A Dynamic Adaptive Computing Framework for Self-Provisioning Cloud Systems

• Main Authors: Bo-min Jiang, 江博閔
• Other Authors: Wei-Jen Wang
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/51503641073384861847

碩士 === 國立中央大學 === 資訊工程研究所 === 99 === The emerging cloud computing technology has become one of the urgent development goals in most developed countries. However, existing automatic/adaptive computing solutions in a cloud environment are still primitive. This work combines the concept of multi-agent systems, mobile agents, and cloud computing systems, and develops a framework to support self-provisioning, adaptability, and dynamic load balancing in a cloud environment. In the proposed framework, user can submit their applications, implemented as a group of mobile agents, to the proposed framework. When the system encounters the problem of system overloading or load imbalance, it will use two strategies to handle this problem. First, the mobile agents themselves can dynamically partition and redistribute the tasks to balance the load. Second, the system can notify the mobile agents to migrate to some free machines. If all resources are busy, the system can create more computing resources in the public cloud to increase the computing pool and to reduce the overall workload. In addition, the state of each application is periodically saved by the system to support fault tolerance.