Containerized resource provisioning framework for multimedia big data application

Ye Tao

doi:10.1007/s11042-017-5366-6

Containerized resource provisioning framework for multimedia big data application

2018-02-27

2

PRE

AI

DOI：10.1007/s11042-017-5366-6

原文链接

原文求助

摘要

En 中文

Container, as a light-weight virtualization solution, provides secure and effective approaches to control and limit access to resources for multimedia data and applications. However, due to the complexity of the containerized computing environment, setting up runtime configuration presents a great challenge for non-computational domain specialists without much knowledge of service-oriented computing and virtualization. In this paper, fuzzy-logic-based approaches are proposed to simplify the user preferences representation and automate the processes of container environment setup. By using fuzzy inference techniques, the approach allows users to define non-quantifiable factors and policies to represent their preferences, and automatically converts the vague requirements to numeric parameters and runtime deployment. Compared to classical methods, the proposed approach presents only the information relevant to user's requirements and preferences. The validation results show that with appropriate customization steps and natural interfaces, user preferences can be reflected effectively in the final configurations of containers. Furthermore, a fuzzy-logic-based schedule algorithm for global container resource allocation is also proposed, and the effectiveness of the provisioning policies are validated by sample use cases.

Keyword:

Containerized computing

Fuzzy inference system

Intuitionistic fuzzy value

User preference

Resouce provisioning

Containerized resource provisioning framework for multimedia big data application

摘要

AI总结

期刊

机构

引用论文