Distributed dynamic load balancing for pipelined computations on heterogeneous systems

Riakiotakis, I.; Ciorba, Florina M.; Andronikos, Theodore; Παπακωνσταντίνου, Γ.

doi:10.1016/j.parco.2011.01.003

Cited by 9 publications

(8 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Depending on the scheduling policy and the computational environment and application requirement, various shortcomings are found: (1) Efficiency may not scale as the number of processors increases and (2) High probability that the scheduling is not done by using updated information. In an effort to address these limitations, a distributed load balancing scheme is presented by Riakiotakis [7], in which the scheduling decisions are made by the workers in a distributed fashion and implemented this method along with other two master-worker schemes for three different scientific computational kernels.…”

Section: Dynamic Load Balancing Algorithmmentioning

confidence: 99%

Load Balancing Approaches for Web Servers: A Survey of Recent Trends

Shukla

Kumar

Singh

2018

IJE

View full text Add to dashboard Cite

Numerous works have been done for load balancing of web servers in grid environment. Reason for popularity of grid environment is to allow accessing distributed resources which are located at remote locations. For effective utilization, load must be balanced with all resources. Importance of load balancing is discussed by distinguishing the system between without load balancing and with load balancing. Various performance metrics that needed to be considered for designing an efficient load balancing algorithm are also described. Intensive review of literature of different load balancing approaches for web servers had been carried out and presented in this paper.

show abstract

Section: Dynamic Load Balancing Algorithmmentioning

confidence: 99%

Load Balancing Approaches for Web Servers: A Survey of Recent Trends

Shukla

Kumar

Singh

2018

IJE

View full text Add to dashboard Cite

show abstract

“…In distributed approaches e.g. [36][37][38][39], the load balancing process can be executed by all nodes in the system. In addition, in this approach all nodes can communicate with each other for achieving a global goal in the system which is called cooperative or every node can work independently for just achieving a local goal that is noncooperative form.…”

Section: Load Balancing Literaturementioning

confidence: 99%

Load Balancing in Cloud Computing: A State of the Art Survey

Mesbahi

Rahmani

2016

IJMECS

View full text Add to dashboard Cite

Abstract-Cloud computing has proposed a new perspective for provisioning the large-scale computing resources by using virtualization technology and a payper-use cost model. Load balancing is taken into account as a vital part for parallel and distributed systems. It helps cloud computing systems by improving the general performance, better computing resources utilization, energy consumption management, enhancing the cloud services' QoS, avoiding SLA violation and maintaining system stability through distribution, controlling and managing the system workloads. In this paper we study the necessary requirements and considerations for designing and implementing a suitable load balancer for cloud environments. In addition we represent a complete survey of current proposed cloud load balancing solutions which according to our classification, they can be classified into three categories: General Algorithm-based, Architectural-based and Artificial Intelligence-based load balancing mechanisms. Finally, we propose our evaluation of these solutions based on suitable metrics and discuss their pros and cons.

show abstract

“…Let T i be the wall‐clock time from the beginning of the computation until the i th processing node finishes its computation. So, the data partitioning problem in heterogeneous system can be expressed as an optimization problem to find the optimal partitioning as follows:

\begin{matrix} Minimize max {T_{1}, T_{2}, ⋯, T_{i}, ⋯, T_{P}}, 1 & \leq i \leq P \\ T_{i} & = T_{com p_{i}} + T_{com m_{i}} . \end{matrix}

The uneven distribution of the workload between processing nodes could degrade the performance of scientific applications on distributed systems, especially on heterogeneous systems . So, to balance the computation time,

T_{italiccom p_{i}}

, we should distribute the workload between processing nodes of the heterogeneous system according to their computational powers.…”

Section: The 3ddp: 3d Data Partitioning Algorithmmentioning

confidence: 99%

“…The uneven distribution of the workload between processing nodes could degrade the performance of scientific applications on distributed systems, especially on heterogeneous systems. 42 When one node would do all the work, the communication volume equal to 0 while it would result in a fully imbalanced system. In contrast, a good load balance can cause the amount of extra communication overhead to grow.…”

Section: Problem Statementmentioning

confidence: 99%

3‐D data partitioning for 3‐level perfectly nested loops on heterogeneous distributed systems

Zefreh

Lotfi

Khanli

et al. 2016

Concurrency and Computation

View full text Add to dashboard Cite

SummaryNested loops are the largest source of parallelism in many data-parallel scientific applications.Heterogeneous distributed systems are popular computing platforms for data-parallel applications. Data partitioning is critical in exploiting the computational power of such systems, and existing data partitioning algorithms try to maximize performance of data-parallel applications by finding a data distribution that balances the workload between the processing nodes while minimizing communication costs. This paper addresses the problem of 3-dimensional data partitioning for 3-level perfectly nested loops on heterogeneous distributed systems. The primary aim is to minimize the execution time by improving the load balancing and minimizing the internode communications. We propose a new data partitioning algorithm using dynamic programming, build a theoretical model to estimate the execution time of each partition, and select a partition with minimum execution time as a near-optimal solution. We demonstrate the effectiveness of the new algorithm for 2 data-parallel scientific applications on heterogeneous distributed systems. The new algorithm reduces the execution time by between 7% and 17%, on average, compared with leading data partitioning methods on 3 heterogeneous distributed systems. architecture and the need to meet ever-increasing computing needs of scientific applications, the computational capacity of the clusters is often increased. 8,9 As a first approach, we could replace all processing nodes with newer, faster ones. In this case, the cluster remains homogeneous over time, but a complete replacement of all nodes can be very costly. As a second approach, we could upgrade the cluster by adding more processing nodes that use a newer technology with higher speed. Also, we could aggregate several clusters together to use their computational power for solving computing problems. 8 Another approach is adding graphics processing units (GPUs) to improve performance of existing nodes. 9 In the latter 3 cases, the cluster becomes heterogeneous. [10][11][12] In this way, heterogeneous computing systems have emerged as an important contribution to provide computational capacity in high-performance computing. In fact, the prevalence of heterogeneous systems in the TOP500 list grew from 3.4% to 18.0%

show abstract

Distributed dynamic load balancing for pipelined computations on heterogeneous systems

Cited by 9 publications

References 24 publications

Load Balancing Approaches for Web Servers: A Survey of Recent Trends

Load Balancing Approaches for Web Servers: A Survey of Recent Trends

Load Balancing in Cloud Computing: A State of the Art Survey

3‐D data partitioning for 3‐level perfectly nested loops on heterogeneous distributed systems

Contact Info

Product

Resources

About