Due to increasing circuit densities and data throughput rates, power consumption has become a significant concern in the design and operation of high-performance packet switches. We extend the idea of Dynamic Power Management (DPM) to input queued switches, allowing operators to tradeoff power and delay in a useful way. We frame the problem as a dynamic program and solve a relaxation using techniques from Linear Quadratic Regulation (LQR). This optimal policy is combined with existing, nonpower-aware switch controls to generate two novel scheduling algorithms: 1) LQR Power Aware Maximum Weight Matching (LQR PA MWM) and 2) LQR Power Aware Projective Cone Scheduling (LQR PA PCS). Simulation results suggest that our algorithms result in significant power savings compared to MWM and previous power control schemes with little performance degradation.

This paper studies the slotted time Queueing / Switching Service Structure (QSSS) problem, where service configurations are selected dynamically in response to queue workload backlogs, so as to maintain stability (inflow-outflow balance) when possible. The only assumption on the traffic traces is that the incoming workload in each slot is bounded by some global (per trace), finite, burst ceiling. First, a key condition is identified, based on which each traffic trace can be classified as either admissible or non-admissible. Non-admissible traces drive the QSSS unstable, irrespective of the service configuration algorithm. Under admissible traces, the QSSS will remain stable when a Stabilizing Scheduling Algorithm (SSA) is utilized to control the service configuration. It is shown that Maximum-Weight-Matching (MWM) and Projective-Cone- Scheduling (PCS) belong to the SSA family. In addition, useful, theoretical tools are developed to facilitate trace classification. Second, the impact of rate management (i.e. power management) on stability is theoretically explored. In particular, under mild assumptions, it is proven that 'low workload backlog' decisions have little effect on the stability of a QSSS, provided that above a certain backlog threshold SSA-based service configuration controls are used.

We investigate the important problem of patching vulnerabilities in highly utilized data centers with heterogeneous groups of servers. In particular, our goal is to select the patching rate so as to efficiently (if not optimally), balance the trade-off between service disruption and risk exposure due to potential exploitation of vulnerabilities. We formulate the problem using a dynamic programming approach, that captures the aforementioned tradeoff, and study the structural properties of the optimal solution. Furthermore, we focus on insightful special cases and develop low-complexity justified heuristics, which achieve significant performance gains over standard benchmarks. We also demonstrate that the heuristics are very efficient, in the sense that they perform very close to the optimal solution obtained via dynamic programming.

We investigate autonomic power control policies for Internet servers and data centers. In particular, by monitoring the system load and thermal status, we decide how to vary the utilized processing resources to achieve acceptable delay and power performance. We formulate the problem using a dynamic programming approach that captures the power-performance tradeoff. We study the structural properties of the optimal solution and develop low-complexity justified heuristics, which achieve significant performance gains over standard benchmarks. The performance gains are higher when the load exhibits stronger temporal variations. We also demonstrate that the heuristics are very efficient, in the sense that they perform very close to the optimal solution obtained via dynamic programming.