No description available.

No description available.

No description available.

No description available.

No description available.

MapReduce is a parallel programming paradigm used for processing huge datasets on certain classes of distributable problems using a cluster. Budgetary constraints and the need for better usage of resources in a MapReduce cluster often make organizations rent or share hardware resources for their main data processing and analysis tasks. Thus, there may be many competing jobs from different clients performing simultaneous requests to the MapReduce framework on a particular cluster. Schedulers like Fair Share and Capacity have been specially designed for such purposes. Administrators and users run into performance problems, however, because they do not know the exact meaning of different task scheduler settings and what impact they can have with respect to the resource allocation scheme across organizations for a shared MapReduce cluster. In this work, Capacity Scheduler is integrated into an existing MRPERF simulator to predict the performance of MapReduce jobs in a shared cluster under different settings for Capacity Scheduler. A few case studies on the behaviour of Capacity Scheduler across different job patterns etc. using integrated simulator are also conducted.

The rapid pace of climate and environmental changes requires some degree of adaptation, to forestall or avoid severe impacts. Adaptive capacity and water security are concepts used to guide the ways in which resource managers plan for and manage change. Yet the assessment of adaptive capacity and water security remains elusive, due to flaws in guiding concepts, paucity or inadequacy of data, and multiple difficulties in measuring the effectiveness of management prescriptions at scales relevant to decision-making. We draw on conceptual framings and empirical findings of the articles in this special issue and seek to respond to key questions with respect to metrics for the measurement, governance, information accessibility, and robustness of the knowledge produced in conjunction with ideas related to adaptive capacity and water security. Three overarching conclusions from this body of work are (a) systematic cross-comparisons of metrics, using the same models and indicators, are needed to validate the reliability of evaluation instruments for adaptive capacity and water security, (b) the robustness of metrics to applications across multiple scales of analysis can be enhanced by a “metrics plus” approach that combines well-designed quantitative metrics with in-depth qualitative methods that provide rich context and local knowledge, and (c) changes in the governance of science-policy can address deficits in public participation, foster knowledge exchange, and encourage the co-development of adaptive processes and approaches (e.g., risk-based framing) that move beyond development and use of static indicators and metrics.

PhD / In recent years, mobile communication networks have experienced significant evolution. The 3G mobile communication system, UMTS, employs WCDMA as the air interface standard, which leads to quite different mobile network planning and dimensioning processes compared with 2G systems. The UMTS system capacity is limited by the received interference at NodeBs due to the unique features of WCDMA, which is denoted as `soft capacity'. Consequently, the key challenge in UMTS radio network planning has been shifted from channel allocation in the channelized 2G systems to blocking and outage probabilities computation under the `cell breathing' effects which are due to the relationship between network coverage and capacity. The interference characterization, especially for the other-cell interference, is one of the most important components in 3G mobile networks planning. This monograph firstly investigates the system behavior in the operation of UMTS uplink, and develops the analytic techniques to model interference and system load as fully-characterized random variables, which can be directly applicable to the performance modeling of such networks. When the analysis progresses from single-cell scenario to multi-cell scenario, as the target SIR oriented power control mechanism is employed for maximum capacity, more sophisticated system operation, `feedback behavior', has emerged, as the interference levels at different cells depend on each other. Such behaviors are also captured into the constructed interference model by iterative and approximation approaches. The models are then extended to cater for the features of the newly introduced HSUPA, which provides enhanced dedicated channels for the packet switched data services such that much higher bandwidth can be achieved for best-effort elastic traffic, which allows network operators to cope with the coexistence of both circuit-switched and packet-switched traffic and guarantee the QoS requirements. During the derivation, we consider various propagation models, traffic models, resource allocation schemes for many possible scenarios, each of which may lead to different analytical models. All the suggested models are validated with either Monte-Carlo simulations or discrete event simulations, where excellent matches between results are always achieved. Furthermore, this monograph studies the optimization-based resource allocation strategies in the UMTS uplink with integrated QoS/best-effort traffic. Optimization techniques, both linear-programming based and non-linear-programming based, are used to determine how much resource should be assigned to each enhanced uplink user in the multi-cell environment where each NodeB possesses full knowledge of the whole network. The system performance under such resource allocation schemes are analyzed and compared via Monte-Carlo simulations, which verifies that the proposed framework may serve as a good estimation and optimal reference to study how systems perform for network operators.

In this dissertation we investigate joint optimal capacity investment, pricing and production decisions for a multinational manufacturer who faces exchange rate uncertainties. We consider a manufacturer that sells its product in both domestic and foreign markets over a multiperiod season. Because of long-lead times, the capacity investment must be committed before the selling season begins. The exchange rate between the two countries fluctuates across period and the demand in both markets is price dependent. In the first part, the model considers three scenarios: (1) early commitment to price and quantity with central sourcing, (2) postponement of prices and quantities with central sourcing, and (3) local sourcing. We derive the optimal capacity and the optimal prices for each scenario, and investigate the impact of the exchange rate parameters and the length of the selling season on optimal capacity investment, production allocation, and pricing decisions. We observe that while the price and production decisions in the domestic market are independent of the exchange rate under early commitment and local sourcing scenarios, the exchange rate between two countries directly impacts these decisions under the postponement setting. We identify thresholds and gain insights on investment costs, market potentials, exchange rate drifts, and selling season length for the choice of entering a foreign market under all scenarios. In the second part of this dissertation, we consider a duopoly competition in the foreign country. We consider a single period setting and we model the exchange rate as a random variable. We assume two scenarios: (1) Exogenous Model, where the price of the foreign manufacturer is set a priori, and (2) Endogenous Model, where the prices are set simultaneously based on a Nash Game outcome. In the Exogenous Model, we study the impact of exchange rate and foreign manufacturer's price on optimal capacity and prices. In the Endogenous Model, we investigate the impact of competition and exchange rate on optimal capacities and optimal prices. We show how competition can impact the decision of the home manufacturer to enter the foreign market.

The primary goal of this dissertation is to gain a better understanding, in thecontext of a dynamic stochastic general equilibrium framework, of the role of inventories and capacity utilization (of both capital and labor) and, in particular, therelationship among them. These are variables which have long been recognized asplaying an important role in the business cycle. An analysis of the association between inventories and capital utilization seems natural, for physical capital could beseen as a stock ultimately destined to be transformed into an inventory of finishedgoods. In the same way, inventories could be seen as a stock of physical capital already transformed into finished goods. Introducing variable rates of utilization ofcapacity, then both can be seen as providing a short-run adjustment "buffer stock"mechanism.The analysis of the relationship between those variables is centered on the effectsof two possible shocks: preference (demand) shocks and technology shocks. Impulse-response experiments show that inventories and the rate of capital utilization aremostly complements, while inventories and the rate of labor utilization are mostlysubstitutes. Moreover, low-persistence shocks emphasize the role of inventories asbeing a "shock absorber", whereas high-persistence shocks emphasize the role of inventories as being a complement to consumption. Consistent with the stylized facts inthe literature, simulation results show that inventory holdings are pro-cyclical, while the inventory-to-sales ratio is counter-cyclical.Two additional "themes" are explored. The first has to do with the treatmentof uncertainty and the consequences of using, as it is done in most of the literature, afirst-order approximation. By approximating the decision rules to a second order, weobserve that higher exogenous uncertainty enhances the importance of the precautionary motive to holding inventories. The second additional theme is a more generalframework for the analysis of capital utilization. We find that the two most commonways of modeling capital utilization can t in a more general specification that incorporates spending on capital maintenance. Though the aforementioned results do notvary qualitatively after that concept is introduced, quantitative answers do.