CN117076133B - Cloud game platform heterogeneous resource allocation method, computer device and storage medium - Google Patents
Cloud game platform heterogeneous resource allocation method, computer device and storage medium Download PDFInfo
- Publication number
- CN117076133B CN117076133B CN202311325788.9A CN202311325788A CN117076133B CN 117076133 B CN117076133 B CN 117076133B CN 202311325788 A CN202311325788 A CN 202311325788A CN 117076133 B CN117076133 B CN 117076133B
- Authority
- CN
- China
- Prior art keywords
- resource
- task
- allocation
- resource pool
- request
- 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.)
- Active
Links
- 238000013468 resource allocation Methods 0.000 title claims abstract description 87
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000012544 monitoring process Methods 0.000 claims abstract description 36
- 230000007246 mechanism Effects 0.000 claims abstract description 7
- 230000008447 perception Effects 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 15
- 230000004044 response Effects 0.000 claims description 13
- 230000008901 benefit Effects 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000003062 neural network model Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 239000013598 vector Substances 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 description 5
- 230000010354 integration Effects 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000002955 isolation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
-
- 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
Abstract
The invention discloses a cloud game platform heterogeneous resource allocation method, a computer device and a storage medium, comprising the following steps: continuously acquiring task requests provided by users, classifying the task types by adopting a data perception mechanism, and dynamically configuring a resource scheduling file; the resource scheduling files with different task types are communicated with each other through a resource pool of a cloud platform server, and a monitoring model is arranged on the resource pool to dynamically acquire the load of the resource pool; monitoring the task state and the resource state of the running resource pool in real time through the monitoring model, and sequentially sending the resource load into a resource allocation model through the task waiting sequence; and configuring a plurality of resource pool allocation schemes capable of meeting the resource load by adopting a resource allocation algorithm for the resource allocation model, combining the task request with the resource utility through a resource allocation utility model, and acquiring a resource pool with the highest allocation utility, so that the work load of the resource pool can be accurately predicted, and the resource utilization rate of the cloud game is improved.
Description
Technical Field
The invention relates to the technical field of cloud game resource allocation, in particular to a cloud game platform heterogeneous resource allocation method, a computer device and a storage medium.
Background
Under the heterogeneous cloud computing architecture, due to different resources such as CPU, memory, hard disk and the like among different frameworks, unavoidable performance difference exists when different types of services are processed, and as the cloud games are increasingly exploded, the user quantity and the access request quantity are increased, and the heterogeneous resources are required to be adopted before each framework to complete adjustment in aspects such as resource cluster scale deployment, optimization parameters and the like, so that efficient utilization of resources is maintained.
The existing cloud game resource allocation calculation mostly adopts a virtualization technology to solve the resource allocation problem and virtualizes a plurality of server resources into a common resource pool, but a plurality of server resources are different in type, performance and architecture, and the whole scale of the calculation resources formed by a plurality of servers is large, so the cloud calculation resource allocation problem is a key problem in cloud computing research .
The current heterogeneous resource allocation method for cloud games mainly considers a plurality of requirements from the perspective of resource service providers or users to meet the requirements of the users and improve the service quality or simply consider the overall resource utilization efficiency, the system running stability and the like, but the method also has the following problems:
(1) In the cloud computing heterogeneous network environment, the existing resource allocation method allocates network resources distributed in different structures to user tasks in different network environments, and the problem of uncoordinated matching of data transmission and data protocols in different networks needs to be overcome, so that cloud game resource allocation calculation is complex, and resource utility is low;
(2) In the current cloud computing, along with the dynamic change of game user demands, the operation information of cloud computing nodes is continuously updated, and the resource workload condition cannot be accurately predicted in a dynamic change environment, so that the system throughput is low.
Disclosure of Invention
The invention aims to provide a cloud game platform heterogeneous resource allocation method, a computer device and a storage medium, which are used for solving the technical problems that in the prior art, a cloud game resource allocation calculation model is complex, the resource utility is low, and the resource workload condition cannot be accurately predicted in a dynamically-changing environment, so that the system throughput is low.
In order to solve the technical problems, the invention specifically provides the following technical scheme:
in a first aspect of the present invention, there is provided a cloud game platform heterogeneous resource allocation method, including the steps of:
continuously acquiring task requests provided by users, inquiring the task requests through a task inquiry interface, identifying task types and task resource demand corresponding to the task requests by adopting a data identification table, classifying the task types by adopting a data perception mechanism, and dynamically configuring resource scheduling files for the task types after classification;
the resource scheduling files with different task types are communicated with each other through a resource pool of a cloud platform server, a monitoring model is arranged on the resource pool to dynamically acquire resource pool loads, the resource scheduling files are ordered according to the resource pool loads by taking task request moments as starting points, task waiting sequences are configured in the resource scheduling files, and resource loads corresponding to the task requests in the task waiting sequences are acquired through a resource load model;
the task state and the resource state of the running resource pool are monitored in real time through the monitoring model, the quantity of the heterogeneous resources remained in the current resource pool is obtained, and the resource load is sequentially sent into a resource allocation model through the task waiting sequence;
and configuring a plurality of resource pool allocation schemes capable of meeting the resource load for the resource allocation model by adopting a resource allocation algorithm, combining the task request with the resource utility through a resource allocation utility model, and obtaining the resource pool with the highest allocation utility.
As a preferred scheme of the invention, the data identification table identifies the task request through a symbol priority identifier, acquires corresponding data content characteristics, orderly arranges the data content characteristics according to symbol priority, and establishes a mapping relation matrix between the ordered data content characteristics and resource demand of the corresponding task request one by one in a fixed period, and caches the type and the resource demand of the current task request in a matrix form.
As a preferred scheme of the invention, the task types are classified by adopting the data perception mechanism, and the resource scheduling file is dynamically configured for the classified task types, which comprises the following steps:
matching the corresponding task requests through the mapping relation matrix, updating the mapping relation in the data identification table, and sequencing the task requests through the priority identifiers;
acquiring the ordered data content characteristics of the task requests, and storing the effective data of the data content characteristics into an input cache according to a mapping relation;
and creating an Overlay type security domain for the input cache, wherein the security domain is connected with the task request through a tree service cloud structure, and an independent resource scheduling file is built.
As a preferred scheme of the invention, the monitoring model dynamically acquires the resource pool load by monitoring the resource attribute in the corresponding task request, and the method comprises the following steps:
initializing the task request to obtain the name, type and task request time of a monitoring object, setting a timing task through a Scheduled function, and polling the time and resource idle information of the task request message reaching a cloud server;
when idle resources or resource requests arrive in the cloud server, acquiring content features of the task requests through polling, and triggering event monitoring time;
when any one of the task request or the resource scheduling file gets a response, a corresponding event object is set, the event object is transmitted to an event resource allocation module, the resource allocation processing task request is executed, the monitoring time is terminated after the task request is ended, and the load condition of the resource pool in the current state is recorded.
As a preferred solution of the present invention, according to the load condition of the resource pool in the current state, dynamically configuring a task waiting sequence in the resource scheduling file, including:
defining a plurality of resource pool sets formed by heterogeneous servers in the resource scheduling file, introducing a time variable taking the task request time as a starting point into the resource pool sets, and dividing resource category time nodes by the time variable;
defining resource vectors of corresponding resource servers according to the resource type time nodes, regularizing all the resource vectors in the resource pool set to be 1, and sequentially configuring task waiting sequences in the resource pool set according to the task request time.
As a preferable scheme of the invention, carrying out resource load prediction on the task waiting sequence through a resource load model to obtain a resource load predicted value corresponding to the task request, comprising the following steps:
the resource load data in one period is configured with the corresponding task waiting sequence by taking the time node intervals of different types of resource types as the period in the resource pool set;
and taking the task waiting sequence in each period as input data, inputting the input data into a neural network model for training, carrying out convolution operation on a convolution layer of the neural network model by taking the period offset as a time domain, extracting data characteristics, and obtaining a resource load predicted value of the task waiting sequence in a task request and processing process.
As a preferred embodiment of the present invention, the step of sequentially sending the task waiting sequence to a resource allocation model by using the resource load predicted value includes:
dividing the task requests into a plurality of sets containing dependency relationships according to the resource load predicted values, and dividing the resource responses into a plurality of clusters according to the dependency relationships among the task requests;
comprehensively planning the resource quantity required by the task requests according to the task waiting sequence by a plurality of clusters, predicting the resource quantity required by each task in the resource pool set, comparing the resource quantity with the existing resource quantity in the resource pool set, and judging whether enough resources meet the current task requests;
and if the current task request is met, carrying out resource scheduling, classifying the task request into a plurality of request response clusters according to the dependency relationship among the task requests, and carrying out resource allocation on the response clusters in the resource allocation model.
As a preferred embodiment of the present invention, the resource allocation model configures a plurality of resource pool allocation schemes by using a resource allocation algorithm, including:
acquiring a task state and a resource state in the current task execution through the monitoring model, and acquiring the resource usage condition, the resource residual quantity and the current resource workload condition of the resource pool;
according to the ordering of the task requests in the task waiting sequence, sequentially responding to the task resource requests, obtaining a plurality of resource pool allocation schemes meeting the resource requests, and predicting the resource load of the resource pool meeting the resource request conditions by reading the resource workload of executing the current task requests;
the method for configuring the corresponding resource pool by adopting the resource allocation utility model comprises the following steps:
calculating resource allocation utility for a plurality of resource pool allocation schemes meeting the resource request through a resource allocation utility function, wherein the resource allocation utility function calculates resource pool normalization benefit by adopting a unified utility function framework;
and selecting a resource pool with high benefit value to execute the resource request by comparing the benefit values of different resource pool allocation schemes.
In a second aspect of the invention, a computer apparatus is provided,
comprising the following steps: at least one processor; and a memory communicatively coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor, the instructions being executable by the at least one processor.
In a third aspect of the present invention, a computer-readable storage medium is provided,
the computer-readable storage medium has stored therein computer-executable instructions that are executed by a processor.
Compared with the prior art, the invention has the following beneficial effects:
according to the invention, a resource load model is added into a utility function resource allocation algorithm to establish cloud game resource allocation, a task request deployment cluster with a dependency relationship is established for the resource allocation model, so that efficient internal service integration is realized, a resource pool allocation scheme corresponding to resource request arrangement is realized by the deployment cluster, a single task request is converted into an integral cluster to carry out resource overall planning and resource scheduling, the calculated amount in the resource scheduling process is simplified, the workload of a resource pool can be accurately predicted, the resource allocation is reasonably calculated for game application requirements under different environments, the resource utilization rate of a cloud game is improved, and the throughput of a system is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It will be apparent to those of ordinary skill in the art that the drawings in the following description are exemplary only and that other implementations can be obtained from the extensions of the drawings provided without inventive effort.
Fig. 1 is a flowchart of a heterogeneous resource allocation method of a cloud game platform according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the invention provides a cloud game platform heterogeneous resource allocation method, which comprises the following steps:
continuously acquiring task requests provided by users, inquiring the task requests through a task inquiry interface, identifying task types and task resource demand corresponding to the task requests by adopting a data identification table, classifying the task types by adopting a data perception mechanism, and dynamically configuring resource scheduling files for the task types after classification;
in this embodiment, when the resource scheduling file is configured, the getStatus method of the Task class is used to obtain the resource usage condition and the resource pool load condition of each time node in the Task type, a set List of resource pool usage conditions is constructed and maintained, the information stored in the class is updated each time before the Task is to be allocated to the corresponding resource node, and when the resource scheduling file is configured, the Task resource is reasonably scheduled to the corresponding time node for execution according to the resource usage information of each time node stored in the Task type.
The resource scheduling files with different task types are communicated with each other through a resource pool of a cloud platform server, a monitoring model is arranged on the resource pool to dynamically acquire resource pool loads, the resource scheduling files are ordered according to the resource pool loads by taking task request moments as starting points, task waiting sequences are configured in the resource scheduling files, and resource loads corresponding to the task requests in the task waiting sequences are acquired through a resource load model;
the task state and the resource state of the running resource pool are monitored in real time through the monitoring model, the quantity of the heterogeneous resources remained in the current resource pool is obtained, and the resource load is sequentially sent into a resource allocation model through the task waiting sequence;
in this embodiment, the monitoring model performs isolation monitoring on the used resources such as CPU, kernel and hard disk based on the Docker container, so that more virtual resource containers can be deployed and monitored at the same time, and the system resource monitoring efficiency and resource utilization rate are greatly improved.
And configuring a plurality of resource pool allocation schemes capable of meeting the resource load for the resource allocation model by adopting a resource allocation algorithm, combining the task request with the resource utility through a resource allocation utility model, and obtaining the resource pool with the highest allocation utility.
In this embodiment, in the resource allocation model, the task request is divided into a plurality of sets including dependency relationships by the resource load, a deployment cluster with the dependency relationships is established, so that efficient integration of internal services is realized, a resource pool allocation scheme corresponding to resource request arrangement is performed by the deployment cluster, a single task request is converted into an integral cluster to perform resource overall planning and resource scheduling, and resource allocation is reasonably performed on game application requirements under different environments, thereby improving the resource utilization rate of the cloud game.
The data identification table identifies the task requests through symbol priority identifiers, obtains corresponding data content features, orderly arranges the data content features according to symbol priorities, corresponds the ordered data content features to resource demand quantities of the corresponding task requests one by one, establishes a mapping relation matrix between the task requests and the resource demand quantities according to the priorities in a fixed period, and caches the types and the resource demand quantities of the current task requests in a matrix form.
Classifying the task types by adopting the data perception mechanism, and dynamically configuring a resource scheduling file for the classified task types, wherein the method comprises the following steps:
matching the corresponding task requests through the mapping relation matrix, updating the mapping relation in the data identification table, and sequencing the task requests through the priority identifiers;
acquiring the ordered data content characteristics of the task requests, and storing the effective data of the data content characteristics into an input cache according to a mapping relation;
and creating an Overlay type security domain for the input cache, wherein the security domain is connected with the task request through a tree service cloud structure, and an independent resource scheduling file is built.
In this embodiment, by creating an Overlay type security domain to implement an Overlay virtualization network on an existing security domain network, and adopting an internal IP manner, for a service container of the same service system, we access the service container to the same virtual security domain network, so as to ensure secure access between virtualization containers on different hosts in the security domain; all service containers of different service systems are accessed to the special security domain network of the service system, so that the application containers can form a special security domain subnet, and isolation protection across security domains is achieved.
In this embodiment, an independent resource scheduling file is established in the security domain subnet, and the effective data of the task requests are continuously cached through the tree service cloud structure, so that the independence of different task requests of the cloud game application is ensured.
The monitoring model dynamically acquires the load of the resource pool by monitoring the resource attribute in the corresponding task request, and the monitoring model comprises the following steps:
initializing the task request to obtain the name, type and task request time of a monitoring object, setting a timing task through a Scheduled function, and polling the time and resource idle information of the task request message reaching a cloud server;
when idle resources or resource requests arrive in the cloud server, acquiring content features of the task requests through polling, and triggering event monitoring time;
when any one of the task request or the resource scheduling file gets a response, a corresponding event object is set, the event object is transmitted to an event resource allocation module, the resource allocation processing task request is executed, the monitoring time is terminated after the task request is ended, and the load condition of the resource pool in the current state is recorded.
In this embodiment, a monitoring model is used to monitor resource utilization conditions in the task execution process in real time, valid data of a task request is used as basic data to be sent to the monitoring model, an event object of the task request is generated in the monitoring model, an event attribute of the task request is obtained through the event object, and operations in the monitoring model are executed.
According to the load condition of the resource pool in the current state, dynamically configuring a task waiting sequence in the resource scheduling file, wherein the task waiting sequence comprises the following steps:
defining a plurality of resource pool sets formed by heterogeneous servers in the resource scheduling file, introducing a time variable taking the task request time as a starting point into the resource pool sets, and dividing resource category time nodes by the time variable;
defining resource vectors of corresponding resource servers according to the resource type time nodes, regularizing all the resource vectors in the resource pool set to be 1, and sequentially configuring task waiting sequences in the resource pool set according to the task request time.
In this embodiment, the time variable is divided into an estimated completion time of the task request and an average completion time of the task request, and if the estimated completion time is greater than the average completion time, the task request is indicated to be a long task class, otherwise, the task request type is marked as a short task class;
in this embodiment, based on the long task type and the short task type, the tasks are divided into different subclasses according to the starting time of the task requests, the task requests with the same starting time are divided into one class, the same subclass job is allocated to the corresponding queue, and each queue supports FIFO and priority scheduling modes, thereby dividing the corresponding resource types.
Predicting the resource load of the task waiting sequence through a resource load model to obtain a resource load predicted value corresponding to the task request, wherein the resource load predicted value comprises the following steps:
the resource load data in one period is configured with the corresponding task waiting sequence by taking the time node intervals of different types of resource types as the period in the resource pool set;
and taking the task waiting sequence in each period as input data, inputting the input data into a neural network model for training, carrying out convolution operation on a convolution layer of the neural network model by taking the period offset as a time domain, extracting data characteristics, and obtaining a resource load predicted value of the task waiting sequence in a task request and processing process.
In this embodiment, the resource load data is used as the workload of the application server, the workload data of the cloud game platform in the period from eight hours a day to six hours a day later is intercepted and processed as a data set, invalid data in the intercepted data set, which is caused by the reasons of downtime or abnormal fluctuation of the server, is discarded, a training data set and a testing data set are finally obtained, and a long-period memory neural network is used for training the data set;
in this embodiment, the number of neural modules of the long-short-term memory neural network is 80, the number of convolution layers is 10, the size of each convolution kernel is 2*5, the data processed by convolution is sent to the pooling layer, the kernel size in the pooling layer is 1*2, the data features extracted from the data processed by the convolution layer and the pooling layer are sent to the long-short-term memory layer, and finally the load predicted value is output.
In the embodiment, 80 neural modules are adopted to circularly process data, so that the training speed can be effectively improved and the loss can be reduced in training, and the Dropout is adopted to sparsify the neural network model, so that the model effect is optimized.
And sending the task waiting sequence into a resource allocation model sequentially through the resource load predicted value, wherein the task waiting sequence comprises the following steps:
dividing the task requests into a plurality of sets containing dependency relationships according to the resource load predicted values, and dividing the resource responses into a plurality of clusters according to the dependency relationships among the task requests;
comprehensively planning the resource quantity required by the task requests according to the task waiting sequence by a plurality of clusters, predicting the resource quantity required by each task in the resource pool set, comparing the resource quantity with the existing resource quantity in the resource pool set, and judging whether enough resources meet the current task requests;
and if the current task request is met, carrying out resource scheduling, classifying the task request into a plurality of request response clusters according to the dependency relationship among the task requests, and carrying out resource allocation on the response clusters in the resource allocation model.
In this embodiment, the task scheduling step of the resource allocation model is as follows:
traversing each task request in the deployment cluster, checking the running state of the task through a getStatus method, and if the task state is running, searching the next task; if the task is a local task of the current resource pool, marking the current position, and adding the current job to an assignedtask schedule; if the task is a non-local task, saving the task information, and continuing to search for the next task;
when the residual resource pool is larger than the resource pool required by the current task, the task can be distributed to the node, and when the residual resource pool exists in the task tracker, the task is distributed to the node for execution according to the priority of the task in the queue; and if the node load rates in the clusters are all close to full load at the moment, waiting until the resources are enough for task execution.
In this embodiment, after a task request of a cloud game is divided into a plurality of sets including dependency relationships, a resource response is divided into a plurality of deployment clusters, a resource allocation scheme matrix is constructed for each single task in a cluster form, when the resource allocation scheme matrix is given, the same task cluster considers the utility value of the scheme for the whole under the condition that a plurality of allocation schemes exist, and therefore an allocation scheme with the highest whole utility value is selected to allocate resources to the whole task group.
The resource allocation model adopts a resource allocation algorithm to configure a plurality of resource pool allocation schemes, and the method comprises the following steps:
acquiring a task state and a resource state in the current task execution through the monitoring model, and acquiring the resource usage condition, the resource residual quantity and the current resource workload condition of the resource pool;
according to the ordering of the task requests in the task waiting sequence, sequentially responding to the task resource requests, obtaining a plurality of resource pool allocation schemes meeting the resource requests, and predicting the resource load of the resource pool meeting the resource request conditions by reading the resource workload of executing the current task requests;
the method for configuring the corresponding resource pool by adopting the resource allocation utility model comprises the following steps:
calculating resource allocation utility for a plurality of resource pool allocation schemes meeting the resource request through a resource allocation utility function, wherein the resource allocation utility function calculates resource pool normalization benefit by adopting a unified utility function framework;
and selecting a resource pool with high benefit value to execute the resource request by comparing the benefit values of different resource pool allocation schemes.
And selecting a resource pool with high benefit value to execute the resource request by comparing the benefit values of different resource pool allocation schemes.
In this embodiment, the load prediction value is converted into a cost vector in a load balancing manner, and a resource allocation scheme with the highest benefit value is selected in combination with a resource allocation utility function.
Second embodiment: a computer device for the computer of a computer system,
comprising the following steps: at least one processor; and a memory communicatively coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor, the instructions being executable by the at least one processor.
Third embodiment: a computer-readable storage medium comprising a memory, a storage medium, and a memory,
the computer-readable storage medium has stored therein computer-executable instructions that are executed by a processor.
According to the invention, a resource load model is added into a utility function resource allocation algorithm to establish cloud game resource allocation, a task request deployment cluster with a dependency relationship is established for the resource allocation model, so that efficient internal service integration is realized, a resource pool allocation scheme corresponding to resource request arrangement is realized by the deployment cluster, a single task request is converted into an integral cluster to carry out resource overall planning and resource scheduling, the calculated amount in the resource scheduling process is simplified, the workload of a resource pool can be accurately predicted, the resource allocation is reasonably calculated for game application requirements under different environments, the resource utilization rate of a cloud game is improved, and the throughput of a system is improved.
The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application, the scope of which is defined by the claims. Various modifications and equivalent arrangements may be made to the present application by those skilled in the art, which modifications and equivalents are also considered to be within the scope of the present application.
Claims (9)
1. The heterogeneous resource allocation method for the cloud game platform is characterized by comprising the following steps of:
continuously acquiring task requests provided by users, inquiring the task requests through a task inquiry interface, identifying task types and task resource demand corresponding to the task requests by adopting a data identification table, classifying the task types by adopting a data perception mechanism, and dynamically configuring resource scheduling files for the task types after classification;
the resource scheduling files with different task types are communicated with each other through a resource pool of a cloud platform server, a monitoring model is arranged on the resource pool to dynamically acquire resource pool loads, the resource scheduling files are ordered according to the resource pool loads by taking task request moments as starting points, task waiting sequences are configured in the resource scheduling files, and resource loads corresponding to the task requests in the task waiting sequences are acquired through a resource load model;
the task state and the resource state of the running resource pool are monitored in real time through the monitoring model, the quantity of the heterogeneous resources remained in the current resource pool is obtained, and the resource load is sequentially sent into a resource allocation model through the task waiting sequence;
a resource allocation algorithm is adopted for the resource allocation model to configure a plurality of resource pool allocation schemes capable of meeting the resource load, and the task request is combined with the resource utility through a resource allocation utility model to obtain a resource pool with highest allocation utility;
the monitoring model dynamically acquires the load of the resource pool by monitoring the resource attribute in the corresponding task request, and the monitoring model comprises the following steps:
initializing the task request to obtain the name, type and task request time of a monitoring object, setting a timing task through a Scheduled function, and polling the time and resource idle information of the task request message reaching a cloud server;
when idle resources or resource requests arrive in the cloud server, acquiring content features of the task requests through polling, and triggering event monitoring time;
when any one of the task request or the resource scheduling file gets a response, a corresponding event object is set, the event object is transmitted to an event resource allocation module, the resource allocation processing task request is executed, the monitoring time is terminated after the task request is ended, and the load condition of the resource pool in the current state is recorded.
2. The cloud gaming platform heterogeneous resource allocation method of claim 1, wherein,
the data identification table identifies the task requests through symbol priority identifiers, obtains corresponding data content features, orderly arranges the data content features according to symbol priorities, corresponds the ordered data content features to resource demand quantities of the corresponding task requests one by one, establishes a mapping relation matrix between the task requests and the resource demand quantities according to the priorities in a fixed period, and caches the types and the resource demand quantities of the current task requests in a matrix form.
3. The cloud gaming platform heterogeneous resource allocation method of claim 2, wherein,
classifying the task types by adopting the data perception mechanism, and dynamically configuring a resource scheduling file for the classified task types, wherein the method comprises the following steps:
matching the corresponding task requests through the mapping relation matrix, updating the mapping relation in the data identification table, and sequencing the task requests through the priority identifiers;
acquiring the ordered data content characteristics of the task requests, and storing the effective data of the data content characteristics into an input cache according to a mapping relation;
and creating an Overlay type security domain for the input cache, wherein the security domain is connected with the task request through a tree service cloud structure, and an independent resource scheduling file is built.
4. The method for heterogeneous resource allocation of a cloud gaming platform of claim 3,
according to the load condition of the resource pool in the current state, dynamically configuring a task waiting sequence in the resource scheduling file, wherein the task waiting sequence comprises the following steps:
defining a plurality of resource pool sets formed by heterogeneous servers in the resource scheduling file, introducing a time variable taking the task request time as a starting point into the resource pool sets, and dividing resource category time nodes by the time variable;
defining resource vectors of corresponding resource servers according to the resource type time nodes, regularizing all the resource vectors in the resource pool set to be 1, and sequentially configuring task waiting sequences in the resource pool set according to the task request time.
5. The cloud gaming platform heterogeneous resource allocation method of claim 4, wherein,
predicting the resource load of the task waiting sequence through a resource load model to obtain a resource load predicted value corresponding to the task request, wherein the resource load predicted value comprises the following steps:
the resource load data in one period is configured with the corresponding task waiting sequence by taking the time node intervals of different types of resource types as the period in the resource pool set;
and taking the task waiting sequence in each period as input data, inputting the input data into a neural network model for training, carrying out convolution operation on a convolution layer of the neural network model by taking the period offset as a time domain, extracting data characteristics, and obtaining a resource load predicted value of the task waiting sequence in a task request and processing process.
6. The cloud gaming platform heterogeneous resource allocation method of claim 5, wherein,
and sending the task waiting sequence into a resource allocation model sequentially through the resource load predicted value, wherein the task waiting sequence comprises the following steps:
dividing the task requests into a plurality of sets containing dependency relationships according to the resource load predicted values, and dividing the resource responses into a plurality of clusters according to the dependency relationships among the task requests;
comprehensively planning the resource quantity required by the task requests according to the task waiting sequence by a plurality of clusters, predicting the resource quantity required by each task in the resource pool set, comparing the resource quantity with the existing resource quantity in the resource pool set, and judging whether enough resources meet the current task requests;
and if the current task request is met, carrying out resource scheduling, classifying the task request into a plurality of request response clusters according to the dependency relationship among the task requests, and carrying out resource allocation on the response clusters in the resource allocation model.
7. The cloud gaming platform heterogeneous resource allocation method of claim 6, wherein,
the resource allocation model adopts a resource allocation algorithm to configure a plurality of resource pool allocation schemes, and the method comprises the following steps:
acquiring a task state and a resource state in the current task execution through the monitoring model, and acquiring the resource usage condition, the resource residual quantity and the current resource workload condition of the resource pool;
according to the ordering of the task requests in the task waiting sequence, sequentially responding to the task resource requests, obtaining a plurality of resource pool allocation schemes meeting the resource requests, and predicting the resource load of the resource pool meeting the resource request conditions by reading the resource workload of executing the current task requests;
the method for configuring the corresponding resource pool by adopting the resource allocation utility model comprises the following steps:
calculating resource allocation utility for a plurality of resource pool allocation schemes meeting the resource request through a resource allocation utility function, wherein the resource allocation utility function calculates resource pool normalization benefit by adopting a unified utility function framework;
and selecting a resource pool with high benefit value to execute the resource request by comparing the benefit values of different resource pool allocation schemes.
8. A computer device, characterized in that,
comprising the following steps: at least one processor; and a memory communicatively coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor to cause the processor to perform the method of any of claims 1-7.
9. A computer-readable storage medium comprising,
the computer readable storage medium having stored therein computer executable instructions which, when executed by a processor, implement the method of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311325788.9A CN117076133B (en) | 2023-10-13 | 2023-10-13 | Cloud game platform heterogeneous resource allocation method, computer device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311325788.9A CN117076133B (en) | 2023-10-13 | 2023-10-13 | Cloud game platform heterogeneous resource allocation method, computer device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117076133A CN117076133A (en) | 2023-11-17 |
| CN117076133B true CN117076133B (en) | 2024-01-26 |
Family
ID=88702832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311325788.9A Active CN117076133B (en) | 2023-10-13 | 2023-10-13 | Cloud game platform heterogeneous resource allocation method, computer device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117076133B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7249179B1 (en) * | 2000-11-09 | 2007-07-24 | Hewlett-Packard Development Company, L.P. | System for automatically activating reserve hardware component based on hierarchical resource deployment scheme or rate of resource consumption |
| CN114741200A (en) * | 2022-04-27 | 2022-07-12 | 国网智能电网研究院有限公司 | Data center station-oriented computing resource allocation method and device and electronic equipment |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5939740B2 (en) * | 2011-04-11 | 2016-06-22 | インターナショナル・ビジネス・マシーンズ・コーポレーションＩｎｔｅｒｎａｔｉｏｎａｌ Ｂｕｓｉｎｅｓｓ Ｍａｃｈｉｎｅｓ Ｃｏｒｐｏｒａｔｉｏｎ | Method, system and program for dynamically allocating resources |
| US20140337435A1 (en) * | 2011-12-13 | 2014-11-13 | Gerald Kaefer | Device and Method for the Dynamic Load Management of Cloud Services |
| US11194572B2 (en) * | 2016-02-22 | 2021-12-07 | International Business Machines Corporation | Managing external feeds in an event-based computing system |
-
2023
- 2023-10-13 CN CN202311325788.9A patent/CN117076133B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7249179B1 (en) * | 2000-11-09 | 2007-07-24 | Hewlett-Packard Development Company, L.P. | System for automatically activating reserve hardware component based on hierarchical resource deployment scheme or rate of resource consumption |
| CN114741200A (en) * | 2022-04-27 | 2022-07-12 | 国网智能电网研究院有限公司 | Data center station-oriented computing resource allocation method and device and electronic equipment |
Non-Patent Citations (1)
| Title |
|---|
| 异构自主平台上云资源分配方法研究与实现;冯硕;中国优秀硕士学位论文全文数据库信息科技辑（月刊）(第03期);I139-129 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117076133A (en) | 2023-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110764912B (en) | Self-adaptive task scheduler and method | |
| Zhang et al. | Online adaptive interference-aware VNF deployment and migration for 5G network slice | |
| US8843929B1 (en) | Scheduling in computer clusters | |
| CN109564528B (en) | System and method for computing resource allocation in distributed computing | |
| CN110221920B (en) | Deployment method, device, storage medium and system | |
| CN110308984B (en) | Cross-cluster computing system for processing geographically distributed data | |
| CN107977271B (en) | Load balancing method for data center integrated management system | |
| CN107070709B (en) | NFV (network function virtualization) implementation method based on bottom NUMA (non uniform memory Access) perception | |
| Khalifa¹ et al. | Collaborative autonomic resource management system for mobile cloud computing | |
| Seth et al. | Dynamic heterogeneous shortest job first (DHSJF): a task scheduling approach for heterogeneous cloud computing systems | |
| El Khoury et al. | Energy-aware placement and scheduling of network traffic flows with deadlines on virtual network functions | |
| Al-Sinayyid et al. | Job scheduler for streaming applications in heterogeneous distributed processing systems | |
| CN110601994A (en) | Load balancing method for micro-service chain perception in cloud environment | |
| CN114679451B (en) | Service dispatching system and dispatching method for edge computing | |
| CN114666335A (en) | DDS-based distributed system load balancing device | |
| CN113014611A (en) | Load balancing method and related equipment | |
| Liu et al. | KubFBS: A fine‐grained and balance‐aware scheduling system for deep learning tasks based on kubernetes | |
| CN117076133B (en) | Cloud game platform heterogeneous resource allocation method, computer device and storage medium | |
| Chunlin et al. | Elastic resource provisioning in hybrid mobile cloud for computationally intensive mobile applications | |
| US11182189B2 (en) | Resource optimization for virtualization environments | |
| Gao et al. | JCSP: Joint caching and service placement for edge computing systems | |
| CN115499432A (en) | Family terminal computing resource management system and computing resource scheduling method | |
| Wang et al. | Model-based scheduling for stream processing systems | |
| Soryani et al. | Improving inter-node communications in multi-core clusters using a contention-free process mapping algorithm | |
| Uchechukwu et al. | Scalable analytic models for performance efficiency in the cloud |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |