User satisfaction perception-based service placement method for data centers

Abstract

The invention discloses a user satisfaction perception-based service placement method for data centers. The method comprises the following steps: first sequencing expenditures of all the data centers at a moment t from low expenditures to high expenditures; then sequentially allocating service requests to each data center under the conditions that the center data has enough virtual machines and that the requirements of the user requests are met, reserving a group of allocation with maximum profit for a computing service provider consistent with the conditions, and placing service in the data center; and after the corresponding service is placed for all the requests, finishing the service placement of the round, and making preparations for the next moment. According to the method, the service is placed in the data centers for the user requests with minimum expenditures by taking the maximization of return of the service providers as a target on the premise of meeting satisfaction of users; and the method is easy in computation, high in operation speed and applied to online large-scale service request scenes.

Description

Technical field [0001] The present invention is a data center service deployment method based on user satisfaction perception, and particularly relates to service providers deploying services in geographically distributed data centers to maximize service providers through the service deployment method on the premise of ensuring user satisfaction Income. Background technique [0002] With the rapid development of cloud computing, service providers provide a variety of applications in the cloud to meet the needs of users, such as existing applications such as Google’s Docs and AppEngine, Amazon’s EC2 and S3, etc. These application services are mostly Realized through the arrangement in the data center. Therefore, a large number of service requests will not only increase the demand on the data center, but also increase the power consumption of the data center. According to relevant statistics, Google's data center has about 500,000 servers, and the cost of electricity for a year o...

AI Technical Summary

Problems solved by technology

The first type mainly studies the energy consumption and time delay of traditional data center processing requests. The problem was raised earlier, and there are a lot of researches; the second type of problem is an extension of the first type of problem, which introduces The new target parameters, the model also mainly refers to the original multi-core scheduling problem, and does not reflect the characteristics of the data center; the third type is also an extension of the first type of problem, mainly by introducing the existing real-time system utility function to measure user performance. Income problem, in this method, one soft job corresponds to multiple utility equations, which can only be applied when the problem scale is small; the fourth method is to solve the problem through theoretical proof, without doing simulation or simulation experiments

Images