloud computing environments. Rather, embodiments of the application can be implemented in connection with any other type of computing environment, whether now known or later developed.
In a cloud computing scenario, resources provided by multiple hosts form a resource pool to provide services for multiple tenants. The user referred to herein means a tenant, and may purchase resources of a cloud computing provider through funding to build a cloud computing product meeting the needs of the user, such as a cloud server, a cloud database, a container, or a virtual machine, and the user may register a user account at the cloud computing provider, and the cloud computing provider may distinguish different users through the user account.
One or more cloud computing products can run in one host, one cloud computing product runs in one host, and when the cloud computing product is created, resource scheduling operation needs to be executed to select a proper host to allocate corresponding resources so as to create the cloud computing product.
The current resource scheduling mode is generally that a host capable of providing required resources is randomly selected to allocate resources to a cloud computing product, and the inventor finds that the host can generally provide multiple resources, such as a CPU, a memory, storage, and the like, in practical application, after some resource allocation is used up in one host, another resource is still idle, and more idle resources occur, which are fragmented, resulting in host resource waste and failure to be fully utilized, and finally resulting in a reduction in resource allocation rate of the host, and multiple resources are unbalanced, and resource cost is high.
In order to improve the resource allocation rate, reduce the resource cost and the like, the inventor provides the technical scheme of the application through a series of researches, in the embodiment of the application, a resource loss model based on vectorization is constructed, the resource loss of a host is evaluated by using the resource loss model based on an allocable resource vector, so that a candidate host set with a loss value meeting the scheduling requirement can be screened, a target host is selected from the candidate host set to allocate resources for a resource scheduling request, the allocable resource vector comprises a plurality of resource dimensions, the resource scheduling is performed by comprehensively considering the resource dimensions, the resource fragmentation phenomenon can be reduced, the resource balance is kept, the accuracy of the resource scheduling is improved, the resource allocation rate is improved, and the resource cost is reduced.
The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In a cloud computing scenario, the technical solution of the embodiment of the present application may be applied to a cloud computing system as shown in fig. 1, where the cloud computing system includes one or more cloud computing nodes 10, the one or more cloud computing nodes 10 include a physical server, i.e., a host, and are shown in the form of a general purpose computing device in the figure, and the cloud computing nodes may include one or more processing components, storage components, and the like, and buses and the like that couple other system components including the storage components to the processing components.
A user may communicate with cloud computing node 10 using a local computing device, such as a Personal Digital Assistant (PDA) or cellular telephone 20A, a desktop computer 20B, a laptop computer 20C, and/or an automobile computer system 20N. Cloud computing nodes 10 may also communicate with each other, and they may be grouped physically or virtually in one or more networks (not shown), such as a private, community, public, or hybrid cloud, or a combination thereof. This allows the cloud computing system to provide the infrastructure, platform, and/or software as a service without the user needing to maintain resources on the local computing device. It should be understood that the type of computing device shown in fig. 1 is for illustration only, and that cloud computing node 10 may communicate with any type of computerized device over any type of network and/or network addressable connection (e.g., using a web browser).
A user, through a local computing device, may request a cloud computing system to create a cloud computing product, a creation task may be performed by one or more cloud computing nodes, a target host is selected by performing a resource scheduling operation to create a cloud computing product, and so on. In addition, the cloud computing system may migrate, for example, a cloud computing product operated by a certain host in the event of a failure of the host, and at this time, may also execute the resource scheduling operation provided in the embodiment of the present application to select a target host to migrate the cloud computing product, and the like.
The hosts provided by the cloud computing system can be divided according to areas, and the like, so that more efficient services can be provided for users in different areas, and the like.
It should be noted that the resource scheduling method provided in the embodiment of the present application is generally executed by one or more cloud computing nodes in a cloud computing system, and a corresponding scheduling apparatus is generally disposed in the cloud computing nodes.
It should be noted that, in the embodiments of the present application, the user data may be used, and in practical applications, the user-specific personal data may be used in the schemes described herein within the scope permitted by applicable laws and regulations under the condition of meeting the requirements of applicable laws and regulations in the country (for example, clear agreement of the user, actual notification to the user, and the like).
It should be understood that the number of cloud computing nodes in fig. 1 is merely illustrative. There may be any number of cloud computing nodes, as desired for implementation.
The details of implementation of the technical solution of the embodiments of the present application are set forth in the following.
Fig. 2 is a flowchart of an embodiment of a resource scheduling method provided in an embodiment of the present application, where the method may include the following steps:
201: in response to a resource scheduling request, a resource loss model is determined.
The resource scheduling request may be generated for the cloud computing product to be created based on the user creation request, or may be generated for the cloud computing product requested to be migrated based on the migration request of the cloud computing product.
The resource scheduling request may include resource components of multiple resource dimensions, where different resource dimensions correspond to different types of resources, such as a processor, for example, a CPU, a memory, and a storage, and a resource component of a certain resource dimension is a resource request amount of a corresponding resource type, and the resource components of the multiple resource dimensions may generate a request resource vector, and may perform calculation processing in a vectorization manner, and the like.
The resource loss model may be constructed in advance, and optionally, may be constructed based on at least one parameter affecting resource scheduling; thus, at least one parameter affecting the scheduling of resources may be included; the at least one parameter affecting resource scheduling may include, for example, measurement parameters for measuring allocable resources, resource balance, and resource fragmentation, respectively. Allocable resources may measure the situation of allocable resources in a host; the resource balance degree is used for measuring the distribution condition of different types of resources in the host; the resource fragmentation degree is used for measuring the allocation condition of idle resources in the host, and the like, and the at least one resource scheduling influence factor influences the resource allocation rate, the idle resource residual amount and the like of the host.
The resource allocation rate referred to herein may refer to a resource sale rate, where the resource sale rate represents a ratio of a total amount of sold resources of a certain kind to a total amount of resources of a non-empty host in a cluster, and the non-empty host refers to a host in the cluster that has performed resource scheduling.
The resource loss model constructed based on the at least one parameter affecting resource scheduling may evaluate the resource loss, for example, the larger the loss value is, the more idle resources are indicated, the lower the resource allocation rate is, the smaller the loss value is, the less idle resources are indicated, and the higher the resource allocation rate is.
Wherein the metric parameter may be generated based on a resource vector parameter, the resource vector parameter being composed of resource components of a plurality of resource dimensions.
202: and calculating loss values of different hosts by using a resource loss model based on the allocable resource vectors of the different hosts.
Wherein the allocable resource vector is composed of resource components of a plurality of resource dimensions. For example, X = { r 1 ,r 2 ,…，r d X denotes an allocable resource vector, r i I.e. the resource component representing the ith resource dimension, i ≦ d ≦ 1.
The resource component of each resource dimension in the allocable resource vector is also the remaining resource amount available for allocation in the resource class corresponding to the resource dimension in the host.
The allocable resource vector may be input as a resource vector parameter value to a resource loss model to calculate a resulting loss value. For example, the resource loss model is represented as L (R), R represents a resource vector parameter, and the allocable resource vector may be input as a resource vector parameter value to the resource loss model, i.e., L (X) is calculated to obtain a loss value.
In addition, as another alternative, the loss values of different hosts may be calculated by using a resource loss model based on the allocable resource vector and the request resource vector of different hosts. Thus, optionally, the calculating the loss value of the different hosts using the resource loss model based on the request resource vector and the allocable resource vectors of the different hosts may include:
determining the residual resource vectors of the allocable resource vectors of different hosts after the request resource vectors are respectively removed; and inputting the residual resource vectors into a resource loss model, and calculating to obtain loss values of different hosts.
For example, the allocable resource vector is subtracted by the request resource vector to obtain a residual resource vector; and then, inputting the residual resource vector as a resource vector parameter value into a resource loss model, and calculating to obtain loss values of different hosts. For example, the resource loss model is represented as L (R), R represents a resource vector parameter, and the allocable resource vector X subtracts the request resource vector M to obtain a residual resource vector X-M; the remaining resource vector may be input to the resource loss model as resource vector parameter values, i.e. L (X-M) is calculated to obtain the loss value. I.e. the loss value may be the loss value that would be incurred after allocating resources for a resource scheduling request in the host. The smaller the loss value is, the smaller the idle resource is after the host allocates the resource for the resource scheduling request, and the higher the resource allocation rate is.
203: and determining a candidate host set with the loss value meeting the scheduling requirement.
204: and determining a target host for allocating resources for the resource scheduling request from the candidate host set.
According to the loss value used for evaluating the host resource loss, hosts with loss values meeting the scheduling requirement can be screened so as to form a candidate host set, and then a target host for allocating resources for the resource scheduling request can be determined in the candidate host set.
The scheduling requirement can be set according to actual conditions, and can be determined according to the relationship between the loss value and the amount of the idle resources and the resource allocation rate, so that the host with the loss value meeting the scheduling requirement can ensure that the idle resources are smaller and the resource allocation rate is higher.
As an alternative, the smaller the loss value is, the smaller the idle resource is, and the higher the resource allocation rate is, the determining that the loss value satisfies the candidate host set of the scheduling requirement may include: a set of candidate hosts having a penalty value less than or equal to a first penalty threshold is determined.
The first loss threshold may be set in conjunction with the actual situation, which may be greater than or equal to a value of 0. That is, the host corresponding to the loss value between 0 and the first loss threshold may be selected as the candidate host. The first loss threshold may be, for example, a minimum loss value among loss values corresponding to the respective hosts.
As another alternative, the smaller the loss value, the smaller the free resource and the higher the resource allocation rate, and the determining that the loss value satisfies the candidate host set of the scheduling requirement may include:
a set of candidate hosts is selected for which the difference between the penalty value and the minimum penalty value is less than or equal to a second penalty threshold.
The second loss threshold may be set in combination with an actual situation, and is used to screen candidate hosts whose loss values approach the minimum loss value, and optionally, a sum of the minimum loss value and the second loss threshold may be equal to the first loss threshold.
The set of candidate hosts for which the loss value is determined to be less than or equal to the first loss threshold may be represented by the following equation:
wherein L represents a resource loss model; x c An allocable resource vector representing a candidate host; x i Indicating assignability of ith hostA resource vector; m represents a request resource vector; m represents the number of hosts; ε represents the second loss threshold.
In the embodiment, the resource loss of the hosts is evaluated by the resource loss model based on the allocable resource vector through the constructed resource loss model, so that a candidate host set with a loss value meeting the scheduling requirement can be screened, the target host is selected from the candidate host set to allocate resources for the resource scheduling request, the allocable resource vector comprises a plurality of resource dimensions, the resource scheduling is performed by comprehensively considering the plurality of resource dimensions, the resource fragmentation phenomenon can be reduced, the resource balance is kept, the accuracy of the resource scheduling is improved, the resource allocation rate is improved, and the resource cost is reduced.
In order to further improve the resource allocation rate and reduce the resource cost, as shown in another embodiment of the resource scheduling method shown in fig. 3, the scheduling method may include the following steps:
301: in response to a resource scheduling request, a resource loss model is determined.
302: and calculating loss values of different hosts by using the resource loss model based on the allocable resource vectors of different hosts.
An allocable resource vector is made up of resource components of multiple resource dimensions.
303: and determining a candidate host set with the loss value meeting the scheduling requirement.
304: and determining a target host with resource components of multiple resource dimensions meeting balance requirements from the candidate host set based on the request resource vector of the resource scheduling request.
Wherein the request resource vector is composed of resource components of a plurality of resource dimensions.
The operations in step 301 to step 303 may be detailed in the operations in step 201 to step 203 in the embodiment shown in fig. 2, and are not repeated here.
In this embodiment, in the determined hosts that satisfy the scheduling requirement from the overall perspective, the resource components in the multiple resource dimensions may still be unbalanced, and the requested resources of the resource scheduling request in the multiple resource dimensions cannot be effectively satisfied. The target host can further ensure the resource balance degree and the like, further improve the resource allocation rate, reduce the problem of resource fragmentation and the like, and further reduce the resource cost.
As an optional way, determining a plurality of target hosts whose resource dimensions satisfy the balancing requirement from the candidate host set may include:
and determining the included angle between the residual resource vector of each candidate host in the candidate host set and the main direction vector, and selecting the candidate host with the minimum included angle as the target host.
The main direction vector may refer to a unit vector having the same positive angle with each coordinate axis. The remaining resource vector may refer to the allocable resource vector minus the request resource vector.
Alternatively, the angle between the residual resource vector and the main direction vector may be calculated according to the following angle calculation formula:
the above representation mode for selecting the candidate host with the smallest included angle as the target host may be:
wherein, X belongs to hosts, hosts = { A, B, C, ... }; theta is belonged to 0, pi/2.
Where θ (X-M) represents the residual resource vector and the principal direction vectorAngle of (d) r s Representing a target host, and X representing an allocable resource vector of a candidate host; m represents a request resource vector; p represents a principal direction vector; X-M represents a residual resource vector; II X M II represents the modulus value of the remaining resource vector; hosts represent the candidate host set, and A, B and C represent the candidate hosts.
The included angle is smaller, the resource allocation of each resource dimension representing the residual resource vector is closer to the main direction vector, and at the moment, the corresponding loss value is smaller, the corresponding idle resource is smaller, and the resource allocation rate is higher.
Further, as another alternative, determining a plurality of target hosts from the candidate host set whose resource dimensions meet the balance requirement may include: and determining the included angle between the allocable resource vector of each candidate host in the candidate host set and the main direction vector, and selecting the candidate host with the minimum included angle as the target host.
That is, through the two optional manners, the target host with smaller resource loss may be selected to perform resource scheduling, so as to preferentially select the host with smaller resource loss to perform resource scheduling, or the target host with smaller resource loss may be selected to perform resource scheduling after allocating resources for the resource scheduling request, so as to ensure that the target host after performing resource scheduling has smaller resource loss, and so on, and ensure the resource allocation rate after performing resource scheduling for the target host.
The calculation of the included angle between the allocable resource vector and the main direction vector may refer to the calculation of the included angle between the remaining resource vector and the main direction vector, which is not repeated herein.
As yet another alternative, determining a plurality of target hosts from the set of candidate hosts for which the resource dimension meets the balancing requirement may include:
and determining a target resource dimension with the minimum resource residual amount in the residual resource vectors of all candidate hosts in the candidate host set, and selecting the candidate host with the maximum resource residual amount corresponding to the target resource dimension as the target host.
The target resource dimensions corresponding to different candidate hosts may be different. For example, assuming that the candidate host set includes three candidate hosts a, B, and C, each candidate host includes three resource dimensions u, v, and w, the target resource dimension with the minimum resource remaining amount of the candidate host a is u, the corresponding resource remaining amount is 10, the target resource dimension with the minimum resource remaining amount of the candidate host B is v, the corresponding resource remaining amount is 20, the target resource dimension with the minimum resource remaining amount of the candidate host C is w, and the corresponding resource remaining amount is 15, it is known that the target resource dimension u of the candidate host a has the minimum resource remaining amount, and the candidate host a may be the target host.
It should be noted that, because the statistical units corresponding to different resource dimensions may be different, for convenience of calculation processing and the like, the resource amount referred to in the embodiment of the present application may specifically refer to a numerical value after normalization to facilitate comparison and the like, and the specific normalization method is not limited in the present application.
The candidate host with the largest resource residual amount corresponding to the selected target resource dimension may be represented by the following formula:
wherein r is s Representing a target host, X representing an allocable resource vector of a candidate host; m represents a request resource vector; X-M represents a residual resource vector; (X-M) (i) represents a resource component of an ith resource dimension in the residual resource vector; d represents the number of resource dimensions.
The resources corresponding to the target resource dimension are also the bottleneck resources of the host, the host with the largest bottleneck resource is preferentially selected for resource scheduling, the overall allocation of the host resources can be improved, the fairness is better, and the overall resource allocation rate is better.
As yet another alternative, determining a target host from the candidate host set for which the plurality of resource dimensions meet the balancing requirement may include:
and determining the similarity between the allocable resource vector and the request resource vector of each candidate host in the candidate host set, and selecting the candidate host with the maximum similarity as the target host.
The similarity between the allocable resource vector and the request resource vector may be calculated in various manners, for example, a dot product of the allocable resource vector and the request resource vector of each candidate host in the candidate host set is calculated, and a result of the dot product is used as the similarity between the allocable resource vector and the request resource vector of each candidate host. The candidate host with the largest similarity can be selected as the target host by adopting the following formula:
of course, the similarity may also be calculated in other manners, for example, calculating a cosine distance or a euclidean distance, etc. to represent the similarity between two vectors, which is not limited in this application.
The larger the click is, the more similar the request resource vector and the allocable resource vector of the host are, the more matched the resource request quantity and the allocable resource quantity is, and the better the overall resource allocation rate can be ensured.
In some embodiments, the resource loss model may be constructed as follows:
determining at least one parameter affecting resource scheduling;
and constructing a resource loss model based on at least one parameter influencing resource scheduling and the weight coefficients respectively corresponding to the parameters.
The weight coefficients corresponding to the parameters that affect the resource scheduling may be set in combination with actual conditions, for example, the weight coefficients may be set in combination with the degrees of influence of the different parameters on the resource loss, the resource scheduling influencing factor with a larger influence degree may be set as a larger weight coefficient, and the resource scheduling influencing factor with a smaller influence degree may be set as a smaller weight coefficient. As can be seen from the above description, the allocable resource, the resource balance, and the resource fragmentation degree may affect resource scheduling, and the at least one parameter affecting resource scheduling may include measurement parameters for measuring the allocable resource, the resource balance, and the resource fragmentation degree, respectively, and therefore, the weight coefficient may also be determined in combination with different resource scheduling manners, for example, a "compact" resource scheduling manner in which resource allocation of one host is preferentially exhausted and then resource scheduling is performed on another host is adopted, and then the weight coefficient corresponding to the measurement parameter of the allocable resource may be set to a larger weight coefficient, and for example, in a "scattered" resource scheduling manner in which resource scheduling is performed in a plurality of hosts, the weight coefficient corresponding to the measurement parameter of the allocable resource may be set to a smaller weight coefficient, and other measurement parameters are set to a larger weight coefficient, and the like.
As an alternative, determining at least one parameter affecting resource scheduling may include:
determining a measurement parameter for measuring the allocable resource based on the modulus value of the resource vector parameter;
determining a measurement parameter for measuring the resource balance degree based on the included angle between the resource vector parameter and the main direction vector;
and determining a measurement parameter for measuring the fragmentation degree of the resource based on the resource components of different resource dimensions in the resource vector parameter.
Alternatively, for example, the modulus value of the resource vector parameter may be specifically used as a measurement parameter for measuring the allocable resource; of course, the square of the modulus of the resource vector parameter may be used as a metric parameter for measuring the allocable resource.
For example, an angle between the resource vector parameter and the principal direction vector may be used as a measurement parameter for measuring the resource balance.
For example, the sum of the resource components of different resource dimensions in the resource vector parameter and the remainder of the fragmentation degree coefficient may be specifically used as the measurement parameter for measuring the fragmentation degree of the resource. Of course, the sum of the resource components of the multiple resource dimensions may be used as a measurement parameter for measuring the fragmentation degree of the resource, or the sum of the product values of the resource components of the multiple resource dimensions and the fragmentation degree coefficient may be used as a measurement parameter for measuring the fragmentation degree of the resource.
Alternatively, the resource loss model may be represented by the following loss function, for example:
L(R)＝α‖R‖+βθ(R)+γ∑ 1≤i≤d r i modfi；
wherein the content of the first and second substances,
wherein L (R) represents a resource loss model, R represents a resource vector parameter, | R | represents a modulus value of the resource vector parameter, R i The resource component of the ith resource dimension is represented, fi represents a fragmentation degree coefficient corresponding to the ith resource dimension, and d represents the number of the resource dimensions; p represents a principal direction vector; α, β, and γ represent weight coefficients, respectively. The fragmentation degree coefficients corresponding to different resource dimensions may be the same, and certainly may also be different.
As can be seen from the above description, the resource vector parameter may refer to an allocable resource vector in the host, or a remaining resource vector obtained by subtracting the request resource vector from the allocable resource vector in the host.
Furthermore, as another alternative, determining at least one parameter affecting resource scheduling may include:
determining a measurement parameter for measuring the allocable resource based on the modulus value of the resource vector parameter;
determining a measurement parameter for measuring the resource balance degree based on the resource component difference value of any two resource dimensions;
and determining a measurement parameter for measuring the fragmentation degree of the resource based on the resource components of different resource dimensions.
For example, the square of the modulus value of the residual resource vector can be used as a measurement parameter for measuring the resource allocation rate; or the residual resource vector module value can be used as a measurement parameter for measuring the resource allocation rate and the like,
for example, a sum corresponding to a square of the difference values of the plurality of resource components may be used as a metric parameter for measuring the resource balance;
for example, the sum of the resource components of multiple resource dimensions may be used as a measurement parameter for measuring the fragmentation degree of the resource, or the sum of the product values of the resource components of multiple resource dimensions and the fragmentation degree coefficient may be used as a measurement parameter for measuring the fragmentation degree of the resource, or the sum of the resource components of different resource dimensions in the resource vector parameter and the remainder of the fragmentation degree coefficient may be used as a measurement parameter for measuring the fragmentation degree of the resource, or the like.
Alternatively, the resource loss model may be represented by the following loss function, for example:
wherein L (R) represents a resource loss model, R represents a resource vector parameter, | R | represents a modulus value of the resource vector parameter, R i Resource component representing the ith resource dimension, r u A resource component representing a u-th resource dimension; r is v Resource components representing v resource dimensions; fi represents a fragmentation degree coefficient corresponding to the ith resource dimension, and d represents the number of the resource dimensions; α, β, and γ represent weight coefficients, respectively. The fragmentation degree coefficients corresponding to different resource dimensions may be the same, and certainly may also be different.
As can be seen from the above description, the resource vector parameter may refer to an allocable resource vector in the host, or a remaining resource vector obtained by subtracting the request resource vector from the allocable resource vector in the host.
Further, after determining the target host, in some embodiments, the method may further include:
and carrying out resource allocation for the resource scheduling request in the target host.
The resource scheduling request can be a request for resource allocation of resource components of multiple resource dimensions in a resource vector, and in a cloud computing scene, after resource allocation is performed, a cloud computing product and the like corresponding to the resource scheduling request can be created in a target host.
To facilitate understanding, a cloud computing scenario is taken as an example to introduce the technical solution of the present application, the cloud computing product may be, for example, a cloud database, fig. 4 shows an interaction schematic diagram in the cloud computing scenario, and the first cloud computing node 10 may receive a resource scheduling request, where the resource scheduling request may be triggered by a user using the local computing device 20, and of course, may also be generated based on the cloud database to be migrated.
The cloud computing system may include a cluster system of a plurality of cloud computing nodes, each of which may include a host to provide resources. Any one or a designated cloud computing node may be responsible for performing resource scheduling processing and the like.
The first cloud computing node 10, in response to the resource scheduling request, may determine a resource loss model constructed based on at least one parameter affecting resource scheduling; then, based on the allocable resource vector and the request resource vector of different hosts, calculating loss values of different hosts by using a resource loss model; then, determining a candidate host set with loss values meeting the scheduling requirements, and determining a target host with resource components of multiple resource dimensions meeting balance requirements from the candidate host set based on a request resource vector of a resource scheduling request; the specific implementation can be described in the corresponding embodiments above, and is not described herein again.
After the first cloud computing node 10 determines the target host, the resource scheduling request may be sent to the second cloud computing node 30 corresponding to the target host, and the second cloud computing node 30 performs resource allocation for the resource scheduling request and creates a corresponding cloud database.
According to the technical scheme of the embodiment of the application, the resource loss of the hosts is evaluated by the resource loss model based on the vectorization-based resource loss model and the allocable resource vectors, so that the candidate host set with the loss value meeting the scheduling requirement can be screened, the target hosts are selected from the candidate host set to allocate resources for the resource scheduling request, the allocable resource vectors comprise a plurality of resource dimensions, the resource scheduling is carried out by comprehensively considering the resource dimensions, the resource fragmentation phenomenon can be reduced, the resource balance is kept, the accuracy of the resource scheduling is improved, the resource allocation rate is improved, and the resource cost is reduced.
Fig. 5 is a schematic structural diagram of an embodiment of a resource scheduling apparatus according to an embodiment of the present application, where the apparatus may include:
a request response module 501, configured to determine a resource loss model in response to a resource scheduling request;
a loss calculating module 502, configured to calculate loss values of different hosts by using a resource loss model based on allocable resource vectors of the different hosts; the allocable resource vector is composed of resource components of a plurality of resource dimensions;
a first determining module 503, configured to determine a candidate host set whose loss value satisfies a scheduling requirement;
a second determining module 504, configured to determine a target host that allocates resources for the resource scheduling request from the candidate host set.
In some embodiments, the second determining module may specifically determine, from the candidate host set, a target host whose resource components of the multiple resource dimensions meet the balancing requirement based on a request resource vector of the resource scheduling request; wherein the request resource vector is composed of resource components of a plurality of resource dimensions.
In some embodiments, the loss calculation module may specifically calculate the loss values of the different hosts using a resource loss model based on the request resource vector and the allocable resource vectors of the different hosts.
In some embodiments, the apparatus may further comprise:
and the resource allocation module is used for allocating resources for the resource scheduling request in the target host.
In some embodiments, the loss calculating module may specifically determine remaining resource vectors of allocable resource vectors of different hosts after respectively removing the request resource vector; and inputting the residual resource vectors into a resource loss model, and calculating to obtain loss values of different hosts.
In some embodiments, the second determining module may specifically determine an included angle between an allocable resource vector or a remaining resource vector of each candidate host in the candidate host set and the principal direction vector, and select the candidate host with the smallest included angle as the target host;
determining a target resource dimension with the minimum resource residual amount in the residual resource vectors of all candidate hosts in the candidate host set, and selecting the candidate host with the maximum resource residual amount corresponding to the target resource dimension as the target host;
alternatively, the first and second liquid crystal display panels may be,
and determining the similarity between the allocable resource vector and the request resource vector of each candidate host in the candidate host set, and selecting the candidate host with the maximum similarity as the target host.
In some embodiments, the determining the similarity between the allocable resource vector and the request resource vector of each candidate host in the candidate host set by the second determining module, and the selecting the candidate host with the highest similarity as the target host may include: calculating dot products of the allocable resource vectors and the request resource vectors of the candidate hosts in the candidate host set, and taking dot product results as the similarity of the allocable resource vectors and the request resource vectors of the candidate hosts; and selecting the candidate host with the maximum similarity as the target host.
In some embodiments, the apparatus may further comprise:
a model building module for determining at least one parameter affecting resource scheduling; and constructing a resource loss model based on at least one parameter influencing resource scheduling and the respectively corresponding weight coefficients.
In some embodiments, the at least one parameter affecting resource scheduling comprises a metric parameter measuring allocable resources, a resource balance degree and a resource fragmentation degree respectively; the model building module determining at least one parameter affecting the resource scheduling may comprise: determining a measurement parameter for measuring the allocable resource based on the modulus value of the resource vector parameter; determining a measurement parameter for measuring the resource balance degree based on an included angle between the resource vector parameter and the main direction vector; and determining a measurement parameter for measuring the fragmentation degree of the resource based on the resource components of different resource dimensions in the resource vector parameter.
In some embodiments, the model building module determines a metric parameter that measures the allocable resource based on the norm value of the allocable resource vector comprises: taking the modulus value of the resource vector parameter as a measurement parameter for measuring the allocable resource;
the model building module determines a measurement parameter for measuring the resource balance degree based on an included angle between the resource vector and the main direction vector, and the measurement parameter comprises the following steps: taking the included angle between the resource vector parameter and the main direction vector as a measurement parameter for measuring the resource balance degree;
the model building module determines a measurement parameter for measuring the fragmentation degree of the resource based on the resource components of different resource dimensions in the resource vector, and the measurement parameter comprises the following steps: and taking the sum value corresponding to the resource components of different resource dimensions and the surplus result of the fragmentation degree coefficient in the resource vector parameters as a measurement parameter for measuring the fragmentation degree of the resources.
In some embodiments, the at least one parameter affecting resource scheduling comprises a metric parameter measuring allocable resources, a resource balance degree and a resource fragmentation degree respectively; the model building module determining at least one parameter affecting resource scheduling comprises: determining a measurement parameter for measuring the allocable resources based on the modulus value of the residual resource vector; determining a measurement parameter for measuring the resource balance degree based on the resource component difference value of any two resource dimensions; and determining a measurement parameter for measuring the fragmentation degree of the resources based on the resource components of different resource dimensions.
In some embodiments, the first determination module may specifically determine a set of candidate hosts for which the penalty value is less than or equal to a first penalty threshold; alternatively, a set of candidate hosts for which the difference between the penalty value and the minimum penalty value is less than or equal to the second penalty threshold is selected.
The resource scheduling apparatus shown in fig. 5 may execute the resource scheduling method described in the embodiment shown in fig. 2, and details of the implementation principle and the technical effect are not repeated. The specific manner in which each module and unit of the resource scheduling apparatus in the above embodiments perform operations has been described in detail in the embodiments related to the method, and will not be described in detail here.
Embodiments of the present application also provide a computing device, as shown in fig. 6, which may include a storage component 601 and a processing component 602;
the storage component 601 stores one or more computer instructions for execution by the processing component 602 to implement the resource scheduling method shown in fig. 2.
Of course, a computing device may also necessarily include other components, such as input/output interfaces, communication components, and so forth.
The input/output interface provides an interface between the processing components and peripheral interface modules, which may be output devices, input devices, etc. The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, and the like.
Wherein the processing components may include one or more processors executing computer instructions to perform all or part of the steps of the above-described method. Of course, the processing elements may also be implemented as one or more Application Specific Integrated Circuits (ASICs), digital Signal Processors (DSPs), digital Signal Processing Devices (DSPDs), programmable Logic Devices (PLDs), field Programmable Gate Arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic components configured to perform the above-described methods.
The storage component is configured to store various types of data to support operations at the terminal. The memory components may be implemented by any type or combination of volatile and non-volatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.
The display element may be an Electroluminescent (EL) element, a liquid crystal display or a microdisplay having a similar structure, or a retina-directable display or similar laser scanning type display.
It should be noted that the computing device may be a physical device or a flexible computing host provided by a cloud computing platform. It can be implemented as a distributed cluster consisting of a plurality of servers or terminal devices, or as a single server or a single terminal device.
An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a computer, the resource scheduling method in the embodiment shown in fig. 2 may be implemented. The computer-readable medium may be included in the electronic device described in the above embodiment; or may exist separately without being assembled into the electronic device.
An embodiment of the present application further provides a computer program product, which includes a computer program carried on a computer-readable storage medium, and when the computer program is executed by a computer, the method for scheduling resources as described in the embodiment shown in fig. 2 can be implemented. In such embodiments, the computer program may be downloaded and installed from a network, and/or installed from a removable medium. The computer program, when executed by a processor, performs various functions defined in the system of the present application.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on the understanding, the above technical solutions substantially or otherwise contributing to the prior art may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the various embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (14)
1. A resource scheduling method comprises the following steps:
in response to a resource scheduling request, determining a resource loss model;
calculating loss values of different hosts by using the resource loss model based on the allocable resource vectors of the different hosts; the allocable resource vector is composed of resource components of a plurality of resource dimensions;
determining a candidate host set with a loss value meeting the scheduling requirement;
and determining a target host for allocating resources for the resource scheduling request from the candidate host set.
2. The method of claim 1, wherein the determining a target host from the set of candidate hosts to allocate resources for the resource scheduling request comprises:
determining, from the set of candidate hosts, a target host for which resource components of the plurality of resource dimensions meet a balancing requirement based on a request resource vector of the resource scheduling request; wherein the request resource vector is comprised of resource components of the plurality of resource dimensions.
3. The method of claim 1, wherein the calculating loss values for different hosts using the resource loss model based on allocable resource vectors for different hosts comprises:
and calculating loss values of different hosts by utilizing the resource loss model based on the request resource vector and the allocable resource vectors of different hosts.
4. The method of claim 3, wherein the calculating loss values for different hosts using the resource loss model based on the requested resource vector and allocable resource vectors for different hosts comprises:
determining the residual resource vectors after the allocable resource vectors of different hosts are respectively subtracted from the request resource vector;
and inputting the residual resource vector into the resource loss model, and calculating to obtain loss values of different hosts.
5. The method of claim 1, wherein the determining, from the set of candidate hosts, a target host for which a plurality of resource dimensions meet balancing requirements comprises:
determining an included angle between an allocable resource vector or a residual resource vector of each candidate host in the candidate host set and a main direction vector respectively, and selecting the candidate host with the smallest included angle as a target host;
determining a target resource dimension with the minimum resource residual amount in the residual resource vectors of all candidate hosts in the candidate host set, and selecting the candidate host with the maximum resource residual amount corresponding to the target resource dimension as the target host;
alternatively, the first and second electrodes may be,
and determining the similarity between the allocable resource vector of each candidate host in the candidate host set and the request resource vector, and selecting the candidate host with the maximum similarity as the target host.
6. The method of claim 5, wherein the determining similarity between the allocable resource vector and the request resource vector of each candidate host in the candidate host set, and the selecting the candidate host with the highest similarity as the target host comprises:
calculating the dot product of the allocable resource vector and the request resource vector of each candidate host in the candidate host set, and taking the dot product result as the similarity of the allocable resource vector and the request resource vector of each candidate host;
and selecting the candidate host with the maximum similarity as the target host.
7. The method of claim 1, wherein the resource loss model is constructed as follows:
determining at least one parameter affecting resource scheduling;
and constructing the resource loss model based on at least one parameter influencing resource scheduling and the respectively corresponding weight coefficients.
8. The method of claim 7, wherein the at least one parameter affecting resource scheduling comprises a metric parameter measuring allocable resources, a resource balance, and a resource fragmentation, respectively;
the determining at least one parameter affecting resource scheduling comprises:
determining a metric parameter for measuring the allocable resource based on the modulus value of the resource vector parameter;
determining a measurement parameter for measuring the resource balance degree based on the included angle between the resource vector parameter and the main direction vector;
and determining a measurement parameter for measuring the fragmentation degree of the resource based on resource components of different resource dimensions in the resource vector parameter.
9. The method of claim 8, wherein the determining a metric parameter that measures the allocable resource based on a modulus value of a resource vector parameter comprises:
using the modulus value of the resource vector parameter as a measurement parameter for measuring the allocable resource;
the determining the measurement parameter for measuring the resource balance degree based on the included angle between the resource vector and the main direction vector comprises:
taking the included angle between the resource vector parameter and the main direction vector as a measurement parameter for measuring the resource balance degree;
the determining, based on resource components of different resource dimensions in the resource vector, a metric parameter that measures the fragmentation degree of the resource comprises:
and taking the sum value corresponding to the remainder result of the resource components of different resource dimensions and the fragmentation degree coefficient in the resource vector parameters as a measurement parameter for measuring the fragmentation degree of the resources.
10. The method of claim 7, wherein the at least one parameter affecting resource scheduling comprises a metric parameter measuring allocable resources, a resource balance, and a resource fragmentation, respectively;
the determining at least one parameter affecting resource scheduling comprises:
determining a measurement parameter for measuring the allocable resource based on the modulus value of the resource vector parameter;
determining a measurement parameter for measuring the resource balance degree based on the resource component difference value of any two resource dimensions;
and determining a measurement parameter for measuring the fragmentation degree of the resources based on the resource components of different resource dimensions.
11. The method of claim 1, wherein the determining the set of candidate hosts for which the penalty value meets the scheduling requirement comprises:
determining a set of candidate hosts having a penalty value less than or equal to a first penalty threshold; alternatively, the first and second electrodes may be,
a set of candidate hosts is selected for which the difference between the penalty value and the minimum penalty value is less than or equal to a second penalty threshold.
12. The method of claim 1, further comprising:
and carrying out resource allocation on the resource scheduling request in the target host.
13. A computing device, comprising a processing component and a storage component;
the storage component stores one or more computer instructions; the one or more computer instructions to be invoked for execution by the processing component to implement the resource scheduling method of any one of claims 1 to 12.
14. A computer storage medium in which a computer program is stored which, when executed by a computer, implements the resource scheduling method of any one of claims 1 to 12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310127021.9A CN115981868A (en) | 2023-01-31 | 2023-01-31 | Resource scheduling method and computing equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310127021.9A CN115981868A (en) | 2023-01-31 | 2023-01-31 | Resource scheduling method and computing equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115981868A true CN115981868A (en) | 2023-04-18 |
Family
ID=85970439
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310127021.9A Pending CN115981868A (en) | 2023-01-31 | 2023-01-31 | Resource scheduling method and computing equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115981868A (en) |
-
2023
- 2023-01-31 CN CN202310127021.9A patent/CN115981868A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10339152B2 (en) | Managing software asset environment using cognitive distributed cloud infrastructure | |
| US11263052B2 (en) | Determining optimal compute resources for distributed batch based optimization applications | |
| CN110233873B (en) | Block chain-based virtual certificate value distribution method, contract running device and storage medium | |
| US20120131172A1 (en) | Managing service level agreements using statistical process control in a networked computing environment | |
| US20140289245A1 (en) | Optimizing a clustered virtual computing environment | |
| US9785474B2 (en) | Managing a shared pool of configurable computing resources using a set of scaling factors and a set of workload resource data | |
| CN112640388A (en) | Suspicious activity detection in computer networks | |
| US20130024573A1 (en) | Scalable and efficient management of virtual appliance in a cloud | |
| US8806483B2 (en) | Determining starting values for virtual machine attributes in a networked computing environment | |
| EP2994832A2 (en) | User-influenced placement of virtual machine instances | |
| WO2006112988A2 (en) | Prescriptive architecture recommendations | |
| US20120323821A1 (en) | Methods for billing for data storage in a tiered data storage system | |
| US10956541B2 (en) | Dynamic optimization of software license allocation using machine learning-based user clustering | |
| US20150149608A1 (en) | Information technology resource management | |
| CN112187670B (en) | Networked software shared resource allocation method and device based on group intelligence | |
| US20140244311A1 (en) | Protecting against data loss in a networked computing environment | |
| US20170243268A1 (en) | Method and system for determining an optimized service package model for market participation | |
| CN110689425A (en) | Method and device for pricing quota based on income and electronic equipment | |
| CN108845860B (en) | Method and device for managing quota and electronic equipment | |
| CN111771222A (en) | Method and system for optimizing resource reallocation | |
| CN115981868A (en) | Resource scheduling method and computing equipment | |
| CN111210109A (en) | Method and device for predicting user risk based on associated user and electronic equipment | |
| AU2020418595B2 (en) | Implementing workloads in a multi-cloud environment | |
| US20220083242A1 (en) | Metering computing power in memory subsystems | |
| CN114020469A (en) | Edge node-based multi-task learning method, device, medium and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |