Workflow engine automatic scheduling method of AI service platform
A technology of workflow engine and service platform, which is applied in the field of automatic scheduling of workflow engine of AI service platform, can solve problems such as AI service error reporting and AI service scheduling, and achieve the effect of improving completion rate and utilization rate
Pending Publication Date: 2022-01-18
SOUTH CHINA UNIV OF TECH
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] The purpose of the present invention is to overcome the problem of AI service scheduling in the prior art in the case of insufficient resources, and proposes a workflow engine automatic scheduling method for the AI service platform, which can solve the problem when the computing resources of the AI service platform are insufficient In this case, solve the problem of direct error reporting by the created AI service, and execute the created AI service within the user's acceptable waiting time by guiding the user to modify the parameter information of the AI service
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0021] The present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings, but the embodiments of the present invention are not limited thereto.
[0022] see figure 1 As shown, this embodiment discloses an automatic scheduling method for a workflow engine of an AI service platform, and the specific steps are as follows:
[0023] 1) The AI service platform customizes the process definition of AI services. The AI service platform customizes differentiable AI service process definitions for different users. For example, for model training, user A's service process includes image preprocessing, while user B does not. Different AI services have different process tasks. For AI services that do not include tasks that occupy computing resources, the AI service platform directly executes them in the order defined in the process until the end of the process. At the same time, the AI service platform maintains a que...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a workflow engine automatic scheduling method of an AI service platform, which comprises the following steps: 1) the AI service platform customizes a flow definition of AI service for a user, and maintains an AI service instance queue Q to be activated; 2) a user sets the values of an execution duration limit Tclimit and a waiting execution duration limit TwLimit for a task occupying computing resources in the AI service, and starts an instance of the AI service; 3) the AI service platform calculates the execution duration Tc of the task occupying the computing resource in the AI service instance started by the user; and 4) the AI service platform schedules the AI service instance started by the user according to the calculated execution duration Tc in combination with the use condition of the current computing resource of the AI service platform and the information of the queue to be activated. According to the invention, under the condition that the computing resources of the AI service platform are insufficient, the execution duration, the execution waiting duration and other information of the AI service are set, so that the completion rate of the AI service and the utilization rate of the computing resources of the AI service platform are improved within the acceptable waiting time range of a user.
Description
technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to an automatic scheduling method for a workflow engine of an AI service platform. Background technique [0002] With the rapid development of artificial intelligence technology, research and development of scientific research platforms that manage artificial intelligence algorithms are also in full swing. There are many general-purpose AI development platforms on the market, such as ModelArts launched by Huawei Cloud, and smart machine learning launched by Tencent Cloud. The platform TI-ONE and AWS Machine Learning provided by Amazon provide users with full-process development support from data preprocessing, model training, model evaluation to model service. [0003] However, the above general AI development platforms are all based on sufficient underlying resources. For university laboratories and research institutes, when resources such as GPUs are not su...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06F9/48G06F9/50
CPCG06F9/4881G06F9/485G06F9/5038
Inventor 胡金龙虢鲁新董守斌
Owner SOUTH CHINA UNIV OF TECH