Memory resource balancing for virtualized computing

Virtualization has become a common abstraction layer in modern data centers. By multiplexing hardware resources into multiple virtual machines (VMs) and thus enabling several operating systems to run on the same physical platform simultaneously, it can effectively reduce power consumption and building size or improve security by isolating VMs. In a virtualized system, memory resource management plays a critical role in achieving high resource utilization and performance. Insufficient memory allocation to a VM will degrade its performance dramatically. On the contrary, over-allocation causes waste of memory resources. Meanwhile, a VM’s memory demand may vary significantly. As a result, effective memory resource management calls for a dynamic memory balancer, which, ideally, can adjust memory allocation in a timely manner for each VM based on their current memory demand and thus achieve the best memory utilization and the optimal overall performance. In order to estimate the memory demand of each VM and to arbitrate possible memory resource contention, a widely proposed approach is to construct an LRU-based miss ratio curve (MRC), which provides not only the current working set size (WSS) but also the correlation between performance and the target memory allocation size. Unfortunately, the cost of constructing an MRC is nontrivial. In this dissertation, we first present a low overhead LRU-based memory demand tracking scheme, which includes three orthogonal optimizations: AVL-based LRU organization, dynamic hot set sizing and intermittent memory tracking. Our evaluation results show that, for the whole SPEC CPU 2006 benchmark suite, after applying the three optimizing techniques, the mean overhead of MRC construction is lowered from 173% to only 2%. Based on current WSS, we then predict its trend in the near future and take different strategies for different prediction results. When there is a sufficient amount of physical memory on the host, it locally balances its memory resource for the VMs. Once the local memory resource is insufficient and the memory pressure is predicted to sustain for a sufficiently long time, a relatively expensive solution, VM live migration, is used to move one or more VMs from the hot host to other host(s). Finally, for transient memory pressure, a remote cache is used to alleviate the temporary performance penalty. Our experimental results show that this design achieves 49% center-wide speedup.

Reducing main memory access latency through SDRAM address mapping techniques and access reordering mechanisms

As the performance gap between microprocessors and memory continues to increase, main memory accesses result in long latencies which become a factor limiting system performance. Previous studies show that main memory access streams contain significant localities and SDRAM devices provide parallelism through multiple banks and channels. These locality and parallelism have not been exploited thoroughly by conventional memory controllers. In this thesis, SDRAM address mapping techniques and memory access reordering mechanisms are studied and applied to memory controller design with the goal of reducing observed main memory access latency. The proposed bit-reversal address mapping attempts to distribute main memory accesses evenly in the SDRAM address space to enable bank parallelism. As memory accesses to unique banks are interleaved, the access latencies are partially hidden and therefore reduced. With the consideration of cache conflict misses, bit-reversal address mapping is able to direct potential row conflicts to different banks, further improving the performance. The proposed burst scheduling is a novel access reordering mechanism, which creates bursts by clustering accesses directed to the same rows of the same banks. Subjected to a threshold, reads are allowed to preempt writes and qualified writes are piggybacked at the end of the bursts. A sophisticated access scheduler selects accesses based on priorities and interleaves accesses to maximize the SDRAM data bus utilization. Consequentially burst scheduling reduces row conflict rate, increasing and exploiting the available row locality. Using a revised SimpleScalar and M5 simulator, both techniques are evaluated and compared with existing academic and industrial solutions. With SPEC CPU2000 benchmarks, bit-reversal reduces the execution time by 14% on average over traditional page interleaving address mapping. Burst scheduling also achieves a 15% reduction in execution time over conventional bank in order scheduling. Working constructively together, bit-reversal and burst scheduling successfully achieve a 19% speedup across simulated benchmarks.

Soft-decision equalization techniques for frequency selective MIMO channels

Multi-input multi-output (MIMO) technology is an emerging solution for high data rate wireless communications. We develop soft-decision based equalization techniques for frequency selective MIMO channels in the quest for low-complexity equalizers with BER performance competitive to that of ML sequence detection. We first propose soft decision equalization (SDE), and demonstrate that decision feedback equalization (DFE) based on soft-decisions, expressed via the posterior probabilities associated with feedback symbols, is able to outperform hard-decision DFE, with a low computational cost that is polynomial in the number of symbols to be recovered, and linear in the signal constellation size. Building upon the probabilistic data association (PDA) multiuser detector, we present two new MIMO equalization solutions to handle the distinctive channel memory. With their low complexity, simple implementations, and impressive near-optimum performance offered by iterative soft-decision processing, the proposed SDE methods are attractive candidates to deliver efficient reception solutions to practical high-capacity MIMO systems. Motivated by the need for low-complexity receiver processing, we further present an alternative low-complexity soft-decision equalization approach for frequency selective MIMO communication systems. With the help of iterative processing, two detection and estimation schemes based on second-order statistics are harmoniously put together to yield a two-part receiver structure: local multiuser detection (MUD) using soft-decision Probabilistic Data Association (PDA) detection, and dynamic noise-interference tracking using Kalman filtering. The proposed Kalman-PDA detector performs local MUD within a sub-block of the received data instead of over the entire data set, to reduce the computational load. At the same time, all the inter-ference affecting the local sub-block, including both multiple access and inter-symbol interference, is properly modeled as the state vector of a linear system, and dynamically tracked by Kalman filtering. Two types of Kalman filters are designed, both of which are able to track an finite impulse response (FIR) MIMO channel of any memory length. The overall algorithms enjoy low complexity that is only polynomial in the number of information-bearing bits to be detected, regardless of the data block size. Furthermore, we introduce two optional performance-enhancing techniques: cross- layer automatic repeat request (ARQ) for uncoded systems and code-aided method for coded systems. We take Kalman-PDA as an example, and show via simulations that both techniques can render error performance that is better than Kalman-PDA alone and competitive to sphere decoding. At last, we consider the case that channel state information (CSI) is not perfectly known to the receiver, and present an iterative channel estimation algorithm. Simulations show that the performance of SDE with channel estimation approaches that of SDE with perfect CSI.

Optimal game theoretic distributed control methodologies in small scale power systems

This dissertation presents the competitive control methodologies for small-scale power system (SSPS). A SSPS is a collection of sources and loads that shares a common network which can be isolated during terrestrial disturbances. Micro-grids, naval ship electric power systems (NSEPS), aircraft power systems and telecommunication system power systems are typical examples of SSPS. The analysis and development of control systems for small-scale power systems (SSPS) lacks a defined slack bus. In addition, a change of a load or source will influence the real time system parameters of the system. Therefore, the control system should provide the required flexibility, to ensure operation as a single aggregated system. In most of the cases of a SSPS the sources and loads must be equipped with power electronic interfaces which can be modeled as a dynamic controllable quantity. The mathematical formulation of the micro-grid is carried out with the help of game theory, optimal control and fundamental theory of electrical power systems. Then the micro-grid can be viewed as a dynamical multi-objective optimization problem with nonlinear objectives and variables. Basically detailed analysis was done with optimal solutions with regards to start up transient modeling, bus selection modeling and level of communication within the micro-grids. In each approach a detail mathematical model is formed to observe the system response. The differential game theoretic approach was also used for modeling and optimization of startup transients. The startup transient controller was implemented with open loop, PI and feedback control methodologies. Then the hardware implementation was carried out to validate the theoretical results. The proposed game theoretic controller shows higher performances over traditional the PI controller during startup. In addition, the optimal transient surface is necessary while implementing the feedback controller for startup transient. Further, the experimental results are in agreement with the theoretical simulation. The bus selection and team communication was modeled with discrete and continuous game theory models. Although players have multiple choices, this controller is capable of choosing the optimum bus. Next the team communication structures are able to optimize the players’ Nash equilibrium point. All mathematical models are based on the local information of the load or source. As a result, these models are the keys to developing accurate distributed controllers.

EVALUATING INFRASTRUCTURE PERFORMANCE TO ASSIST FACILITY MANAGEMENT TOWARD SUSTAINABLE SYSTEMS

Civil infrastructure provides essential services for the development of both society and economy. It is very important to manage systems efficiently to ensure sound performance. However, there are challenges in information extraction from available data, which also necessitates the establishment of methodologies and frameworks to assist stakeholders in the decision making process. This research proposes methodologies to evaluate systems performance by maximizing the use of available information, in an effort to build and maintain sustainable systems. Under the guidance of problem formulation from a holistic view proposed by Mukherjee and Muga, this research specifically investigates problem solving methods that measure and analyze metrics to support decision making. Failures are inevitable in system management. A methodology is developed to describe arrival pattern of failures in order to assist engineers in failure rescues and budget prioritization especially when funding is limited. It reveals that blockage arrivals are not totally random. Smaller meaningful subsets show good random behavior. Additional overtime failure rate is analyzed by applying existing reliability models and non-parametric approaches. A scheme is further proposed to depict rates over the lifetime of a given facility system. Further analysis of sub-data sets is also performed with the discussion of context reduction. Infrastructure condition is another important indicator of systems performance. The challenges in predicting facility condition are the transition probability estimates and model sensitivity analysis. Methods are proposed to estimate transition probabilities by investigating long term behavior of the model and the relationship between transition rates and probabilities. To integrate heterogeneities, model sensitivity is performed for the application of non-homogeneous Markov chains model. Scenarios are investigated by assuming transition probabilities follow a Weibull regressed function and fall within an interval estimate. For each scenario, multiple cases are simulated using a Monte Carlo simulation. Results show that variations on the outputs are sensitive to the probability regression. While for the interval estimate, outputs have similar variations to the inputs. Life cycle cost analysis and life cycle assessment of a sewer system are performed comparing three different pipe types, which are reinforced concrete pipe (RCP) and non-reinforced concrete pipe (NRCP), and vitrified clay pipe (VCP). Life cycle cost analysis is performed for material extraction, construction and rehabilitation phases. In the rehabilitation phase, Markov chains model is applied in the support of rehabilitation strategy. In the life cycle assessment, the Economic Input-Output Life Cycle Assessment (EIO-LCA) tools are used in estimating environmental emissions for all three phases. Emissions are then compared quantitatively among alternatives to support decision making.

The effect of demand, tank parameters, and pumping stations on energy use in municipal drinking water distribution systems

Two of the indicators of the UN Millennium Development Goals ensuring environmental sustainability are energy use and per capita carbon dioxide emissions. The increasing urbanization and increasing world population may require increased energy use in order to transport enough safe drinking water to communities. In addition, the increase in water use would result in increased energy consumption, thereby resulting in increased green-house gas emissions that promote global climate change. The study of multiple Municipal Drinking Water Distribution Systems (MDWDSs) that relates various MDWDS aspects--system components and properties--to energy use is strongly desirable. The understanding of the relationship between system aspects and energy use aids in energy-efficient design. In this study, components of a MDWDS, and/or the characteristics associated with the component are termed as MDWDS aspects (hereafter--system aspects). There are many aspects of MDWDSs that affect the energy usage. Three system aspects (1) system-wide water demand, (2) storage tank parameters, and (3) pumping stations were analyzed in this study. The study involved seven MDWDSs to understand the relationship between the above-mentioned system aspects in relation with energy use. A MDWDSs model, EPANET 2.0, was utilized to analyze the seven systems. Six of the systems were real and one was a hypothetical system. The study presented here is unique in its statistical approach using seven municipal water distribution systems. The first system aspect studied was system-wide water demand. The analysis involved analyzing seven systems for the variation of water demand and its impact on energy use. To quantify the effects of water use reduction on energy use in a municipal water distribution system, the seven systems were modeled and the energy usage quantified for various amounts of water conservation. It was found that the effect of water conservation on energy use was linear for all seven systems and that all the average values of all the systems' energy use plotted on the same line with a high R 2 value. From this relationship, it can be ascertained that a 20% reduction in water demand results in approximately a 13% savings in energy use for all seven systems analyzed. This figure might hold true for many similar systems that are dominated by pumping and not gravity driven. The second system aspect analyzed was storage tank(s) parameters. Various tank parameters: (1) tank maximum water levels, (2) tank elevation, and (3) tank diameter were considered in this part of the study. MDWDSs use a significant amount of electrical energy for the pumping of water from low elevations (usually a source) to higher ones (usually storage tanks). The use of electrical energy has an effect on pollution emissions and, therefore, potential global climate change as well. Various values of these tank parameters were modeled on seven MDWDSs of various sizes using a network solver and the energy usage recorded. It was found that when averaged over all seven analyzed systems (1) the reduction of maximum tank water level by 50% results in a 2% energy reduction, (2) energy use for a change in tank elevation is system specific, and (2) a reduction of tank diameter of 50% results in approximately a 7% energy savings. The third system aspect analyzed in this study was pumping station parameters. A pumping station consists of one or more pumps. The seven systems were analyzed to understand the effect of the variation of pump horsepower and the number of booster stations on energy use. It was found that adding booster stations could save energy depending upon the system characteristics. For systems with flat topography, a single main pumping station was found to use less energy. In systems with a higher-elevation neighborhood, however, one or more booster pumps with a reduced main pumping station capacity used less energy. The energy savings for the seven systems was dependent on the number of boosters and ranged from 5% to 66% for the analyzed five systems with higher elevation neighborhoods (S3, S4, S5, S6, and S7). No energy savings was realized for the remaining two flat topography systems, S1, and S2. The present study analyzed and established the relationship between various system aspects and energy use in seven MDWDSs. This aids in estimating the amount of energy savings in MDWDSs. This energy savings would ultimately help reduce Greenhouse gases (GHGs) emissions including per capita CO 2 emissions thereby potentially lowering the global climate change effect. This will in turn contribute to meeting the MDG of ensuring environmental sustainability.