Title:
SCHEDULING METHOD AND APPARATUS BASED ON MICROSERVICE LINK ANALYSIS AND REINFORCEMENT LEARNING
Document Type and Number:
WIPO Patent Application WO/2023/185090
Kind Code:
A1
Abstract:
The present invention relates to a scheduling method and apparatus based on microservice link analysis and reinforcement learning. According to the method and the apparatus, a cloud server cluster is scheduled on the basis of a deep learning model, and three different scheduling modes can be selected for reinforcement learning: vertical scaling, horizontal scaling, and brownout. According to the present invention, a workload and link analyzer based on a deep learning model is used to perform analysis and decision-making on a link of a microservice to select a key path having the longest delay and a key node. According to the present invention, load data of a cloud server is trained by a reinforcement learning algorithm based on deep Q-learning, and the algorithm is used to train a deep learning model suitable for different load states. The present invention solves the possible problem that a link length and a microservice delay do not match, and can obtain an optimization target for a delay itself; and resource scheduling performed on a cloud server according to the optimization target can effectively relieve the existing problem of microservice link delays.
More Like This:
Inventors:
XU MINXIAN (CN)
SONG CHENGHAO (CN)
YE KEJIANG (CN)
XU CHENGZHONG (CN)
SONG CHENGHAO (CN)
YE KEJIANG (CN)
XU CHENGZHONG (CN)
Application Number:
PCT/CN2022/138189
Publication Date:
October 05, 2023
Filing Date:
December 09, 2022
Export Citation:
Assignee:
SHENZHEN INST ADV TECH (CN)
International Classes:
G06F9/50
Foreign References:
| CN114780233A | 2022-07-22 | |||
| CN113553149A | 2021-10-26 | |||
| CN113553150A | 2021-10-26 | |||
| US20210304063A1 | 2021-09-30 |
Attorney, Agent or Firm:
BEIJING ZHONG XUN TONG DA INTELLECTUAL PROPERTY AGENCY CO., LTD. (CN)
Download PDF: