CN113434275A - Remote batch deployment system and method for artificial intelligence algorithm model - Google Patents
Remote batch deployment system and method for artificial intelligence algorithm model Download PDFInfo
- Publication number
- CN113434275A CN113434275A CN202110729428.XA CN202110729428A CN113434275A CN 113434275 A CN113434275 A CN 113434275A CN 202110729428 A CN202110729428 A CN 202110729428A CN 113434275 A CN113434275 A CN 113434275A
- Authority
- CN
- China
- Prior art keywords
- algorithm model
- deployment
- node
- server
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/451—Execution arrangements for user interfaces
- G06F9/452—Remote windowing, e.g. X-Window System, desktop virtualisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5066—Algorithms for mapping a plurality of inter-dependent sub-tasks onto a plurality of physical CPUs
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Stored Programmes (AREA)
Abstract
The invention relates to a system and a method for remote batch deployment of an artificial intelligence algorithm model, which comprises an algorithm model management server, a plurality of algorithm model deployment node servers and a plurality of terminal devices, wherein the algorithm model management server is used for managing the algorithm model; the algorithm model management server is used for managing a plurality of algorithm model deployment node servers and configuring node services of the algorithm model deployment node servers in two ways; the algorithm model deployment node server is used for executing a job task of issuing deployment algorithm models to a plurality of terminal devices in batches; and the terminal equipment is used for receiving and using the algorithm model distributed and deployed by the algorithm model deployment node server. The deployment task allocation module can reasonably allocate tasks according to the performance pressure condition of the service node, fully utilize service resources and realize automatic control of service performance and improve efficiency; the plurality of node services simultaneously execute the deployment task in parallel and carry out full automatic processing without manual intervention operation.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a system and a method for remote batch deployment of an artificial intelligence algorithm model.
Background
With the rapid development of the artificial intelligence technology, the artificial intelligence algorithm model is more and more mature, the threshold of entrance is reduced with the appearance of a deep learning frame, codes do not need to be written from a complex neural network, the existing model can be selected according to needs, model parameters are obtained through training, the layer of the model can be increased on the basis of the existing model, or a classifier and an optimization algorithm which are needed by the user are selected at the top end; of course, as such, none of the frames are perfect, and the areas in which the different frames are applicable are not exactly the same; in general, the deep learning framework provides a series of deep learning components (for general algorithms, the implementation of the deep learning components can be realized), and when a new algorithm needs to be used, a user needs to define the new algorithm by himself, and then the function interface of the deep learning framework is called to use the new algorithm customized by the user. Therefore, currently, most of the time, existing algorithm models can be directly copied for use.
The traditional way for the application program to copy the pull algorithm model from the remote server is as follows: 1. copying the algorithm model from an artificial intelligence algorithm model training server to terminal equipment needing to use the algorithm model in a manual copying and pasting mode; 2. the login terminal equipment remotely copies the algorithm model from the artificial intelligence training server to the local terminal equipment by using a network; 3. the login terminal device selects to download the algorithm model to the local terminal device by using a special terminal application or a web application. However, the disadvantages of the conventional method are: point-to-point, one manual live application program copies the pull algorithm model from the remote server, and like the current downloading application of a mobile phone, the algorithm model cannot be deployed rapidly in batches, so that the efficiency is low; moreover, if a plurality of terminal equipment colleagues have huge pressure on the server through network downloading, the server cannot automatically control the pressure and the flow by adopting the mode.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a system and a method for remote batch deployment of an artificial intelligence algorithm model, and solves the defects of the traditional method.
The purpose of the invention is realized by the following technical scheme: a remote batch deployment system for an artificial intelligence algorithm model comprises an algorithm model management server, a plurality of algorithm model deployment node servers, a plurality of terminal devices and an algorithm model storage library;
the algorithm model management server is used for managing a plurality of algorithm model deployment node servers and configuring node services of the algorithm model deployment node servers in two ways;
the algorithm model deployment node server is used for executing a job task of issuing deployment algorithm models to a plurality of terminal devices in batches;
the terminal equipment is used for receiving and using the algorithm model distributed and deployed by the algorithm model deployment node server;
the algorithm model repository is used for storing algorithm model files and is called to pull and push by a plurality of algorithm model deployment node servers.
The algorithm model management server comprises a task progress collection module and a deployment task distribution module; the deployment task allocation module is used for inquiring the performance condition of each algorithm model deployment node server, sequencing according to performance integrals and preferentially issuing a task to the algorithm model deployment node server with the best performance state to execute the task; the task progress collecting module is used for obtaining the task execution progress of the algorithm model deployment node server.
The algorithm model management server configures the node service of the algorithm model deployment node server in two ways, including: deploying node service information of a node server through a WEB interface operation algorithm model; when the node service of the algorithm model deployment node server is started, the configuration information of the node server is automatically transmitted to the algorithm model management server, and the algorithm model management server modifies and adds the related service node information according to the node service of the existing algorithm model deployment node server.
The number of the terminal devices is larger than that of the algorithm model deployment node servers, each algorithm model deployment node server is communicated with at least one terminal device, and each terminal device is communicated with at least one algorithm model deployment node server.
A method for remote batch deployment of a system using an artificial intelligence algorithmic model, the method comprising:
s1, selecting one or more terminal devices to be deployed and one or more related algorithm models by a user and initiating a deployment task;
s2, the algorithm model management server analyzes the algorithm model deployment node servers and then distributes deployment tasks to each algorithm model deployment node server;
s3, each algorithm model deployment node server pushes the assigned algorithm models in the tasks to the assigned terminal equipment in batches according to the assigned tasks issued by the algorithm model management server;
the method further comprises the following steps:
s4, the algorithm model management server acquires the task execution progress in each algorithm model deployment node server in real time;
and S5, when the task execution progress in each acquired algorithm model deployment node server reaches 100%, indicating that the algorithm model is deployed in the corresponding terminal equipment.
The analysis of the algorithm model deployment node server by the algorithm model management server comprises the following steps:
inquiring the working performance condition of each algorithm model deployment node server, and calculating according to the performance data to obtain corresponding performance integral;
sorting the integrals according to the height, wherein the higher the integral is, the better the performance is;
and preferably issuing the human tasks to the algorithm model deployment node server with the best performance according to the sequencing condition.
The invention has the following advantages: a is used in artificial intelligence algorithmic model remote batch deployment system and method, users operate the model algorithm of batch deployment to the terminal installation through the administrative interface, it is simple and apt to use, needn't log in the terminal installation one by one to choose to download the algorithmic model or duplicate to the terminal installation; the deployment task allocation module can reasonably allocate tasks according to the performance pressure condition of the service node, fully utilize service resources and realize automatic control of service performance and improve efficiency; the plurality of node services simultaneously execute deployment tasks in parallel and perform comprehensive automatic processing without manual intervention operation; the distributed architecture system is adopted, so that more terminal devices can be corresponded by simply expanding node services when expansion is needed, node services can be reduced when more terminal devices are not needed to be managed, and resources are released to other services to ensure that the resources are fully utilized.
Drawings
FIG. 1 is a schematic diagram of the system of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the detailed description of the embodiments of the present application provided below in connection with the appended drawings is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application. The invention is further described below with reference to the accompanying drawings.
As shown in fig. 1, the invention adopts a remote delivery deployment mode, and a server can control batch delivery and deployment to each terminal device according to network and server performance indexes through multi-node management and control; the method adopts a WEB application mode for operation, but is not limited to WEB application, and a user can also operate the method in a command line mode; the method specifically comprises the following steps:
algorithm model management server (management service): the server role is mainly used for managing and storing the algorithm model file and configuring the node service by the following two ways: the method comprises the steps of configuring node service information (such as adding, deleting and modifying the node service information) through WEB interface operation; and when the node service is started (the node service points to the management service interface address through the configuration file), automatically transmitting the node service configuration information (the configuration information is stored in the configuration file) to the management service, and modifying and adding the related service node information by the management service according to the existing node service information.
Further, the algorithm model management server comprises a task progress collection module and a deployment task distribution module; the deployment task allocation module is used for inquiring the performance condition of each algorithm model deployment node server, sequencing according to performance integrals and preferentially issuing a task to the algorithm model deployment node server with the best performance state to execute the task; the task progress collecting module is used for obtaining the task execution progress of the algorithm model deployment node server.
The user can arrange the deployment task through the management service and directly observe the progress of the deployment task; the deployment task allocation module in the server allocates tasks according to performance pressure assumed by the node service executing the tasks.
Algorithm model deployment node server (node service): the server role can simultaneously execute the task of issuing the deployment algorithm model to the terminal equipment, can configure a plurality of node services, and can also deploy a plurality of node services under one server by using the design scheme of the container cloud. Each node can be connected with all terminal devices or connected in groups.
An algorithm model storage library: the algorithm model files are stored in the storage library in the same way, can be pulled and pushed by the service call of the node, and can be subjected to version control;
the terminal equipment: the user uses the terminal devices of the algorithm model, and the algorithm model is distributed to the devices and used.
The flow realized by the method comprises the following steps:
s1, selecting a terminal device to be deployed with the algorithm model and a relevant algorithm model by a user and initiating a deployment task;
s2, the deployment task allocation module reasonably allocates deployment tasks to each node for service through the following processes;
further, S21, inquiring the actual working performance of each node service (such as the number of unprocessed tasks served by the current node, the CPU occupancy rate, the memory use condition and the like), and comprehensively calculating a performance integral (weight: the number of unprocessed tasks (70%), the CPU occupancy rate (20%) and the memory use condition (10%));
s22, sorting the integrals according to height, wherein the higher the integral is, the better the performance is;
s23, issuing the task to the node with the best performance according to the sorting condition;
s3, each node service pushes the specified algorithm model in the task to the specified terminal equipment according to the task distributed by the task distribution module;
s4, each node service returns the execution condition to the task progress collection module in the task execution process;
and S5, the user can check the current task execution condition of each terminal device through the task progress interface, and prove that the algorithm model is deployed in the task progress interface when 100 percent is reached.
The foregoing is illustrative of the preferred embodiments of this invention, and it is to be understood that the invention is not limited to the precise form disclosed herein and that various other combinations, modifications, and environments may be resorted to, falling within the scope of the concept as disclosed herein, either as described above or as apparent to those skilled in the relevant art. And that modifications and variations may be effected by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (7)
1. A remote batch deployment system for an artificial intelligence algorithm model is characterized in that: the method comprises an algorithm model management server, a plurality of algorithm model deployment node servers, a plurality of terminal devices and an algorithm model storage library;
the algorithm model management server is used for managing a plurality of algorithm model deployment node servers and configuring node services of the algorithm model deployment node servers in two ways;
the algorithm model deployment node server is used for executing a job task of issuing deployment algorithm models to a plurality of terminal devices in batches;
the terminal equipment is used for receiving and using the algorithm model distributed and deployed by the algorithm model deployment node server;
the algorithm model repository is used for storing algorithm model files and is called to pull and push by a plurality of algorithm model deployment node servers.
2. The system for remote batch deployment of an artificial intelligence algorithmic model according to claim 1, characterized in that: the algorithm model management server comprises a task progress collection module and a deployment task distribution module; the deployment task allocation module is used for inquiring the performance condition of each algorithm model deployment node server, sequencing according to performance integrals and preferentially issuing a task to the algorithm model deployment node server with the best performance state to execute the task; the task progress collecting module is used for obtaining the task execution progress of the algorithm model deployment node server.
3. The system for remote batch deployment of an artificial intelligence algorithmic model according to claim 1, characterized in that: the algorithm model management server configures the node service of the algorithm model deployment node server in two ways, including: deploying node service information of a node server through a WEB interface operation algorithm model; when the node service of the algorithm model deployment node server is started, the configuration information of the node server is automatically transmitted to the algorithm model management server, and the algorithm model management server modifies and adds the related service node information according to the node service of the existing algorithm model deployment node server.
4. The system for remote batch deployment of an artificial intelligence algorithmic model according to claim 1, characterized in that: the number of the terminal devices is larger than that of the algorithm model deployment node servers, each algorithm model deployment node server is communicated with at least one terminal device, and each terminal device is communicated with at least one algorithm model deployment node server.
5. The method for the remote batch deployment system of the artificial intelligence algorithm model according to any one of claims 1-4, wherein: the method comprises the following steps:
s1, selecting one or more terminal devices to be deployed and one or more related algorithm models by a user and initiating a deployment task;
s2, the algorithm model management server analyzes the algorithm model deployment node servers and then distributes deployment tasks to each algorithm model deployment node server;
and S3, each algorithm model deployment node server pushes the assigned algorithm models in the tasks to the assigned terminal equipment in batches according to the assigned tasks issued by the algorithm model management server.
6. The method for the remote batch deployment system of the artificial intelligence algorithmic model according to claim 5, characterized in that: the method further comprises the following steps:
s4, the algorithm model management server acquires the task execution progress in each algorithm model deployment node server in real time;
and S5, when the task execution progress in each acquired algorithm model deployment node server reaches 100%, indicating that the algorithm model is deployed in the corresponding terminal equipment.
7. The method for the remote batch deployment system of the artificial intelligence algorithmic model according to claim 5, characterized in that: the analysis of the algorithm model deployment node server by the algorithm model management server comprises the following steps:
inquiring the working performance condition of each algorithm model deployment node server, and calculating according to the performance data to obtain corresponding performance integral;
sorting the integrals according to the height, wherein the higher the integral is, the better the performance is;
and preferably issuing the human tasks to the algorithm model deployment node server with the best performance according to the sequencing condition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110729428.XA CN113434275A (en) | 2021-06-29 | 2021-06-29 | Remote batch deployment system and method for artificial intelligence algorithm model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110729428.XA CN113434275A (en) | 2021-06-29 | 2021-06-29 | Remote batch deployment system and method for artificial intelligence algorithm model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113434275A true CN113434275A (en) | 2021-09-24 |
Family
ID=77757768
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110729428.XA Withdrawn CN113434275A (en) | 2021-06-29 | 2021-06-29 | Remote batch deployment system and method for artificial intelligence algorithm model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113434275A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114268661A (en) * | 2021-11-19 | 2022-04-01 | 科大讯飞股份有限公司 | Service scheme deployment method, device, system and equipment |
-
2021
- 2021-06-29 CN CN202110729428.XA patent/CN113434275A/en not_active Withdrawn
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114268661A (en) * | 2021-11-19 | 2022-04-01 | 科大讯飞股份有限公司 | Service scheme deployment method, device, system and equipment |
| CN114268661B (en) * | 2021-11-19 | 2024-04-30 | 科大讯飞股份有限公司 | Service scheme deployment method, device, system and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108924217B (en) | Automatic deployment method of distributed cloud system | |
| CN107766889B (en) | Cloud edge computing fused deep learning computing system and method | |
| CN107463582B (en) | Distributed Hadoop cluster deployment method and device | |
| CN112506444A (en) | Kubernetes cluster-based expansion and contraction capacity control method and device and electronic equipment | |
| WO2020168761A1 (en) | Model training method and apparatus | |
| CN109814951A (en) | The combined optimization method of task unloading and resource allocation in mobile edge calculations network | |
| CN111414233A (en) | Online model reasoning system | |
| CN109150987B (en) | Two-layer container cluster elastic expansion method based on host layer and container layer | |
| CN104699736B (en) | A kind of distributed larger scale data acquisition system and method based on movable equipment | |
| CN111880936B (en) | Resource scheduling method, device, container cluster, computer equipment and storage medium | |
| CN113051075A (en) | Kubernetes intelligent capacity expansion method and device | |
| CN114090176A (en) | Kubernetes-based container scheduling method | |
| CN110933178B (en) | Method for adjusting node configuration in cluster system and server | |
| CN110086664A (en) | A kind of access device fault handling method and device | |
| CN107979498B (en) | Mesh network cluster and large file transmission method based on cluster | |
| CN111597014B (en) | Secondary-mirror-image-free release method for containerized JavaEE application server | |
| CN111612132A (en) | Artificial intelligence algorithm development system, training method, device and medium | |
| CN113434275A (en) | Remote batch deployment system and method for artificial intelligence algorithm model | |
| CN116231860B (en) | Cloud edge end cooperation-based intelligent power load identification system, method and equipment | |
| JP2021121921A (en) | Method and apparatus for management of artificial intelligence development platform, and medium | |
| CN112685179A (en) | Resource deployment system and method based on cost on cloud | |
| CN112261125B (en) | Centralized unit cloud deployment method, device and system | |
| CN110809043A (en) | Data processing method, device, equipment and storage medium based on Internet of things | |
| CN117435306A (en) | Cluster container expansion and contraction method, device, equipment and storage medium | |
| CN108595178A (en) | A kind of collecting method, device and equipment based on hook |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210924 |