Method for service capacity expansion and shrinkage and related equipment

Abstract

The invention provides a method for service capacity expansion and shrinkage and related equipment. The method comprises the following steps: obtaining index information and a workload of a current period service, and obtaining a workload prediction value of a next period according to the workload by utilizing a workload prediction model; predicting the state of the service in the next period by using a state prediction model according to the workload prediction value and the index information; and according to the predicted state, determining an instance expansion and shrinkage strategy corresponding to the service and expanding and shrinking the instance corresponding to the service. According to the method, the instance expansion and shrinkage strategy can be determined in time, and theaccuracy of the determined instance expansion and shrinkage strategy is improved.

Description

technical field [0001] The present invention relates to the technical field of cloud computing, in particular to a method for expanding and contracting services and related equipment. Background technique [0002] As an emerging industry in recent years, cloud computing has received extensive attention from the scientific research community and the industrial circle. An important goal in cloud computing is that cloud service providers can save resource consumption as much as possible under the premise of guaranteeing the service level agreement (SLA), and only allocate resources actually needed by users. However, user needs are not static, and resources need to be flexibly and dynamically allocated to users. Therefore, automatic scaling strategies have become one of the core research contents in cloud computing. [0003] At present, a variety of automatic scaling strategies for cloud platforms have been proposed, such as scaling strategies based on thresholds, scaling strat...

AI Technical Summary

Problems solved by technology

However, these strategies have problems such as slow response speed and inaccurate scaling strategies in the application.

For example, when formulating an auto-scaling policy based on the Q-learning algorithm (a reinforcement learning algorithm), the slow convergence speed of the Q-learning algorithm will lead to untimely resource scheduling, which cannot guarantee that the SLA is met, affecting user experience

Images