The Experts below are selected from a list of 78 Experts worldwide ranked by ideXlab platform
Ngo Van Long - One of the best experts on this subject based on the ideXlab platform.
-
welfare implications of leadership in a Resource Market under bilateral monopoly
Social Science Research Network, 2010Co-Authors: Kenji Fujiwara, Ngo Van LongAbstract:Does a country strictly gain if it acts as a leader in a Resource Market under bilateral monopoly? Using differential games, we show that the answer is "yes" when leadership can be exercised globally (global Stackelberg leadership), but possibly "no" when it is exercised only at each stage (stagewise Stackelberg leadership). On the other hand, world welfare under Nash equilibrium is strictly higher than under global Stackelberg equilibrium. Regardless of which country is the leader, world welfare under stagewise Stackelberg leadership is higher than under global Stackelberg leadership.
Kenji Fujiwara - One of the best experts on this subject based on the ideXlab platform.
-
Welfare Implications of Leadership in a Resource Market under Bilateral Monopoly
Dynamic Games and Applications, 2011Co-Authors: Kenji Fujiwara, Ngo Van LongAbstract:Formulating a dynamic game model of a world exhaustible Resource Market, in this paper, we study welfare implications of Stackelberg leaderships for an individual country and the world. We overcome the problem of time-inconsistency by imposing a “credibility condition” on the Markovian strategy of the Stackelberg leader. Under this condition, we show that the presence of a global Stackelberg leader leaves the follower worse off relative to the Nash equilibrium. Moreover, the world welfare is highest in the Nash equilibrium as compared with the two Stackelberg equilibria.
-
welfare implications of leadership in a Resource Market under bilateral monopoly
Social Science Research Network, 2010Co-Authors: Kenji Fujiwara, Ngo Van LongAbstract:Does a country strictly gain if it acts as a leader in a Resource Market under bilateral monopoly? Using differential games, we show that the answer is "yes" when leadership can be exercised globally (global Stackelberg leadership), but possibly "no" when it is exercised only at each stage (stagewise Stackelberg leadership). On the other hand, world welfare under Nash equilibrium is strictly higher than under global Stackelberg equilibrium. Regardless of which country is the leader, world welfare under stagewise Stackelberg leadership is higher than under global Stackelberg leadership.
Jiang Qingye - One of the best experts on this subject based on the ideXlab platform.
-
Executing Large-Scale Workloads in Public Clouds
Engineering Computer Science, 2020Co-Authors: Jiang QingyeAbstract:In this research, we systematically study the execution of large scale workloads in public clouds. We analyze the microeconomics behavior of the global enterprise computing Resource Market. The analysis results clearly indicate that public clouds represent the future of the enterprise computing Resource Market. To address the challenges involved in migrating from traditional computing Resources to public clouds, we develop a set of methods to optimize the execution of large scale workloads in public clouds, using scientific workflow, quality of service, and video transcoding as examples. Furthermore, we study the limit of the horizontal scaling technique in public clouds by identifying sources of cloudscale bottlenecks, then quantitatively measuring their impact on the capacity of horizontally scalable applications. We start with an analysis on the price elasticity of demand of the global enterprise computing Resource Market, including the server sales business, server rental business, and public clouds. We reveal that from a microeconomics point of view public clouds are fundamentally different from traditional server sales business and server rental business. The analysis results clearly indicate that public clouds represent the future of the enterprise computing Resource Market. To address the challenges involved in migrating from traditional computing Resources to public clouds, we develop a set of techniques and tools to optimize the execution of large scale workloads in public clouds, using scientific workflow, quality of service, and video transcoding as examples. We present DEWE v3, a workflow management system with function-as-a-service (FaaS) as the target execution environment. DEWE v3 reduces the effort needed to execute large-scale scientific workflows. It liberates scientist from the tedious administrative tasks involved in the traditional cluster approach, allowing them to focus on their own research work. We present the design and implementation of Janus - a generic and scalable QoS framework for admission control purposes. We demonstrate that Janus achieves linear scalability both vertically and horizontally. We also use a photo sharing application to demonstrate that Janus can be used to provide QoS service for a wide range of SaaS applications. We analyze the challenges involved in large-scale video transcoding, leading to the design and implementation of our own video transcoding system. Large-scale evaluations indicate that the improved application maintains linear horizontal scalability at 10,100 vCPU cores. To understand the limit of public clouds, we develop ScaleBench as a distributed and parallel benchmark framework. ScaleBench is capable of generating substantial and sustainable workload on public clouds, simulating the Resource consumption pattern of various horizontally scalable applications. In particular, it detects cloud-scale bottlenecks in compute unit, block storage, networking and object storage. In addition, we use a real-life video transcoding application to demonstrate that the horizontal scaling technique can fail to gain more capacity when such cloud-scale bottlenecks are reached. We propose the concept of capacity degradation index (CDI) to describe the degree of capacity degradation at scale. We perform extensive empirical studies on four public clouds. We observe significant capacity degradation in three of them. This confirms that on multiple public clouds cloud-scale capacity bottlenecks not only exist, but also can be easily detected by an ordinary cloud user. With as little as 20 worker nodes, we observe up to 24%, 52%, 14% and 90% capacity degradation in overall system performance, block storage, networking and object storage, respectively. We conduct large-scale experiments using a real-life video transcoding application, where the largest worker fleet utilizes 3200 vCPU cores. We demonstrate that when the above-mentioned cloud-scale bottleneck is reached the capacity of the horizontally scalable application stops growing regardless of the growth in the number of nodes.Access is restricted to staff and students of the University of Sydney . UniKey credentials are required. Non university access may be obtained by visiting the University of Sydney Library
-
Executing Large-Scale Workloads in Public Clouds
Engineering Computer Science, 2020Co-Authors: Jiang QingyeAbstract:In this research, we systematically study the execution of large scale workloads in public clouds. We analyze the microeconomics behavior of the global enterprise computing Resource Market. The analysis results clearly indicate that public clouds represent the future of the enterprise computing Resource Market. To address the challenges involved in migrating from traditional computing Resources to public clouds, we develop a set of methods to optimize the execution of large scale workloads in public clouds, using scientific workflow, quality of service, and video transcoding as examples. Furthermore, we study the limit of the horizontal scaling technique in public clouds by identifying sources of cloudscale bottlenecks, then quantitatively measuring their impact on the capacity of horizontally scalable applications. We start with an analysis on the price elasticity of demand of the global enterprise computing Resource Market, including the server sales business, server rental business, and public clouds. We reveal that from a microeconomics point of view public clouds are fundamentally different from traditional server sales business and server rental business. The analysis results clearly indicate that public clouds represent the future of the enterprise computing Resource Market. To address the challenges involved in migrating from traditional computing Resources to public clouds, we develop a set of techniques and tools to optimize the execution of large scale workloads in public clouds, using scientific workflow, quality of service, and video transcoding as examples. We present DEWE v3, a workflow management system with function-as-a-service (FaaS) as the target execution environment. DEWE v3 reduces the effort needed to execute large-scale scientific workflows. It liberates scientist from the tedious administrative tasks involved in the traditional cluster approach, allowing them to focus on their own research work. We present the design and implementation of Janus - a generic and scalable QoS framework for admission control purposes. We demonstrate that Janus achieves linear scalability both vertically and horizontally. We also use a photo sharing application to demonstrate that Janus can be used to provide QoS service for a wide range of SaaS applications. We analyze the challenges involved in large-scale video transcoding, leading to the design and implementation of our own video transcoding system. Large-scale evaluations indicate that the improved application maintains linear horizontal scalability at 10,100 vCPU cores. To understand the limit of public clouds, we develop ScaleBench as a distributed and parallel benchmark framework. ScaleBench is capable of generating substantial and sustainable workload on public clouds, simulating the Resource consumption pattern of various horizontally scalable applications. In particular, it detects cloud-scale bottlenecks in compute unit, block storage, networking and object storage. In addition, we use a real-life video transcoding application to demonstrate that the horizontal scaling technique can fail to gain more capacity when such cloud-scale bottlenecks are reached. We propose the concept of capacity degradation index (CDI) to describe the degree of capacity degradation at scale. We perform extensive empirical studies on four public clouds. We observe significant capacity degradation in three of them. This confirms that on multiple public clouds cloud-scale capacity bottlenecks not only exist, but also can be easily detected by an ordinary cloud user. With as little as 20 worker nodes, we observe up to 24%, 52%, 14% and 90% capacity degradation in overall system performance, block storage, networking and object storage, respectively. We conduct large-scale experiments using a real-life video transcoding application, where the largest worker fleet utilizes 3200 vCPU cores. We demonstrate that when the above-mentioned cloud-scale bottleneck is reached the capacity of the horizontally scalable application stops growing regardless of the growth in the number of nodes
Li Layuan - One of the best experts on this subject based on the ideXlab platform.
-
a distributed utility based two level Market solution for optimal Resource scheduling in computational grid
Parallel Computing, 2005Co-Authors: Li Chunlin, Li LayuanAbstract:This paper investigates the interactions between agents representing users, services and Resources to solve Resource scheduling optimization in computational grid. In order to reduce the computational complexity, we further decompose the grid Resource allocation optimization into subproblems: grid user agent-grid service agent in service Market and grid service agent-grid Resource agent in Resource Market. Two-level Market converges to its optimal points; a globally optimal point is achieved. Total user benefit of the computational grid is maximized when the equilibrium prices are obtained through the service Market level optimization and Resource Market level optimization. It demonstrates a practical approach to Market responsive Resource pricing that can benefit grid providers and users alike. The paper presents two-level Market grid Resource pricing that is an iterative algorithm used to perform optimal Resource allocation. The experiment shows that two-level Market based Resource pricing scheme outperforms one level Market scheme in terms of task completion time and Resource allocation efficiency.
Ngo Van Long - One of the best experts on this subject based on the ideXlab platform.
-
Welfare Implications of Leadership in a Resource Market under Bilateral Monopoly
Dynamic Games and Applications, 2011Co-Authors: Kenji Fujiwara, Ngo Van LongAbstract:Formulating a dynamic game model of a world exhaustible Resource Market, in this paper, we study welfare implications of Stackelberg leaderships for an individual country and the world. We overcome the problem of time-inconsistency by imposing a “credibility condition” on the Markovian strategy of the Stackelberg leader. Under this condition, we show that the presence of a global Stackelberg leader leaves the follower worse off relative to the Nash equilibrium. Moreover, the world welfare is highest in the Nash equilibrium as compared with the two Stackelberg equilibria.