Resource allocation and QoS control in evolving cloud computing systems

点击翻译按钮获取中文摘要

Cloud computing facilitates on-demand and scalable resource provisioning as services. This is implemented in a "pay-as-you go" manner, thus making resources available at all times from anywhere at affordable cost. This networking/service paradigm has gained considerable growth lately and is expected to be dominant in the future. Therefore, it is a rich area not only for academic research, but also an area of crucial importance for product/service development by Internet service/telecom providers. Cloud providers have to offer scalability along with reliable resources and services. To that end, Service Level Agreements (SLAs) state usage terms and penalties in case of performance violations. Provisioning of resources both at the computing as well as routing and service request scheduling levels should be performed in an "elastic" and scalable fashion. SLAs are constantly monitored with a goal to minimize performance violations while at the same time maximize user satisfaction and cloud provider revenue. Resource allocation and quality of service control and monitoring has been an active area of research in traditional data networks. However, the (geographically) distributed and diverse (in terms of service offered) nature of the "cloud" may be challenging to implement using traditional networking paradigms. Software Defined Networking (SDN) can contribute towards efficient, robust and readily reconfigurable cloud data centres. The goal of the proposal is to investigate present and evolving networking practices and their application for the advancement of next generation cloud computing systems and services.