Micro-service simulation model deployment system and method based on K8s and Docker technologies
By adopting an automated deployment system based on K8s and Docker in the service simulation system, the time-consuming and labor-intensive deployment of microservice simulation models is solved, and rapid and automated deployment and recycling are achieved, reducing costs and improving simulation efficiency.
Patent Information
- Application Number
- CN202411898376.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-23
- Publication Date
- 2025-05-02
AI Technical Summary
In service-based simulation systems, the deployment of microservice simulation models requires a lot of manpower and time, which leads to the drag of simulation personnel and cannot focus on simulation business.
The microservice simulation model deployment system is adopted based on K8s and Docker technology. The system includes a model image generation module, a model management module, a deployment design module, a deployment execution module and a deployment undo module. By automatically generating K8s scripts and Docker images, it realizes the automatic deployment and recycling of large-scale microservice simulation models.
It greatly reduces the deployment and operation and maintenance costs of service-based simulation systems, reduces the deployment time of simulation experiments, frees up the energy of simulation personnel, and allows them to focus on simulation business.
Smart Images
Figure CN119917118A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer simulation technology and virtualization technology, and in particular to a microservice simulation model deployment system and method based on K8s and Docker technology. Background Art
[0002] Docker is an innovation in cloud service technology. It is implemented in Go language and is a virtualization technology based on the Linux kernel's cgroup, namespace and other technologies. Docker technology makes application deployment and operation and maintenance simpler and more efficient, greatly reducing the deployment and operation and maintenance costs of applications. The full name of K8s is Kubernetes, which is an open source tool for automatically deploying, expanding and managing large-scale containerized applications. K8s technology is mainly used to solve problems such as automatic deployment and automatic expansion of containers in large-scale containerized application scenarios.
[0003] With the rapid development of software technology, simulation software in the field of computer simulation is gradually evolving from a monolithic, component-based architecture to a distributed, service-based architecture. In a service-based simulation architecture, all simulation models are encapsulated as microservices and called by the simulation engine. However, due to the large number of simulation models, hundreds or thousands of model services may be required to simulate just one business direction, and each simulation experiment requires a lot of manpower and time to deploy these services. Summary of the invention
[0004] In order to solve the technical problem of deploying microservice simulation models in service-oriented simulation systems, the present invention proposes a microservice simulation model deployment system and method based on K8S and Docker technology. The present invention can generate K8s scripts when designing simulation scenarios, and complete the deployment of large-scale microservice simulation models based on K8s and Docker technology, which greatly reduces the deployment and operation and maintenance costs of service-oriented simulation systems, frees simulation personnel from complex and tedious deployment work, and can focus on simulation business.
[0005] The technical solution adopted by the present invention is:
[0006] A microservice simulation model deployment system based on K8s and Docker technology includes a model image generation module, a model management module, a deployment design module, a deployment execution module, and a deployment cancellation module; wherein:
[0007] Model image generation module is used to make model images. After the development of the simulation model is completed, the image name, version number, model package location, image storage path, external port number, and dependent package information are input through this module. According to this information, the Dockerfile file of the model image is automatically generated, and the docker build command is automatically executed to generate the corresponding model image package;
[0008] The model management module is responsible for maintaining a MySQL database and a Docker-Registry private warehouse, managing models and model images, and providing operations for creating, deleting, and querying models;
[0009] When creating a model, first input the model information, including name, basic properties, and version number, import the model image file, store the input model information in the Mysql database, and map the corresponding model image with the model information. At the same time, call the docker upload command in the background service to upload the imported model images to the Docker-Registry private warehouse.
[0010] The query operation on the model is used to query all model information and model images in the current model management library;
[0011] The delete operation on the model is used to delete the selected model from the MySQL database and delete the corresponding model image in the Docker-Registry private warehouse;
[0012] The deployment design module is used to design the model image deployment location, including the deployed host, port number, name, and the relationship between Pod and Deployment. The module first reads the simulation scenario file, parses all model services used in the simulation scenario, and then provides a human-computer interaction interface. Through the human-computer interaction interface, the model deployment host, IP address, port number, container, and Deployment information are planned and designed, and a K8s script file is formed, as well as a yaml format file required to create a model container Pod and a yaml format file required to create a Deployment object. The names of the K8s script file and the yaml format file both contain information about the simulation scenario.
[0013] The deployment execution module is responsible for executing the K8s script file and deploying the model image to the corresponding container. Before the simulation scenario starts running, the deployment execution module is first run. This module searches for the corresponding K8s script file through the simulation scenario file, and then automatically executes these K8s deployment scripts to deploy the model image required for the simulation scenario to the planned host and container, and start these models. After the deployment execution module monitors that all the models used have been started normally, it notifies the simulation scenario to start running.
[0014] The deployment cancellation module is used to recycle the model container used in the simulation scenario after the simulation scenario ends. The deployment cancellation module monitors the simulation end event in real time. When the simulation end event is received, the kubectl delete command is called to delete the model container used in this simulation scenario, and the status of the model container is monitored. When all model containers have been deleted, the kubectl delete command is called to delete the Deployment object used in this simulation task.
[0015] A microservice simulation model deployment method based on K8s and Docker technology includes the following steps:
[0016] S1: Generate the simulation model as a model mirror; the specific method is:
[0017] According to the requirements of the simulation task, develop the required model service, then open the human-computer interface of the model image generation module, create a model image, enter the model name, version number, model package path, image storage path, external port number, and dependent packages, and replace the corresponding information in the template based on the built-in Dockerfile template in the image generation module. Use the FROM field to specify the built-in Linux image in the model image generation module as the base image, and generate the Dockerfile file of the model image;
[0018] After generating the Dockerfile file of the model, the model image generation module automatically calls the dockerbuild command to make the model service package into a model image, and then automatically calls the docker save command to save the model image as a tar file and store it in the image storage path;
[0019] S2: Store the simulation model in the model database and upload the corresponding model image to the Docker-Registry; the specific method is:
[0020] Open the interface of the model management module, select Import Model, enter the model name, basic properties, model image tag, version number, select the tar file of the model image, the model management module automatically uses the model image name, calls the docker images command to match and query the model image ID, and then stores the model name, basic properties, model image tag, version number, and model image ID into the model database MySql, and calls the dockerpush command to push the model image tar file to the Docker-Registry private warehouse;
[0021] S3: Plan the deployment location of the model in the simulation model deployment design module, including the host, port number, name, and the relationship between Pod and Deployment, and automatically generate the corresponding K8s script file; the specific method is:
[0022] Open the model deployment design module, first read the simulation scenario file generated by the simulation task, obtain the identification number of the simulation model used in the simulation scenario, query the model image ID, version number and model image tag corresponding to each model in the model database based on these identifications, and present the model name in the human-computer interface in the form of a list;
[0023] According to the host resource situation, configure the host, port number, and container name information of each model deployment in the human-computer interface. After completing the deployment design, the model deployment design module uses the built-in yaml format K8s configuration file template to automatically generate a yaml format file for creating a model container Pod and a yaml format file for the K8s deployment management object Deployment;
[0024] S4: Execute the K8s script file in the model deployment execution module, deploy the model image to the planned host and container, and automatically start the model to start model simulation; the specific method is:
[0025] After receiving the deployment instruction from the simulation system, the model deployment execution module automatically calls the kubectl apply command, first applies the YAML format file of the K8s deployment management object Deployment, and then calls kubectlapply again to apply the YAML format file for creating the model container Pod, thereby deploying the model image to the Pod specified by the script file in sequence, and managing the Pod object through the Deployment object;
[0026] S5: After the simulation is finished, the deployment cancellation module deletes the model container deployed in this simulation; the specific method is:
[0027] After the simulation starts, the deployment cancellation module is in a listening state, waiting for the simulation end event. After receiving the simulation end event, the deployment cancellation module automatically calls the kubectl delete command to delete the model container Pod used in the simulation experiment scenario. When all model container Pods are deleted, the kubectl delete command is called again to delete the Deployment object used for this simulation task.
[0028] The beneficial effects of the present invention are:
[0029] 1. By calling the K8s command tool and the Docker command tool, the present invention allows the simulation personnel to directly perform simple operations such as clicking and selecting on the software interface of the microservice model deployment system, and then enter some simple parameters to complete the one-click production of the model image and the automatic deployment of the model container.
[0030] 2. When using the system of the present invention, there is no need to write any code throughout the process. The operation is simple and fast, which greatly reduces the deployment time of each simulation test in the service-oriented simulation architecture. It can free developers from a large amount of repetitive manual operation work and save the time and cost of simulation experiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] Figure 1 It is a flowchart of a microservice model deployment method based on K8s and Docker technology in an embodiment of the present invention.
[0032] Figure 2 It is a schematic diagram of the model image import process of the microservice model deployment method based on K8s and Docker technology in an embodiment of the present invention.
[0033] Figure 3 It is a schematic diagram of the composition of a microservice model deployment system based on K8s and Docker technology in an embodiment of the present invention.
[0034] Figure 4 It is a specific flow chart of the microservice model deployment method based on K8s and Docker technology in an embodiment of the present invention. DETAILED DESCRIPTION
[0035] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.
[0036] A microservice simulation model deployment system based on K8S and Docker technology, such as Figure 3 As shown, it includes a model image generation module, a model management module, a deployment design module, a deployment execution module, and a deployment revocation module. The model image generation module implements the one-click creation function of the model image based on the Docker operating environment; the model management module provides a management interface for model information and model images, on which models can be created, deleted, and queried, and the model image can be uploaded based on the Docker operating environment and Docker-Registry; the deployment design module provides an interface to configure the host, port number, container, and Deployment information of the model image deployment; the deployment execution module can deploy the model image to the corresponding host and container according to the designed deployment information based on the K8s operating environment; the deployment revocation module provides the function of recycling the model container after the simulation is completed.
[0037] Specifically, the model image generation module is used to make model images. After completing the development of the simulation model, by using the functions of this module, input the image name, version number, storage path, external port number, dependent packages and other information, click the generate operation, and automatically generate the Dockerfile file of the model image, and automatically execute the docker build command to generate the corresponding model image package;
[0038] The model management module is responsible for maintaining a MySQL database and a Docker-Registry private warehouse, managing models and model images, and can create, delete, and query models;
[0039] Specifically, when creating a model, you can enter the model name, basic properties, application scope and other information, and associate the corresponding model image with the model information. Then, based on the model information, form the Dockerfile script file of the model, and execute the Dockerfile file through the Docker command to build the model image, so as to encapsulate the model microservice model into a Docker container, and upload the encapsulated model to the Docker-Registry private warehouse.
[0040] In the model query operation, you can query all model information and model images in the current model management library;
[0041] In the model deletion operation, you can delete the selected model from the MySQL database and delete the corresponding model image in the Docker-Registry private repository;
[0042] The deployment design module is used to design the model image deployment location, including the deployed host, port number, and the relationship between Pod and Deployment. First, it reads the simulation scenario file and parses all the model services used in the simulation scenario. Then, it provides an interface to assist operators in planning and designing the model deployment host, IP address, port number, container, and Deployment information, and forms a deployment script file and a Deployment file in the yaml format required to create a model container. The names of these script files need to contain the simulation scenario information.
[0043] The deployment execution module is responsible for executing the deployment script file to deploy the model image to the corresponding container. Before the simulation scenario starts running, the deployment execution module is first run. This module searches for the corresponding K8s script file through the simulation scenario file, and then automatically executes these K8s scripts to deploy the model image required for the simulation scenario to the planned host and container, and start these models. After the deployment execution module monitors that all the models used have started normally, it notifies the simulation scenario to start running;
[0044] The deployment cancellation module is mainly used to recycle the model container required for the simulation scenario after the simulation scenario ends. The deployment cancellation module monitors the simulation end event in real time. When the simulation end event is received, the kubectl delete command is called to delete the model container used in this simulation scenario and monitor the status of the model container. When all model containers are deleted, the kubectl delete command is called to delete the Deployment object used for this simulation task.
[0045] A microservice simulation model deployment method based on K8S and Docker technology, comprising:
[0046] In the model image generation module, a human-computer interaction interface is provided to configure the model image name, version number, image storage path, external port number, dependent packages and other information, and then automatically generate a Dockerfile file based on this information, and then automatically run the docker build command to generate the model image;
[0047] In the model management module, you can import model images, configure basic model information, store model information in the MySQL database, and establish a mapping relationship between model information and model images. At the same time, call the docker upload command in the background service to upload the imported model images to the private image library Docker-Registry.
[0048] In the simulation model deployment design module, select the required simulation model image according to the simulation scenario. The model deployment design module provides a human-computer interaction interface, which can configure the host IP, port number, image name, etc. of the simulation model deployment, and then automatically generate the K8s script file of the deployment model image according to the configuration, including the yaml file required to create the Pod object and the yaml file required to create the Deployment object;
[0049] When performing simulation deduction and loading simulation scenarios, the simulation model deployment execution module is run. This module calls the kubectl command to execute the Pod and Deployment script files of the model image generated in the deployment design module, deploys the model image to the Pod specified by the script file in sequence, and manages the Pod object through the Deployment object.
[0050] After the simulation is completed, the simulation end event triggers the operation of the deployment cancellation module, which deletes the Pod corresponding to the model image used in the simulation scenario by calling the kubectl delete command; when all model Pods are deleted, the Deployment object used for this simulation task is deleted by calling the kubectl delete command.
[0051] The following is an example of a specific microservice simulation model deployment method. The prerequisite for this method is that the deployment of the Docker runtime environment and the K8s runtime environment has been completed, and the creation of the k8s simulation cluster has been completed. In this example, the deployment of the simulation model container Pod and the Deployment management object are both performed in the k8s simulation cluster. Figure 1 , the steps of this embodiment are as follows:
[0052] S1: Generate a model mirror image from the simulation model;
[0053] Specifically, according to the needs of the simulation task, develop the required model service, then open the human-computer interface of the model image generation module, create a model image, enter the model name, version number, model package path, image storage path, external port number, dependent operating environment, etc. After the input is completed, click "One-click creation" to replace the corresponding information in the template based on the built-in Dockerfile template in the image generation module, and use the FROM field to specify the built-in centos7.9 image in the model image generation module as the basic image to generate the Dockerfile file of the model image.
[0054] After generating the Dockerfile file of the model, the model image generation module automatically calls the docker build command to make the model service package into a model image with one click, and then automatically calls the docker save command to save the model image as a tar file and store it in the image storage path entered in the human-computer interface.
[0055] S2: Store the simulation model into the model database and upload the corresponding model image to the Docker-Registry;
[0056] Specifically, open the interface of the model management module, select import model, enter the model name, basic properties, model image tag, version number, and select the tar file of the model image. After confirming this information, the model management module automatically uses the model image name and calls the docker images command to match and query the model image ID. Then, the model name, basic information, model image tag, version number, and model image ID are stored in the model database MySql, and the dockerpush command is called to push the model image tar file to the Docker-Registry.
[0057] Figure 2 The specific process of importing the model into the database and mirror library in steps S1 and S2 is given.
[0058] S3: Deploy the design module in the simulation model, plan the deployment location of the model, including the host, port number, name, etc., and automatically generate the corresponding K8s script file;
[0059] Specifically, open the model deployment design module, first read the simulation scenario file formed by the simulation task, that is, the assumption file usually expressed in the general simulation system, obtain the identification number of the simulation model used in the simulation scenario, and query the model image ID, version number and model image tag corresponding to each model in the model database according to these identifications, and present the model name in the human-computer interface in the form of a list;
[0060] Furthermore, according to the host resource situation, the host, port number, container name and other information of each model deployment are configured in the human-computer interface. After the deployment design is completed, the model deployment design module uses the built-in K8s configuration file template in yaml format to automatically generate the yaml file for creating the model container Pod and the yaml file for the K8s deployment management object Deployment.
[0061] S4: In the model deployment execution module, execute the K8s script file, deploy the model image to the planned host and container, and automatically start the model;
[0062] Specifically, before the simulation starts, the simulation system sends a start deployment instruction to the deployment execution module. After receiving the deployment instruction, the kubectl apply command is automatically called to first apply the yaml file of the deployment management object Deployment, and then call kubectl apply again to apply the yaml file of the model container Pod.
[0063] In this embodiment, deployment is performed based on the K8s environment and the K8s simulation cluster. The management object Deployment of the model container Pod is created through the K8s apply command, and the model container Pod is created through the Deployment. The model microservice is executed in the Pod. Pod is the smallest management unit, in which only one container runs, and each Deployment object manages a Pod container.
[0064] S5: After the simulation is finished, the deployment cancellation module deletes the model container deployed in this simulation;
[0065] Specifically, after the simulation begins, the deployment cancellation module is in a listening state, waiting for the simulation end event. After receiving the simulation end event, the deployment cancellation module automatically calls the kubectl delete command to delete the model container Pod used in the simulation experiment scenario. If the deletion is successful, the kubectl delete command is called again to delete the management service Deployment object of the Pod container.
[0066] Through the above method, through K8s container orchestration technology and Docker container image generation technology, model containers can be conveniently and dynamically deployed and recycled according to the needs of simulation experiments, which greatly reduces the deployment difficulty and workload of the service-oriented simulation system, allowing simulation personnel to focus on simulation business.
[0067] Figure 4 The following is a specific process of model deployment using a microservice model deployment system, including the following steps:
[0068] 1) First, analyze the model requirements of the simulation task and check in the model library whether the existing models meet the simulation requirements;
[0069] 2) If the requirements are met, skip to step 6); otherwise, proceed to the next step;
[0070] 3) Develop simulation models and generate microservice packages of the models;
[0071] 4) Use the model image generation module to generate a model image for the simulation model microservice with one click;
[0072] 5) Use the model management module to store model information and upload the corresponding model image to the Docker-Registry;
[0073] 6) Use the deployment design module to read the simulation scenario file, obtain the model required for the simulation task, and provide an interface to assist in designing the model's deployment host, container, port number and other information, and generate a K8s script file for model deployment with one click;
[0074] 7) Use the deployment execution module to run the K8s script file, automatically pull the model image from the Docker Registry, create a container, deploy the model image to the designed container, and then start the model container;
[0075] 8) Monitor the simulation end event to determine whether the simulation has ended. If not, continue to monitor; if the simulation has ended, execute the next step;
[0076] 9) Call the deployment cancellation module to destroy the simulation model container used in this simulation task.
[0077] In short, the present invention allows simulation personnel to directly perform simple operations such as clicking and selecting on the software interface of the microservice model deployment system, and then complete the one-click creation of model images and automatic deployment of model containers by entering some simple parameters. No code needs to be written throughout the process, and the operation is simple and fast, which greatly reduces the deployment time of each simulation test in the service-oriented simulation architecture and saves the time and cost of simulation experiments.
[0078] The above contents are further detailed descriptions of the present invention in combination with specific preferred implementation cases, and the specific implementation methods of the present invention cannot be limited to these. Ordinary technicians in the technical field to which the present invention belongs can make some other modifications and changes to the above implementation cases without departing from the concept of the present invention. The scope of the present invention is defined by the attached claims and their equivalents.
Claims
1. A microservice simulation model deployment system based on K8s and Docker technology, characterized in that: It includes a model image generation module, a model management module, a deployment design module, a deployment execution module, and a deployment cancellation module; among which: Model image generation module is used to make model images. After the development of the simulation model is completed, the image name, version number, model package location, image storage path, external port number, and dependent package information are input through this module. According to this information, the Dockerfile file of the model image is automatically generated, and the docker build command is automatically executed to generate the corresponding model image package; The model management module is responsible for maintaining a MySQL database and a Docker-Registry private warehouse, managing models and model images, and providing operations for creating, deleting, and querying models; When creating a model, first input the model information, including name, basic properties, and version number, import the model image file, store the input model information in the Mysql database, and map the corresponding model image with the model information. At the same time, call the docker upload command in the background service to upload the imported model images to the Docker-Registry private warehouse. The query operation on the model is used to query all model information and model images in the current model management library; The delete operation on the model is used to delete the selected model from the MySQL database and delete the corresponding model image in the Docker-Registry private warehouse; The deployment design module is used to design the model image deployment location, including the deployed host, port number, name, and the relationship between Pod and Deployment. The module first reads the simulation scenario file, parses all model services used in the simulation scenario, and then provides a human-computer interaction interface. Through the human-computer interaction interface, the model deployment host, IP address, port number, container, and Deployment information are planned and designed, and a K8s script file is formed, as well as a yaml format file required to create a model container Pod and a yaml format file required to create a Deployment object. The names of the K8s script file and the yaml format file both contain information about the simulation scenario. The deployment execution module is responsible for executing the K8s script file and deploying the model image to the corresponding container. Before the simulation scenario starts running, the deployment execution module is first run. This module searches for the corresponding K8s script file through the simulation scenario file, and then automatically executes these K8s deployment scripts to deploy the model image required for the simulation scenario to the planned host and container, and start these models. After the deployment execution module monitors that all the models used have been started normally, it notifies the simulation scenario to start running. The deployment cancellation module is used to recycle the model container used in the simulation scenario after the simulation scenario ends. The deployment cancellation module monitors the simulation end event in real time. When the simulation end event is received, the kubectl delete command is called to delete the model container used in this simulation scenario, and the status of the model container is monitored. When all model containers have been deleted, the kubectl delete command is called to delete the Deployment object used in this simulation task.
2. A microservice simulation model deployment method based on K8s and Docker technology, characterized in that: The following steps are involved: S1: Generate the simulation model as a model mirror; the specific method is: According to the requirements of the simulation task, develop the required model service, then open the human-computer interface of the model image generation module, create a model image, enter the model name, version number, model package path, image storage path, external port number, and dependent packages, and replace the corresponding information in the template based on the built-in Dockerfile template in the image generation module. Use the FROM field to specify the built-in Linux image in the model image generation module as the base image, and generate the Dockerfile file of the model image; After generating the Dockerfile file of the model, the model image generation module automatically calls the dockerbuild command to make the model service package into a model image, and then automatically calls the docker save command to save the model image as a tar file and store it in the image storage path; S2: Store the simulation model in the model database and upload the corresponding model image to the Docker-Registry; the specific method is: Open the interface of the model management module, select Import Model, enter the model name, basic properties, model image tag, version number, select the tar file of the model image, the model management module automatically uses the model image name, calls the dockerimages command to match and query the model image ID, and then stores the model name, basic properties, model image tag, version number, and model image ID into the model database MySql, and calls the dockerpush command to push the model image tar file to the Docker-Registry private warehouse; S3: Plan the deployment location of the model in the simulation model deployment design module, including the host, port number, name, and the relationship between Pod and Deployment, and automatically generate the corresponding K8s script file; the specific method is: Open the model deployment design module, first read the simulation scenario file generated by the simulation task, obtain the identification number of the simulation model used in the simulation scenario, query the model image ID, version number and model image tag corresponding to each model in the model database based on these identifications, and present the model name in the human-computer interface in the form of a list; According to the host resource situation, configure the host, port number, and container name information of each model deployment in the human-computer interface. After completing the deployment design, the model deployment design module uses the built-in yaml format K8s configuration file template to automatically generate a yaml format file for creating a model container Pod and a yaml format file for the K8s deployment management object Deployment; S4: Execute the K8s script file in the model deployment execution module, deploy the model image to the planned host and container, and automatically start the model to start model simulation; the specific method is: After receiving the deployment instruction from the simulation system, the model deployment execution module automatically calls the kubectl apply command, first applies the yaml format file of the K8s deployment management object Deployment, and then calls kubectl apply again to apply the yaml format file for creating the model container Pod, thereby deploying the model image to the Pod specified by the script file in sequence, and managing the Pod object through the Deployment object; S5: After the simulation is finished, the deployment cancellation module deletes the model container deployed in this simulation; the specific method is: After the simulation starts, the deployment cancellation module is in a listening state, waiting for the simulation end event. After receiving the simulation end event, the deployment cancellation module automatically calls the kubectl delete command to delete the model container Pod used in the simulation experiment scenario. When all model container Pods are deleted, the kubectl delete command is called again to delete the Deployment object used for this simulation task.