Presents the introductory welcome message from the conference proceedings. May include the conference officers' congratulations to all involved withthe conference event and publication of the proceedings record. ...
详细信息
Due to the growth of data scale, distributed machine learning has become more important than ever. Some recent work, like TuX(2), show promising prospect in dealing withdistributed machine learning by leveraging the ...
详细信息
ISBN:
(纸本)9781728111414
Due to the growth of data scale, distributed machine learning has become more important than ever. Some recent work, like TuX(2), show promising prospect in dealing withdistributed machine learning by leveraging the power of graph computation, but still leave some key problems unsolved. In this paper, we propose Cymbalo, a new distributed graph processing framework for large-scale machine learning algorithms. To satisfy the specific characteristics of machine learning, Cymbalo employs a heterogeneity-aware data model, a hybrid computing model and a vector-aware programming model, to ensure small memory footprint, good computation efficiency and expressiveness. the experiment results show that Cymbalo outperforms Spark by 2.4x-3.2x, and PowerGraph by up to 5.8x. Moreover, Cymbalo can also outperform Angel, a recent parameter server system, by 1.6x-2.1x.
this paper proposes and evaluates availability models for blockchain provisioning over cloud computing infrastructures as well as their respective deployment expenses in order to establish a cost × benefit relati...
详细信息
ISBN:
(数字)9781728148328
ISBN:
(纸本)9781728148335
this paper proposes and evaluates availability models for blockchain provisioning over cloud computing infrastructures as well as their respective deployment expenses in order to establish a cost × benefit relationship. To demonstrate these models' feasibility, we provide two case studies considering blockchain provisioning over a baseline architecture, and three other alternative redundant environments.
Remote sensing of plant traits and their environment facilitates non-invasive, high-throughput monitoring of the plant's physiological characteristics. Effective ingestion of these sensing data into a storage subs...
详细信息
ISBN:
(纸本)9781728111414
Remote sensing of plant traits and their environment facilitates non-invasive, high-throughput monitoring of the plant's physiological characteristics. Effective ingestion of these sensing data into a storage subsystem while georeferencing phenotyping setups is key to providing timely access to scientists and modelers. In this study, we propose RADIX, a high-throughput distributed data ingestion framework with support for fine-grained georeferencing. Our methodology includes a novel spatial indexing scheme, the nested hash grid, for fine-grained georeferencing of data while conserving memory footprints and ensuring acceptable latency. We include empirical evaluations performed on a commodity machine cluster with up to 1TB of data. Our benchmarks demonstrate the efficacy of our approach.
Alternating direction method of multipliers (ADMM) has been recognized as an efficient approach for solving many large-scale machine learning problems. However, the ADMM under master-slave mode suffers from several li...
详细信息
ISBN:
(纸本)9781728111414
Alternating direction method of multipliers (ADMM) has been recognized as an efficient approach for solving many large-scale machine learning problems. However, the ADMM under master-slave mode suffers from several limitations, e.g., can't make full use of multi-core cluster environment and single master load is too heavy, resulting in huge time overhead. In this paper, we propose a hierarchical communication structure. Since intra-node communications mostly use shared memory, we divide the processes of the same node into one group, the processes within the group synchronize communication, and each group communicate asynchronously withthe master. Combining this structure withthe ADMM algorithm, a hierarchical asynchronous group-based ADMM algorithm (HAG-ADMM) is proposed. theoretical analysis and experiments show that the hierarchical communication structure can improve the communication efficiency of the algorithm and has no effect on the convergence.
Cloud providers rely on virtualization technology for efficient data center management and service delivery. Live Virtual Machine (VM) Migration is one of the critical features of this technology enabling cloud provid...
详细信息
ISBN:
(纸本)9781728111414
Cloud providers rely on virtualization technology for efficient data center management and service delivery. Live Virtual Machine (VM) Migration is one of the critical features of this technology enabling cloud providers to relocate VMs between servers without disrupting applications currently running on it. One of the key challenges of VM live migration is that it creates elephant flows over the network links connecting the source to destination servers due to the transfer of the entire memory and possibly disks of the VM. In this paper, we leverage software -defined networking (SDN) and propose a dynamic flow scheduling approach called Acinonyx. Our solution is designed to minimize the negative impact of live VM migration on data center network traffic and reduce migration time in cloud data centers with multiple network paths between servers. Acinonyx adaptively installs flow entries on the switches of the shortest path withthe lowest congestion and redirects VM migration traffic to the appropriate path. We describe the implementation of the proposed solution over our testbed environment, where we use OpenStack for managing commodity servers and OpenDaylight SDN controller for managing the OpenFlow switches. Our experimental results show that Acinonyx can significantly reduce the migration time of VMs while improving the overall network throughput for other VM communications.
High-performance computing (HPC) systems can consume substantial electricity during their operation, and consequently incur significant energy cost. Demand response is a scheme adopted by energy consumers to reduce en...
详细信息
ISBN:
(纸本)9781728111414
High-performance computing (HPC) systems can consume substantial electricity during their operation, and consequently incur significant energy cost. Demand response is a scheme adopted by energy consumers to reduce energy use during the high load periods upon request from the power grid. In this paper, we propose a demand response model to allow both HPC operators and HPC users to jointly reduce the energy consumption. We apply the contract theory, originated from economics for studying the contractual arrangements among economic actors, and design a reward mechanism to ensure participation of both HPC operators and HPC users in the demand response program. We perform both analytical and simulation studies of our proposed approach. Our analyses show that our contract-based demand response model is both feasible (in ensuring both individual rationality and incentive compatibility) and optimal. the trace-based simulation demonstrates the practicality of the proposed approach.
Information Technology industry has competitiveness on the basis of technological environment. In this environment, the use of cloud services has been increasing to provide high quality services and fast delivery of p...
详细信息
ISBN:
(纸本)9781728111414
Information Technology industry has competitiveness on the basis of technological environment. In this environment, the use of cloud services has been increasing to provide high quality services and fast delivery of products to cloud users. But still some issues are unresolved especially, related to latency between cloud data center and end user. Fog computing is used to support increasing demand of IT service withthe collaboration of cloud computing. It provides computational and storage services of cloud proximate to IoT devices. Fog computing is enhancement of the cloud-based network and computing services. this paper discusses the concept, architecture of fog computing and implemented application. It also highlights about resource provisioning techniques to identify over utilization of fog nodes. Along withthe resource utilization, different scheduling terminologies have also been discussed on various parameters. the motive of this survey is to understand the application of fog computing to improve the existing smart healthcare system.
this paper proposes an algorithm for the distributed server allocation problem, namely Minimizing the Maximum Delay (MMD), where an optimal solution is obtained when all server-server delays are the same constant valu...
详细信息
ISBN:
(数字)9781728148328
ISBN:
(纸本)9781728148335
this paper proposes an algorithm for the distributed server allocation problem, namely Minimizing the Maximum Delay (MMD), where an optimal solution is obtained when all server-server delays are the same constant value. We prove that MMD obtains an optimal solution withthe polynomial time complexity.
Fog computing extends the Cloud computing paradigm to the edge of the network, relying on computing services intelligently distributed to best meet the applications needs such as low communication latency, data cachin...
详细信息
Fog computing extends the Cloud computing paradigm to the edge of the network, relying on computing services intelligently distributed to best meet the applications needs such as low communication latency, data caching or confidentiality reinforcement. While P2P is especially prone to implement Fog computing platforms, it usually lacks important elements such as controlling where the data is stored and who will handle the computing tasks. In this paper we propose both a mapping approach for data-locality and a location-aware scheduling for P2P-based middlewares, improving the data management performance on fog environments. Experimental results comparing the data access performances demonstrate the interest of such techniques. (C) 2018the Authors. Published by Elsevier Ltd.
暂无评论