No abstract

No abstract

Meeting the performance specifications of consolidated web services in a data center is a challenging research problem, since the control of the underlying cloud computing infrastructure must meet the Service Level Agreement (SLA) requirements and satisfy the system's constraints. In this article, we address the admission control and resource allocation problem jointly, by establishing a unified modeling and control framework. Convergence to a desired reference point and stability and feasibility of the control strategy are guaranteed, while achieving high performance of the co-hosted web applications. The efficacy of the proposed approach is illustrated in a real testbed.

Abstract-Service providers must guarantee high quality of service (QoS) for each web application in a data center and simultaneously achieve optimal utilization of their infrastructure. Meeting the Service Level Objectives (SLOs), such as response time in a dynamic environment with a dense load and varying capacity, and simultaneously minimizing the energy consumption of the data center is an open research problem. This paper presents a control framework that addresses both problems of load balancing and resource allocation of consolidated web services in cloud computing infrastructure. The proposed approach aims at succeeding the customer requirements described in a Service Level Agreement (SLA) while maximizing server utilization. A hierarchical twolayer controller is established. The local (lower) level controllers determine the capacity and admitted workload of Virtual Machines (VMs), which correspond to a set of feasible operating points with performance guarantee. The global (upper) level decides the number and topology of active VMs that serve the total service demand and activates only the minimum number of servers. The cooperation of the two control layers ensures the system stability against the fluctuations of incoming requests and the system constraints.

SUMMARYNowadays traffi c monitoring and analysis tools provide poor information about traffi c volume without giving any clear view of what the hidden rules and relationships that govern these fl ows are. Since the majority of fl ows is generated by services (web browsing, email, p2p) and most of these applications are dependent on many network assets (servers and databases) we should discover the underlying relationships of every application. We present a technique that discovers the hidden relationships among components of a network that consist of parts of specifi c applications. From time information and fl ow attributes, such as IP addresses and service ports, our method using a novel hybrid genetic algorithm produces a small set of fuzzy rules that can reveal the underlying relationships over a network without any guidance. These dependencies build a service graph which can become a useful tool for fault localization, monitoring service performance, designing changes and anomaly detection.