Robust Dynamic CPU Resource Provisioning in Virtualized Servers

We present robust dynamic resource allocation mechanisms to allocate application resources meeting Service Level Objectives (SLOs) agreed between cloud providers and customers. In fact, two filter-based controllers, i.e., $\mathcal {H}_{\infty }$ filter Maximum Correntropy Criterion Kalman (MCC-KF), are proposed. The controllers self-adaptive, with process noise variances covariances calculated using previous measurements within a time window. the process, bounded client mean response ( notation="LaTeX">$\mathop {\mathrm{mRT}}$ ) is maintained. Both deployed evaluated on an experimental testbed hosting RUBiS (Rice University Bidding System) auction benchmark web site. proposed offer improved performance under abrupt workload changes, shown via rigorous comparison current state-of-the-art. On our setup, Single-Input-Single-Output (SISO) can operate same server where performed; while Multi-Input-Multi-Output (MIMO) separate all data collected for decision making. SISO take decisions not dependent other system states (servers), albeit MIMO characterized by increased communication overhead potential delays. While over ones, latter enable more informed making framework problem of multi-tier applications.