Resource dynamic adjustment method and device, equipment and storage medium

A dynamic adjustment and resource technology, applied in the field of training resource adjustment, can solve problems such as long training time and unreasonable use of resources, and achieve the effect of achieving reasonable application and reducing training time
CN110502340AInactive Publication Date: 2019-11-26GUANGDONG INSPUR BIG DATA RES CO LTD
4 Cites 21 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Current Assignee / Owner
GUANGDONG INSPUR BIG DATA RES CO LTD
Publication Date
2019-11-26
Estimated Expiration
Not applicable Β· inactive patent

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a resource dynamic adjustment method, device and equipment based on a Kubernetes cluster and a computer readable storage medium. In the present application, in the training process of a training task, according to the embodiment of the invention, the adjustment information can be generated according to the idle resource information of the cluster and the current resource use information of each training task, so as to recreate the container group of the training tasks through the adjustment information, thereby achieving the dynamic adjustment of the use resources of the training tasks, achieving the reasonable application of cluster resources, and greatly reducing the training time.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates to the technical field of training resource adjustment, and more specifically, relates to a method, device, device and computer-readable storage medium for dynamically adjusting resources based on Kubernetes clusters. Background technique

[0002] When performing deep learning training tasks, container technology can currently be used to perform deep learning training tasks. This method is a fast and effective training method that eliminates some system configuration and environment dependencies for users. In addition, Kubernetes (container cluster management technology) to manage containers, enabling containerized training for large-scale tasks. However, when using container technology to perform deep learning training tasks, each resource is preset and cannot be changed during the training process. Therefore, when the training model is complex or the training data scale is large, failures due to unreasonable use of resou...

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