An element-agnostic approach to auto-scaling network elements for telco cloud

点击翻译按钮获取中文摘要

This CRD project is to develop a new, unified approach to auto-scaling the capacity of virtual network elements (VNEs), such as virtual routers and virtual switches, for the telco cloud. The telco cloud refers to cloud resources owned and managed by telecommunication companies. By shifting network functions and services to cloud, the telco cloud provides telco operators with an unprecedented opportunity for offering their customers better services at a lower cost. One of the core technologies supporting this shift is the agile management of virtual network elements, which is the ultimate goal of this CRD project. Existing efforts in auto-scaling VNEs consider the problem in isolation. In particular, workload prediction and performance modeling are focused on one application without considering the dependency between multiple applications. A better solution is to enforce multiple VNEs to adapt their capacity collaboratively. We thus need to not only address the diversity of VNEs, but also consider the dependency among VNEs. Using network of queues, copula analysis, machine learning, and experimental test, this CRD project will tackle the above problems. This project's output consists of a general solution that is applicable for adaptive capacity scaling of different types of VNEs. As such, this CRD project directly benefits Canadian telecommunication industry and will strengthen the bridge between university research and industry development. The research involved will advance our knowledge in the innovative application of cloud computing in the telecommunication industry. The technology developed can be directly applied in the cloud-based applications and offerings of our industry partner, Ericsson Canada, and highly-qualified personnel (HQP) trained in the project will be valuable assets in the Canadian IT sector.