A management method of a computing power cloud architecture combining a container cloud and a virtual machine cloud

Abstract

The application belongs to the field of cloud computing, and is a management method of a computing power cloud architecture combining a container cloud and a virtual machine cloud, a cloud management platform, a host resource management layer and a resource access layer, and comprises the following steps: S1: a host resource management module registers with the cloud management platform; S2: a registration and scheduling module sends a container creation instruction, and a host creates and schedules a container; S3: the host resource management layer reports idle resource information to the cloud management platform, and the cloud management platform triggers the binding and deployment of the container and the host; S4: a virtual machine access module sends a virtual machine creation instruction, and deploys a virtual machine to a specified host; S5: a resource access layer respectively collects and reports the running states of the container and the virtual machine; S6: a container state monitoring module sends a container unbinding and destruction instruction, and a host resource management module executes the operation; S7: a virtual machine host module sends an unbinding instruction, executes virtual machine unbinding, and a resource access layer releases resources.

Description

Technical Field

[0001] This invention belongs to the field of cloud computing, and in particular, it is a management method for a computing power cloud architecture that combines container cloud and virtual machine cloud. Background Technology

[0002] Traditionally, container cloud platforms run in a container runtime environment. In this environment, containers are isolated from the host machine through namespaces provided by the Linux kernel, which limits the performance and security of containers.

[0003] Virtualization technology refers to the abstraction of limited physical resources into virtualized resources. These virtualized resources are isolated from each other and can simulate the actual physical environment, providing virtualized resource usage for upper-layer applications or operating systems. Each virtual machine can have its own private memory, CPU, and disk, and can serve independent client applications. However, virtualization technology has limited resource isolation capabilities; for example, it cannot isolate network ports and device resources. Furthermore, it requires numerous system calls, resulting in performance degradation and overhead during startup and operation.

[0004] Virtual machine technology has advantages such as resource isolation and comprehensive functionality, but container technology has shortcomings in terms of performance and agility.

[0005] Currently, the development of container cloud platform technology in China is in its early stages, with various vendors offering different container cloud platform solutions, while foreign open-source projects lack broad applicability to the Chinese market. Summary of the Invention

[0006] This invention proposes a management method for a computing cloud architecture that combines container cloud and virtual machine cloud.

[0007] A management method for a computing cloud architecture combining container cloud and virtual machine cloud, comprising a cloud management platform, a host resource management layer, and a resource access layer, including the following steps: S1: The host machine resource management module registers with the cloud management platform and sends registration information. The cloud management platform then returns the registration result. S2: The registration and scheduling module in the resource access layer sends a container creation command. The host machine creates and schedules the container based on the reinforcement learning model. At the same time, the container log module in the resource access layer uploads logs to the cloud management platform. S3: The host machine resource management layer reports idle resource information, including memory and CPU, to the cloud management platform. The cloud management platform triggers the binding and deployment of containers to the host machine according to resource scheduling instructions. S4: The virtual machine access module in the resource access layer sends a virtual machine creation command to deploy the virtual machine to the specified host machine; S5: The container status monitoring module and virtual machine host module in the resource access layer collect and report the running status of containers and virtual machines respectively, and the corresponding container log sending module and virtual machine log sending module synchronously upload the running logs of containers and virtual machines; S6: The container status monitoring module sends container unbinding and destruction commands to the cloud management platform, and the host resource management module executes the container unbinding, migration and destruction operations; S7: The virtual machine host module sends an unbinding command to the cloud management platform to unbind the virtual machine, and the container access module and virtual machine access module in the resource access layer release resources.

[0008] Preferably, the host machine creates and schedules containers based on a reinforcement learning model, which includes the following: the reinforcement learning model includes intelligent scheduling of container deployment by defining resource status, scheduling actions and reward mechanisms, and combining Q-Learning algorithm for iterative optimization based on the current resource status of the host machine as input. The creation and scheduling of containers on the host machine specifically includes: the container creation module creating the container, and the container deployment module scheduling the container to be deployed to the specified host machine according to the instructions. The scheduling instructions include the target host machine ID and the resource parameters required by the container.

[0009] Preferably, the host resource management layer reports idle resource information to the cloud management platform. The idle resource information includes the container resources currently available on the host, including the number of idle CPU cores and the number of idle memory. The number of idle CPU cores is calculated by subtracting the number of CPU cores used from the total number of CPU cores on the host, and the number of idle memory is calculated by subtracting the number of memory used from the total memory on the host.

[0010] Preferably, the corresponding container log sending module and virtual machine log sending module synchronously upload the runtime logs of the container and the virtual machine. The container runtime logs include: logs output to the log pool during container operation, including timestamps, process IDs, operation types, and error messages. The operation types include start and stop. The virtual machine runtime logs include logs output to the log pool during virtual machine operating system operation, including system startup time, CPU usage changes, and peak memory usage.

[0011] Preferably, the container status monitoring module sends container unbinding and destruction commands to the cloud management platform, and the host resource management module executes the container unbinding, migration, and destruction operations. The unbinding and migration specifically include the following: The container status monitoring module sends a container unbinding command to the cloud management platform, the cloud management platform sends an unbinding command to the host resource management module, and the host resource management module unbinds the container. The container unbinding module sends an unbinding completion command to the cloud management platform, the cloud management platform sends a container pull command to the container deployment module, the host machine creates and schedules the container, and the container downtime during the migration process is controlled within the threshold time. The container migration strategy is based on the migration benefit formula, which calculates the benefit based on the utilization rate of the original host machine before migration, the utilization rate of the original host machine after migration, and the migration cost, where the migration cost includes network transmission time.

[0012] This invention also discloses a computing cloud architecture combining container cloud and virtual machine cloud, and a management method for the aforementioned computing cloud architecture combining container cloud and virtual machine cloud, including a cloud management platform, a host resource management layer, and a resource access layer, specifically including the following: The cloud management platform is used to schedule the resources of the underlying host machine according to resource scheduling instructions, which are dynamically generated based on the host machine resource utilization rate. The host resource management layer includes several host machines, each of which is equipped with a host resource management module for managing the resources of the host machine. The resource access layer includes a container access module and a virtual machine access module. The container access module includes several container virtual machines, which are used to implement the access and scheduling of several containers. The virtual machine access module includes several virtual machines, which are used to enable the access and scheduling of these virtual machines.

[0013] Preferably, the host resource management module includes a container binding module, a container startup module, a container running module, a container status query module, a container destruction module, and a container unbinding module, specifically including the following: The container binding module is used to bind the created container to the host machine, and the binding strategy refers to the host machine resource utilization formula; The container startup module is used to start the container; The container runtime module is used to enable the container to run on the host machine; The container status query module is used to obtain the container running status; The container destruction module is used to destroy specified containers; The container unbinding module is used to unbind the container from the host machine.

[0014] Preferably, the container access module includes a registration and scheduling module, a container creation module, a container deployment module, a container log acquisition module, a container log sending module, and a container status monitoring module, specifically including the following: The container creation module is used to create containers according to business needs; The container deployment module schedules containers to be deployed to a specified host machine according to instructions. The scheduling logic is based on the host machine resource utilization and the container priority. The container status monitoring module is used to monitor the container status; The container log acquisition module is used to acquire container runtime logs; The container log sending module is used to send container runtime logs; The registration and scheduling module is used to register and bind containers to the host machine.

[0015] Preferably, the virtual machine access module includes a virtual machine host module, a virtual machine creation module, a virtual machine start / stop module, a virtual machine status query module, a virtual machine unbinding module, a virtual machine log acquisition module, and a virtual machine log sending module, specifically including the following: The virtual machine host module is used to implement the registration and binding of the virtual machine and the host machine, and the binding strategy refers to the matching degree formula between the virtual machine and the host machine; The virtual machine creation module is used to create virtual machines; The virtual machine start / stop module is used to start and stop virtual machines; The virtual machine status query module is used to obtain the running status of the virtual machine; The virtual machine unbinding module is used to unbind the relationship between the virtual machine and the host machine; The virtual machine log acquisition module is used to acquire virtual machine runtime logs; The virtual machine log sending module is used to send virtual machine runtime logs.

[0016] Preferably, the virtual machine host module is used to register and bind the virtual machine and the host machine. The binding strategy refers to the matching degree formula of the virtual machine and the host machine. The matching degree formula of the virtual machine and the host machine referred to by the binding strategy includes the following: the matching degree between the target virtual machine and the corresponding virtual machine is calculated based on the resource utilization rate and resource utilization rate weight of the target host machine, the network latency between the target host machine and the corresponding virtual machine and the network latency weight.

[0017] The present invention has the following beneficial effects: This invention integrates two forms of computing resources: containers and virtual machines. It breaks the resource isolation limitations of large-scale cloud computing platforms and new computing power clouds, and achieves coordinated scheduling of the two through unified management. It not only meets the needs of traditional businesses for virtual machines, but also adapts to the dependence of new businesses on containers, greatly improving the resource compatibility and business coverage of the cloud platform. This invention effectively optimizes resource allocation efficiency by leveraging unified scheduling, monitoring, and data management mechanisms. It reduces resource waste through dynamic scheduling and status monitoring, while simplifying management processes, reducing operational complexity, and significantly enhancing the operational reliability and ease of use of the cloud platform. The computing cloud architecture combining container cloud and virtual machine cloud of the present invention combines the advantages of lightweight and efficient containers with the strong isolation of virtual machines, flexibly responding to the resource needs of different business scenarios. It can quickly deploy elastically scalable container applications and stably run virtual machine services with high environmental isolation requirements, thereby improving the comprehensive service capabilities and market adaptability of the cloud platform. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the steps of a management method for a computing cloud architecture combining container cloud and virtual machine cloud according to the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions in the embodiments of this invention will be clearly described below in conjunction with the examples.

[0020] like Figure 1 As shown, this invention proposes a management method for a computing cloud architecture that combines container cloud and virtual machine cloud.

[0021] In one feasible embodiment, a computing cloud architecture combining container cloud and virtual machine cloud is included, comprising a cloud management platform, a host resource management layer, and a resource access layer. The cloud management platform is used to schedule resources of the underlying host machine according to resource scheduling instructions, which are based on a host machine resource utilization formula. Dynamically generated; the host resource management layer has several host machines, each host machine has a host resource management module, which manages the resources of the host machine; the resource access layer includes a container access module and a virtual machine access module, the container access module has several container virtual machines, which are used to bind and schedule several containers with their corresponding host machines; the virtual machine access module has several virtual machines, which are used to bind and schedule several virtual machines with their corresponding host machines.

[0022] The container access module includes a registration and scheduling module, a container creation module, a container deployment module, a container log acquisition module, a container log sending module, and a container status monitoring module. The container creation module creates containers according to business needs; the container deployment module deploys several containers according to instructions, with deployment decisions considering the compatibility between the containers and the host machine; the container status monitoring module monitors container status, using an LSTM neural network model to detect anomalies in time-series data such as container CPU utilization and memory usage; the container log acquisition module acquires container runtime logs; the container log sending module sends container runtime logs; and the registration and scheduling module registers and binds containers to the host machine.

[0023] The host resource management module includes a container binding module, a container startup module, a container running module, a container status query module, a container destruction module, and a container unbinding module. The container binding module is used to bind created containers to the host machine, and the binding logic is based on the host machine's idle resources. and The container startup module is used to start the container; the container running module is used to enable the container to run on the host machine; the container status query module is used to obtain the container running status; the container destruction module is used to destroy the specified container; and the container unbinding module is used to unbind the container from the host machine.

[0024] The virtual machine access module includes a virtual machine host module, a virtual machine creation module, a virtual machine start / stop module, a virtual machine status query module, a virtual machine unbinding module, a virtual machine log acquisition module, and a virtual machine log sending module. The virtual machine host module is used to register and bind virtual machines to the host machine, and the binding strategy is based on a matching degree formula. The virtual machine creation module is used to create virtual machines; the virtual machine start / stop module is used to start / stop virtual machines; the virtual machine status query module is used to obtain the virtual machine running status; the virtual machine unbinding module is used to unbind the virtual machine from the host machine; the virtual machine log acquisition module is used to acquire virtual machine running logs; and the virtual machine log sending module is used to send virtual machine running logs.

[0025] A management method for a computing cloud architecture combining container cloud and virtual machine cloud includes the following steps: Step 1: The host resource management module registers with the cloud management platform. The registration information includes: Host ID; Total number of CPU cores, such as 32 cores; Total memory, such as 128GB; Total storage, such as 1TB; After registration is complete, the cloud management platform sends a registration success notification to the host machine resource management module; Step 2: The registration and scheduling module sends a container creation command to the host resource management module. The command includes the container ID (e.g., Container-001) and the required resources (2 CPU cores, 4GB memory). The host machine then creates and schedules the container, with the scheduling process based on a reinforcement learning model. The cloud management platform sends a command to the container log acquisition module to read container logs. The container log acquisition module reads the container logs on the host machine, such as the startup logs and error logs of Container-001, and sends them to the cloud management platform. The container log sending module sends container logs to the cloud management platform. Step 3: The host machine resource management module sends host machine idle information to the cloud management platform, such as host machine 1: 10 idle CPU cores and 40GB idle memory; the resource scheduling instruction sends binding and deployment instructions to the container binding module and the container deployment module; the container binding module binds the container to the specified host machine, such as host machine 1, and the container deployment module calls the host machine to create and deploy the container; Step 4: The virtual machine access module sends a virtual machine creation command to the cloud management platform. The command includes the virtual machine ID: VM-001, and the required resources: 8-core CPU and 16GB memory. The virtual machine is then created according to the command. The virtual machine is deployed to the specified host machine, such as host machine 2. Step 5: The container status monitoring module actively collects container status, such as Container-001's CPU utilization of 20% and memory usage of 2GB, and reports it to the cloud management platform; the virtual machine host module actively collects virtual machine status, such as VM-001's CPU utilization of 30% and memory usage of 8GB, and reports it to the cloud management platform; the container log acquisition module acquires container runtime logs and sends them to the container log sending module, which then sends the container runtime logs to the cloud management platform; the virtual machine log acquisition module acquires virtual machine runtime logs, and the virtual machine log sending module sends the virtual machine runtime logs to the cloud management platform. Step 6: The container status monitoring module sends an unbinding command to the cloud management platform for Container-001. The host resource management module unbinds the container from host machine 1, and the container unbinding module releases the binding between the specified container and the host machine, allowing the container to migrate to another host machine, such as host machine 3. The migration is based on the migration benefit formula. The container status monitoring module sends a container destruction command to the cloud management platform. For Container-001, the resource scheduling command sends destruction and scheduling commands to the host machine and the container startup scheduling module. The container startup scheduling module schedules the host machine to destroy the container, so that the container status monitoring module can monitor the container destruction. Step 7: The virtual machine host module sends an unbinding command to the cloud management platform. For VM-001, the virtual machine unbinding module unbinds the virtual machine from the host machine 2. The host machine resource management module and the virtual machine access module send resource release commands to the cloud management platform respectively. The resource management module releases the resources, and the host machine 2 releases the 8-core CPU and 16GB memory.

[0026] Specifically, step 2, which involves the host machine creating and scheduling containers, involves the container creation module creating containers and the container deployment module scheduling the containers to be deployed to the designated host machine according to instructions.

[0027] In step 3, the idle information of the host machine refers to the available container resources on the host machine, namely the number of idle CPU cores and the amount of idle memory.

[0028] In step 5, the container logs refer to the logs output to the log pool during container operation; the virtual machine runtime logs refer to the logs output to the log pool during virtual machine operating system operation.

[0029] In step 6, unbinding the container refers to: the container status monitoring module sending an unbinding command to the cloud management platform, the cloud management platform sending an unbinding command to the host resource management module, and the host resource management module unbinding the container. The migration of the host machine mentioned in step 6 refers to: the container unbinding module sending an unbinding completion instruction to the cloud management platform, the cloud management platform sending a container pull instruction to the container deployment module, and the host machine creating and scheduling the container.

[0030] A computing device for implementing the management method of a computing power cloud architecture combining container cloud and virtual machine cloud, comprising: The cloud management platform is used to receive resource registration information and container scheduling instructions from the resource access module of the host resource management layer; the cloud management platform is also used to send scheduling instructions to the host resource management module, container access module, and virtual machine access module; the cloud management platform is also used to receive data from the container log acquisition module, container log sending module, container status monitoring module, host resource management module, virtual machine status query module, virtual machine log acquisition module, and virtual machine log sending module. The host resource management layer comprises several host machines. Each host machine includes a host resource management module and a container access module. The host resource management module receives instructions from the cloud management platform and sends creation instructions to the container creation module, as well as scheduling instructions to the container deployment module and the container status monitoring module. The container creation module creates containers and sends container deployment instructions to the container deployment module, which then deploys several containers according to the instructions. The container status monitoring module monitors container status and uses an LSTM model for anomaly detection. The container log acquisition module acquires container runtime logs, and the container log sending module sends the container runtime logs. The resource access layer includes a container access module and a virtual machine access module. The container access module has several container virtual machines, which are used to create, schedule, and deploy several containers. The virtual machine access module has several virtual machines, which are used to schedule and deploy several virtual machines. Resource scheduling instructions are used to schedule container creation, deployment, destruction, migration, and operation, and are generated based on a reinforcement learning model. The virtual machine access module is used to respond to instructions sent by the cloud management platform and register the virtual machine on the cloud management platform.

[0031] The virtual machine access module is used to receive instructions from the cloud management platform, create virtual machines, and deploy virtual machines to the host machine specified by the cloud management platform; the virtual machine access module is used to respond to instructions from the cloud management platform and unbind virtual machines; the virtual machine status query module is used to actively collect virtual machine status and report it to the cloud management platform; the virtual machine log acquisition module is used to acquire virtual machine running logs; and the virtual machine log sending module is used to send the virtual machine running logs.

[0032] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a management method for a computing power cloud architecture combining container cloud and virtual machine cloud.

[0033] Example 1: Computing cloud architecture combining container cloud and virtual machine cloud A computing cloud architecture combining container cloud and virtual machine cloud includes a cloud management platform, a host resource management layer, and a resource access layer; the cloud management platform includes a container access module and a virtual machine access module; the host resource management layer includes a host resource management module; the resource access layer is connected to the host resource management module, and the container access module is connected to the host resource management module.

[0034] The container access module includes a host resource management module, a container creation module, a container deployment module, a host resource management module, a container log acquisition module, a container log sending module, and a container status monitoring module. The virtual machine access module includes a virtual machine host module, a virtual machine creation module, a virtual machine start / stop module, a virtual machine status query module, a virtual machine unbinding module, a virtual machine log acquisition module, and a virtual machine log sending module. The container creation module is connected to the container deployment module, which is connected to the cloud management platform. The container log acquisition module is connected to the container log sending module, and the container status monitoring module is connected to the container deployment module. The container status monitoring module has a built-in LSTM anomaly detection model, with the input being the container's CPU usage over the past 5 minutes, a sampling interval of 10 seconds, and a total of 30 data points. The virtual machine host module is connected to the virtual machine creation module, and the virtual machine start / stop module is connected to the virtual machine status query module.

[0035] The host resource management module includes a container binding module, a container startup module, a container running module, a container status query module, a container destruction module, and a container unbinding module. The container binding module is connected to the container deployment module, and the binding decision is based on the host resource utilization rate. ,like The container is bound first; the container startup module is connected to the container running module, the container status query module is connected to the container binding module, and the container destruction module is connected to the container unbinding module.

[0036] Example 2: Management method for a computing cloud architecture combining container cloud and virtual machine cloud A management method for a computing cloud architecture combining container cloud and virtual machine cloud includes the following steps: Step 1: The host resource management module registers with the cloud management platform. The registration information includes: Host 1, ID: H1, CPU: 32 cores, Memory: 128GB; Host machine 2, ID: H2, CPU: 64 cores, memory: 256GB; Host machine 3, ID: H3, CPU: 48 cores, memory: 192GB; After registration is complete, the cloud management platform sends a registration success notification to the host machine resource management module; Step 2: The registration and scheduling module sends a container creation command to the host resource management module. Container ID: C1, requirements: 2 CPU cores, 4GB memory; the container creation module creates container C1; the container deployment module uses a reinforcement learning model. Calculate the reward for scheduling to H1 Higher than those scheduled to H2 and H3 Therefore, C1 is scheduled to H1; the cloud management platform sends a command to the container log acquisition module to read the container logs, and the container log acquisition module reads the logs of C1 on H1, such as "2025-07-18 10:00:00 C1 started successfully", and sends it to the cloud management platform; the container log sending module sends the logs of C1 to the cloud management platform. Step 3: The host resource management module sends host idle information to the cloud management platform. H1, idle CPU: 30 cores, memory: 124GB; H2, idle CPU: 60 cores, memory: 240GB; H3, idle CPU: 45 cores, memory: 180GB; Resource scheduling instructions are sent to the container binding module and the container deployment module for binding; Deployment command, bind H1, deploy C1; The container binding module binds C1 to H1, and the container deployment module calls H1 to create and deploy C1; Step 4: The virtual machine access module sends a virtual machine creation command, VM1, to the cloud management platform. Requirements: 8-core CPU, 16GB memory. The cloud management platform calculates the compatibility between VM1 and each host machine. : H1 , H2 , H3 , Since H1 has the highest matching degree, VM1 is deployed to H1; Step 5: The container status monitoring module collects the status of C1: 10:05 CPU utilization 15%, memory usage 3GB, and reports it to the cloud management platform; the virtual machine host module collects the status of VM1: 10:05 CPU utilization 20%, memory usage 10GB, and reports it to the cloud management platform; the container log acquisition module obtains the log of C1, "10:03 C1 completes data synchronization" and sends it to the container log sending module, which then sends the log to the cloud management platform; the virtual machine log acquisition module obtains the log of VM1, "10:04 VM1 network connection is normal", and the virtual machine log sending module sends the log to the cloud management platform. In step 6, at 10:30, the container status monitoring module detected the resource utilization of H1. However, C1 needs to migrate due to network fluctuations; calculate the migration benefits. Before migration, H1 utilization was 31.25%; after migration, H1 utilization was 8 / 32 = 25%. Migration cost... The network transmission takes 2 seconds, therefore The profit is positive; the container status monitoring module sends an unbind command to the cloud management platform, H1 unbinds C1, and C1 migrates to H3; at 11:00, the container status monitoring module sends a destroy command to the cloud management platform, the resource scheduling command sends a destroy command to H3 and the container startup scheduling module, H3 destroys C1, and the container status monitoring module confirms that C1 has been destroyed. Step 7, at 11:30, the virtual machine host module sends an unbinding command to the cloud management platform, and the virtual machine unbinding module unbinds the relationship between VM1 and H1; H1 and the virtual machine access module send a resource release command to the cloud management platform, and H1 releases 8 CPU cores and 16GB memory. The resource status is updated to H1, with 32 idle CPU cores and 128GB memory.

[0037] Furthermore, the creation and scheduling of containers by the host machine in step 2 specifically involves: the container creation module creating the container, and the container deployment module scheduling the container to be deployed to the specified host machine according to the instructions.

[0038] Furthermore, the idle information of the host machine mentioned in step 3 refers to the available container resources on that host machine, i.e. and .

[0039] Furthermore, the container logs mentioned in step 5 refer to the logs output to the log pool during container operation; the virtual machine runtime logs refer to the logs output to the log pool during virtual machine operating system operation.

[0040] Furthermore, the unbinding of the container in step 6 refers to: the container status monitoring module sending an unbinding instruction to the cloud management platform, the cloud management platform sending an unbinding instruction to the host resource management module, and the host resource management module unbinding the container; the migration of the host machine in step 6 refers to: the container unbinding module sending an unbinding completion instruction to the cloud management platform, the cloud management platform sending a pull instruction to the container deployment module, and the host machine creating and scheduling the container.

[0041] Example 3: Computing Device A computing device implementing the management method of a computing cloud architecture combining container cloud and virtual machine cloud according to Embodiment 2 includes: The cloud management platform is used to receive resource registration information from the resource access modules of the host machine's resource management layer, such as the hardware configuration of H1, H2, and H3, as well as container scheduling instructions. The cloud management platform is also used to send scheduling instructions to the host resource management module, container access module, and virtual machine access module, such as deploying C1 to H1; The cloud management platform is also used to receive data from the container log acquisition module, container log sending module, container status monitoring module, host resource management module, virtual machine status query module, virtual machine log acquisition module, and virtual machine log sending module; Among them, the container status monitoring module includes the anomaly detection results of C1, the host resource management module includes the resource utilization of H1, and the virtual machine status query module includes the running status of VM1. The host resource management layer has several host machines, such as H1, H2, and H3. Each host machine has a host resource management module and a container access module. The host resource management module is used to receive instructions sent by the cloud management platform, send creation instructions to the container creation module (e.g., to create C1), and send scheduling instructions to the container deployment module and the container status monitoring module. The container creation module is used to create containers, such as C1, and send container deployment instructions to the container deployment module. The container deployment module deploys several containers according to the instructions. The container status monitoring module is used to monitor the container status. It adopts an LSTM model and outputs anomaly probabilities. For example, the anomaly probability of C1 is 0.05. If it is below the threshold of 0.1, it is normal. The container log acquisition module is used to acquire container runtime logs, and the container log sending module is used to send container runtime logs. The resource access layer includes a container access module and a virtual machine access module. The container access module is equipped with several container virtual machines, such as C1. The container virtual machines are used to create, schedule and deploy several containers. The virtual machine access module is equipped with several virtual machines, such as VM1, which are used to schedule and deploy the virtual machines. Resource scheduling instructions are used to schedule container creation, deployment, destruction, migration, and operation, and are generated based on a reinforcement learning model; The virtual machine access module is used to respond to instructions sent by the cloud management platform and register the virtual machine to the cloud management platform, such as VM1.

[0042] Furthermore, the virtual machine access module is used to receive instructions sent by the cloud management platform, create a virtual machine, such as VM1, and deploy the virtual machine to a host machine specified by the cloud management platform, such as H1; the virtual machine access module is used to respond to instructions sent by the cloud management platform to unbind the virtual machine; the virtual machine status query module is used to actively collect the virtual machine status, such as the CPU utilization of VM1, and report it to the cloud management platform; the virtual machine log acquisition module is used to acquire the virtual machine running log, and the virtual machine log sending module is used to send the virtual machine running log.

[0043] Furthermore, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the management method for a computing power cloud architecture combining container cloud and virtual machine cloud as described in Embodiment 2.

[0044] Example 4: Specific Applications of Computing Power Cloud Architecture An internet company deployed the computing cloud architecture of this invention to support its e-commerce business, including container-based microservice deployment and virtual machine-based traditional ERP system deployment. The specific application process is as follows: The host resource management module registers three host machines with the cloud management platform: H1, 32-core CPU, 128GB RAM; H2, 64-core CPU, 256GB RAM; H3, 48-core CPU, 192GB RAM; after registration, the cloud management platform reported successful registration.

[0045] Step 2: The registration and scheduling module sends a container creation command, C1, an e-commerce payment microservice, requiring 2 CPU cores and 4GB of memory. The container creation module creates C1, and the container deployment module calculates the highest reward for scheduling to H1 based on a reinforcement learning model. Therefore, C1 is deployed to H1; the cloud management platform reads C1's log "payment service started successfully" and stores it.

[0046] Step 3: The host machine resource management module sends idle information: H1, 30 cores, 124GB; H2, 64 cores, 256GB; H3, 48 cores, 192GB; resource scheduling commands are sent to bind and deploy commands, C1 is bound to H1 and deployment is completed.

[0047] Step 4: The virtual machine access module sends a virtual machine creation command, VM1, for the ERP system, requiring an 8-core CPU and 16GB of memory. The cloud management platform calculates the compatibility between VM1 and H1. Since the value is the highest, VM1 is deployed to H1.

[0048] Step 5: The container status monitoring module collects the C1 status every 10 seconds: the CPU utilization from 10:00 to 10:05 is 12%, 15%, 14%, 16%, and 13%, respectively, all of which are normal and reported; the virtual machine host module collects the VM1 status: the CPU utilization during the same period is 22%, 20%, 19%, 21%, and 23%, all of which are normal and reported; the logs of C1 and VM1 are uploaded to the cloud management platform in real time.

[0049] Step 6, 10:30: Due to load balancing requirements of H1, C1 is migrated to H3. Migration benefits... At 11:00, C1 completed its mission and was destroyed.

[0050] Step 7, 11:30: VM1 is finished using. Unbind and release H1's 8-core CPU and 16GB memory. H1 is restored to its initial resource state.

[0051] Furthermore, a computer-readable storage medium is provided, which stores a computer program. When executed by a processor, the computer program implements the steps of the management method for a computing power cloud architecture combining container cloud and virtual machine cloud as described in Embodiment 2. It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A management method for a computing cloud architecture combining container cloud and virtual machine cloud, comprising a cloud management platform, a host resource management layer, and a resource access layer, including the following steps: S1: The host machine resource management module registers with the cloud management platform and sends registration information. The cloud management platform then returns the registration result. S2: The registration and scheduling module in the resource access layer sends a container creation command. The host machine creates and schedules the container based on the reinforcement learning model. At the same time, the container log module in the resource access layer uploads logs to the cloud management platform. S3: The host machine resource management layer reports idle resource information, including memory and CPU, to the cloud management platform. The cloud management platform triggers the binding and deployment of containers to the host machine according to resource scheduling instructions. S4: The virtual machine access module in the resource access layer sends a virtual machine creation command to deploy the virtual machine to the specified host machine; S5: The container status monitoring module and virtual machine host module in the resource access layer collect and report the running status of containers and virtual machines respectively, and the corresponding container log sending module and virtual machine log sending module synchronously upload the running logs of containers and virtual machines; S6: The container status monitoring module sends container unbinding and destruction commands to the cloud management platform, and the host resource management module executes the container unbinding, migration and destruction operations; S7: The virtual machine host module sends an unbinding command to the cloud management platform to unbind the virtual machine, and the container access module and virtual machine access module in the resource access layer release resources.

2. The management method for a computing cloud architecture combining container cloud and virtual machine cloud according to claim 1, characterized in that, The creation and scheduling of containers by the host machine based on the reinforcement learning model includes the following: The reinforcement learning model includes intelligent scheduling of container deployment by taking the current resource status of the host machine as input, defining resource status, scheduling actions and reward mechanisms, and combining Q-Learning algorithm for iterative optimization. The creation and scheduling of containers on the host machine specifically includes: the container creation module creating the container, and the container deployment module scheduling the container to be deployed to the specified host machine according to the instructions. The scheduling instructions include the target host machine ID and the resource parameters required by the container.

3. The management method for a computing cloud architecture combining container cloud and virtual machine cloud according to claim 1, characterized in that, The host resource management layer reports idle resource information to the cloud management platform. The idle resource information includes the container resources currently available on the host, including the number of idle CPU cores and the number of idle memory. The number of idle CPU cores is calculated by subtracting the number of CPU cores used from the total number of CPU cores on the host, and the number of idle memory is calculated by subtracting the number of memory used from the total memory on the host.

4. The management method for a computing power cloud architecture combining container cloud and virtual machine cloud according to claim 1, characterized in that, The corresponding container log sending module and virtual machine log sending module synchronously upload the runtime logs of the container and virtual machine. The container runtime logs include: logs output to the log pool during container operation, including timestamps, process IDs, operation types, and error messages. The operation types include start and stop. The virtual machine runtime logs include logs output to the log pool during virtual machine operating system operation, including system startup time, CPU utilization changes, and peak memory usage.

5. The management method for a computing cloud architecture combining container cloud and virtual machine cloud according to claim 1, characterized in that, The container status monitoring module sends container unbinding and destruction commands to the cloud management platform, and the host resource management module executes the container unbinding, migration, and destruction operations. The unbinding and migration operations specifically include the following: The container status monitoring module sends a container unbinding command to the cloud management platform, the cloud management platform sends an unbinding command to the host resource management module, and the host resource management module unbinds the container. The container unbinding module sends an unbinding completion command to the cloud management platform, the cloud management platform sends a container pull command to the container deployment module, the host machine creates and schedules the container, and the container downtime during the migration process is controlled within the threshold time. The container migration strategy is based on the migration benefit formula, which calculates the benefit based on the utilization rate of the original host machine before migration, the utilization rate of the original host machine after migration, and the migration cost, where the migration cost includes network transmission time.

6. A computing cloud architecture combining container cloud and virtual machine cloud, applied to the management method of the computing cloud architecture combining container cloud and virtual machine cloud as described in claims 1-5, characterized in that, It includes a cloud management platform, a host resource management layer, and a resource access layer, specifically comprising the following components: The cloud management platform is used to schedule the resources of the underlying host machine according to resource scheduling instructions, which are dynamically generated based on the host machine resource utilization rate. The host resource management layer includes several host machines, each of which is equipped with a host resource management module for managing the resources of the host machine. The resource access layer includes a container access module and a virtual machine access module. The container access module includes several container virtual machines, which are used to implement the access and scheduling of several containers. The virtual machine access module includes several virtual machines, which are used to enable the access and scheduling of these virtual machines.

7. The computing cloud architecture combining container cloud and virtual machine cloud according to claim 6, characterized in that, The host machine resource management module includes a container binding module, a container startup module, a container running module, a container status query module, a container destruction module, and a container unbinding module. Specifically, it includes the following: The container binding module is used to bind the created container to the host machine, and the binding strategy refers to the host machine resource utilization formula. The container startup module is used to start the container; The container runtime module is used to enable the container to run on the host machine; The container status query module is used to obtain the container running status; The container destruction module is used to destroy specified containers; The container unbinding module is used to unbind the container from the host machine.

8. The computing cloud architecture combining container cloud and virtual machine cloud according to claim 6, characterized in that, The container access module shown includes a registration and scheduling module, a container creation module, a container deployment module, a container log acquisition module, a container log sending module, and a container status monitoring module. Specifically, it includes the following: The container creation module is used to create containers according to business needs; The container deployment module schedules containers to be deployed to a specified host machine according to instructions. The scheduling logic is based on the host machine resource utilization and the container priority. The container status monitoring module is used to monitor the container status; The container log acquisition module is used to acquire container runtime logs; The container log sending module is used to send container runtime logs; The registration and scheduling module is used to register and bind containers to the host machine.

9. A computing cloud architecture combining container cloud and virtual machine cloud according to claim 6, characterized in that, The virtual machine access module includes a virtual machine host module, a virtual machine creation module, a virtual machine start / stop module, a virtual machine status query module, a virtual machine unbinding module, a virtual machine log acquisition module, and a virtual machine log sending module. Specifically, it includes the following: The virtual machine host module is used to register and bind the virtual machine to the host machine. The binding strategy refers to the matching degree formula between the virtual machine and the host machine. The virtual machine creation module is used to create virtual machines; The virtual machine start / stop module is used to start and stop virtual machines; The virtual machine status query module is used to obtain the running status of the virtual machine; The virtual machine unbinding module is used to unbind the relationship between the virtual machine and the host machine; The virtual machine log acquisition module is used to acquire virtual machine runtime logs; The virtual machine log sending module is used to send virtual machine runtime logs.

10. A computing cloud architecture combining container cloud and virtual machine cloud according to claim 9, characterized in that, The virtual machine host module is used to register and bind virtual machines and host machines. The binding strategy refers to the matching degree formula between virtual machines and host machines. The matching degree formula between virtual machines and host machines referred to by the binding strategy includes the following: the matching degree between the target virtual machine and the corresponding virtual machine is calculated based on the resource utilization rate and resource utilization rate weight of the target host machine, the network latency between the target host machine and the corresponding virtual machine and the network latency weight.

Images