ADAPTABILITY ANALYSIS OF CLOUD ENVIRONMENT AND LOAD PREDICTION ALGORITHM

Abstract

In the current cloud environment, resource scheduling is an important research field aimed at effectively managing and allocating cloud computing resources to meet user needs and optimize system performance (Yu, 2021). However, resource scheduling and load prediction are two closely related concepts that influence and depend on each other in the cloud environment (Kumar & Sharma, 2020). Load prediction provides an important reference for resource scheduling (Niri et al., 2020; L. Zhang et al., 2021a). By accurately predicting the load situation, resources can be allocated and adjusted in advance before load fluctuations occur, avoiding problems of resource shortage or waste. At the same time, load prediction can also help resource scheduling algorithms better understand load patterns and trends, thereby formulating more reasonable scheduling strategies. It can be said that to a certain extent, load prediction is the basis for resource scheduling. How to carry out precise load prediction has become a typical challenge faced by current research on cloud computing scheduling optimization. This paper first analyses the characteristics of the cloud environment and finds that there are problems such as increasingly obvious dynamic load characteristics, diversified resource requirements, and poor reliability of workflow task execution (Saif et al., 2021; Zhou et al., 2020). Then, starting from the dynamic characteristics of the cloud environment, this paper summarizes and analyzes its impact on cloud resource scheduling (Cao et al., 2022; Peng et al., 2020), and outlines the limitations of traditional load prediction methods (Sideratos et al., 2020; L. Zhang et al., 2021b)in view of the non-stable characteristics of dynamic changes in resource utilization in the cloud environment. The contribution of this paper is to propose a decomposition-prediction algorithm that reduces the impact of the above uncertainties on scheduling by predicting the host load.

Keywords: Cloud environment, dynamic load characteristics, resource scheduling, load prediction methods, prediction algorithm