Software as a Service (SaaS) in Cloud Computing offers reliable access to software\napplications for end users over the Internet without direct investment in infrastructure\nand software. SaaS providers utilize resources of internal datacenters or rent resources\nfrom a public Infrastructure as a Service (IaaS) provider in order to serve their customers.\nInternal hosting can increase cost of administration and maintenance, whereas hiring\nfrom an IaaS provider can impact quality of service due to its variable performance.\nTo surmount these challenges, we propose a knowledge-based admission control\nalong with scheduling algorithms for SaaS providers to effectively utilize public\nCloud resources in order to maximize profit by minimizing cost and improving customers�\nsatisfaction level. In the proposed model, the admission control is based on Service Level\nAgreement (SLA) and uses different strategies to decide upon accepting user requests for\nthat minimal performance impact, avoiding SLA penalties that are giving higher profit.\nHowever, because the admission control can make decisions optimally, there is a\nneed of machine learning methods to predict the strategies. In order to model\nprediction of sequence of strategies, a customized decision tree algorithm has been\nused. In addition, we conducted several experiments to analyze which solution in which\nscenario fit better to maximize SaaS provider�s profit. Results obtained through our\nsimulation shows that our proposed algorithm provides significant improvement\n(up to 38.4 % cost saving) compared to the previous research works
Loading....