| Popis: |
Cloud computing became recently a common paradigm in both industry and academia. Sensitive applications such as medical, military, and financial ones rely on clouds' platform. Therefore, services continuity should be guaranteed as much as possible. However, services are sensitive to the network Quality of Service (QoS). Heavy network traffic may have its negative impact on services execution in a cloud environment. In extreme cases, cloud services may be completely interrupted due to many reasons including natural disasters - such as earthquakes and tornadoes - and human-induced disasters ranging from hackers to terrorists. In such circumstances, services should be automated without human intervention in order to minimize losses as much as possible. In this paper, the Partitioned Services Layer (PSL) architecture is proposed. It categorizes services into real-time and non realtime. Since real-time services are critical, they should not be subject to QoS. Therefore, they run on a Distributed Autonomous Control (DAC) environment in which devices are connected through a Local Area Network (LAN). On the other hand, non real-time services are executed in the cloud environment to reduce costs. Nevertheless, basic services are automated by another autonomous system which is based on case reasoning. Finally, the proposed architecture is analyzed and compared to other related work. |
