Demand for computational grid is growing ever since the rise of social networks, Artificial Intelligence based systems (use of machine learning), scientific application, etc. Service providers are aiming towards achieving maximum grid utilization so that they can serve more customers efficiently. The key enablers to achieve optimal grid utilization and better turnaround time is by efficient scheduling of tasks on computational grids. However, designing efficient grid scheduling algorithms is still a challenge due to its complexity (NP-complete). Hence, several near optimal approximation algorithms are designed based on plethora of techniques such as heuristics, bio inspired, genetic, greedy approaches, etc. Thus there is a scope for further improvement in scheduling algorithm to achieve early task completion and better grid utilization for precedence constrained tasks. Moreover, nowadays with advancement in computing hardware there is a preference for parallel task execution strategy rather than sequential task execution. With this there is also a notion of partial dependency between parallel tasks. Thus, there is a need to revisit designing grid scheduling algorithm. Hence, In this paper we propose a novel grid scheduling algorithm for interdependent parallel tasks with varying dependencies (full, partial, no) on computational grid using a greedy approach. Further the correctness and performance of the proposed scheduling algorithm is evaluated by comparing it with proposed brute force scheduling algorithm.

A Novel Task Scheduling Scheme for Computational Grids - Greedy Approach

In high performance computing environment such as grids and clouds, resource availability is highly dynamic, unpredictable and unreliable, due to various factors such as system load, system failure, available network bandwidth, network failure, introduction of new resources, contention among remote task and the local tasks of resources and so on. In such environments, the dynamic availability of resources can affect execution of parallel applications. Parallel applications modelled by Directed Acyclic Graphs (DAG) scheduled on a network of heterogeneous processors so as to minimize finish time is a known NP complete problem. The Heterogeneous Earliest Finish Time (HEFT) is a static task scheduling algorithm that assumes probability of resource availability is 100% whereas in real life situations the case is not true. The temporal availability of resources is to be considered in task scheduling decisions. We propose a new Probabilistic Availability based Task Scheduling Algorithm (PATSA) where tasks are scheduled not only based on earliest finish time of task but also on resource availability probability so as to minimize makespan. Our algorithm was compared with existing HEFT algorithm for a certain set of random DAGs based on performance metrics such as Makespan, Schedule Length Ratio and Speedup.

Probabilistic Availability based Task Scheduling Algorithm

Many scheduling algorithms are designed based on directed a cyclic graphs. However, intra-iteration data dependencies among tasks are widespread and have an adverse effect on regulation of task execution order so that the advantage of multi-core is impeded. Therefore, this paper addresses the problem of setting appropriate delays on each edge of a task graph representing an application to eliminate intra-iteration data dependencies. Based on the principle of the retiming technique, we propose an algorithm to determine the appropriate retiming value of each node according to the number of nodes along the path with the maximum number of nodes from it to sink node. Scheduling length can be reduced by the above transformation due to eliminating intra-iteration data dependencies. We experiment with random task graphs on 2 and 4 processor cores, respectively. The experimental results show that the proposed algorithm can effectively improve task parallelization for applications with dependent tasks.

An Algorithm Used to Improve Task Parallelization for Directed Acyclic Graphs

Scientific and business application workflows modeled as directed acyclic graphs represent the complex computational problems, which are to be executed in distributed computing environments. Due to sharing, elasticity, complexity, heterogeneity, virtualization, and workload variations, the cloud computing systems offer a host of challenges in terms of performance and reliability. While running a complex workflow application in the IaaS cloud, failures may be triggered by its resource failures or workflow level failures, which lead to large-scale service disruption with adverse consequences. In some static task scheduling algorithms like earliest completion time scheduling (ECTS) and heterogeneous earliest finish time (HEFT), since resource availability and the probability of task completion are not considered, workflow execution delay increases makespan. The proposed compensatory aggregation based failure aware (CAFA) cloud workflow scheduling algorithm considers resource selection based on the availability influenced by CPU load and processor speed for good performance and the probability of task completion for reliability. A method of compensatory aggregation of criteria is used for scoring each resource and choosing the resource with the highest score. The performance of CAFA algorithm is much better in terms of reduced makespan compared to HEFT and ECTS algorithms and is also found to be more reliable.

Compensatory Aggregation based Failure Aware Cloud Workflow Scheduling

Cloud Computing provides access to a shared pool of computing resources such as servers, storage, computer networks and services, which can be rapidly provisioned and released, for the execution of various scientific and business applications. Scheduling scientific workflows modeled by Directed Acyclic Graphs is an NP complete problem. In cloud environment, there are fluctuations in resource availability due to shared resources and vastly varying workloads. The performance variations in virtual machines, have an impact on task execution times and data transfer times. This has a great impact on the performance of scheduling algorithms. Schedulers should map tasks to cloud resources considering the effective utilization of underlying resources. We propose a new static workflow scheduling algorithm called Predictive Probabilistic Resource Availability based Cloud Workflow Scheduling (PPRA) with the objective of minimizing makespan considering the probability of resource availability of cloud resources. This algorithm is compared with existing algorithms which assume full resource availability while making scheduling decisions. The proposed algorithm is found to be more reliable and performing better than existing HEFT and ECTS algorithms, in terms of minimized makespan.

/pdf/predictive-probabilistic-resource-availability-based-cloud-4v3ulv7ty9.pdf

Predictive Probabilistic Resource Availability based Cloud Workflow Scheduling (PPRA)

Energy minimization for multi-core embedded systems is one of the most important issues, as well as performance When the transition time of supply voltage is a constant, appropriate execution order of tasks can decrease the number of voltage transition and save the energy overhead of voltage transition This paper proposes a scheduling algorithm for multi-core embedded systems running periodic dependent tasks with hard time constraints, considering transition overhead as well as communication overhead Based on the retimed task graphs produced by the RDAG algorithm, we execute tasks mapping and reorder the execution order of tasks by descending voltage levels We conduct experiments on a set of random task sets The results show that the proposed algorithm achieves substantial energy savings compared with previous work

Energy minimization for embedded systems with discrete voltage levels

Data dependencies among tasks are inevitable and have an adverse effect on the parallelism of an application Therefore, this paper addresses the problem of setting appropriate delays on each edge of a task graph representing an application to eliminate intra-iteration data dependencies so as to produce a prologue and epilogue as small as possible Based on the principle of the retiming technique, we propose an algorithm called NNP (number of nodes along a path) to determine the minimum retiming value of each node according to the number of nodes along the path with the maximum number of nodes from its immediate successor nodes to leaf nodes Computation time can be decreased by the above transformation due to eliminating intra-iteration data dependencies We experiment with random task graphs on 2, 3 and 4 processor cores, respectively The experimental results show that the algorithm can obtain substantial scheduling length reduction

An algorithm used to eliminate intra-iteration data dependencies

Recently, with the constant development of application of high-dimensional data query, the index-based nearest neighbor retrieval has greatly improved the efficiency and accuracy of retrieval. The approximate nearest neighbor query is proposed and applied to get high performance to ease the curse of dimensionality to a certain extent. However, the heuristic pruning method may prune many local optimal solutions so that it is impossible to make sure to what extent the final results meet users’ needs. With the nearest neighbor as input, this paper proposes a method to get extent how the input meets user’s need based on fuzzy control, and then determines whether or not the input will be an approximate optimal solution with a comparison between the extent and threshold value. At the same time, the genetic algorithm is used to complete the adaptive study process of the threshold value to reflect users’ approximate needs.

High-Dimensional Approximate Nearest Neighbor Query Based on Fuzzy Control and Genetic Algorithm

In order to meet the testing requirements of task scheduling algorithms, this paper proposes two algorithms for generating random adjacency matrixes. One algorithm is used for generating random adjacency matrixes representing either directed cyclic graphs or directed acyclic graphs. The other algorithm is used for generating directed acyclic graphs only. The paper analyses the characters of elements of an adjacency matrix, and applies a square matrix with elements falling into a Gaussian distribution and the technique of descending order to achieve a random adjacency matrix. We conduct experiments on our algorithms using MATLAB. The experimental results show that the algorithms have high efficiencies, and adjacency matrixes which represent task graphs with dozens of nodes and dozens of edges can be generated in one minute. Those random adjacency matrixes generated by the algorithms are suited for task scheduling algorithms which use random adjacency matrixes to verify validities.

Algorithms for Random Adjacency Matrixes Generation Used for Scheduling Algorithms Test

Handle everyday research tasks with reliable, citation-backed results

Your personal Research Agent to handle research tasks with citation-backed results

Popular Tasks used by Researchers

How can I help with your research?

Meet SciSpace

Get more enhanced response by uploading the PDFs you want me to reference.

No relevant PDFs in your library

SciSpace is the AI research assistant for academics. Run systematic literature reviews on 280M+ papers, and write papers with cited sources. Trusted by 1M+ students, PhDs & researchers.

SciSpace | AI for Research

Analyze PDFs

Code & Manuscripts

Funding & Grants

Literature & Patents

Medical & Clinical Data

Systematic Review

Visualize & Present

Web & Data

Build a Google Scholar-like website for your research.

Build a website

Create charts and images for your research

Create a Chart

Write a paper for submission to a journal

Draft a manuscript

Patent Search

Design eye-catching scientific posters in minutes.

Scientific Poster Generation

Systematic Literature Review

One task is running at the moment. Your messages will be shown right after.

Drag and drop or click here to browse

Loved by <highlight>1 million+</highlight> researchers

Extract a list of specific topics and their sources from unstructured text

Topics

Compare and analyze relevant papers that matches with your search

Papers

Get insights from PDFs and bookmarked papers from your library

My library

Recent searches

Try searching for:

Catch AI-generated content in scholarly and non-scholarly content

{ai} Detector

Ai Writer

Get PDF Summaries, highlighted text explanations 

Chat with PDF

Effortlessly create in-text citations and bibliographies in APA and 2,500 other formats

Citation generator

Get explanations, summaries, and answers on academic papers

Ease up your research workflow with {scispace}'s cohort of exciting AI tools

Elevate your academic writing skills and convey your ideas the way you want

Paraphraser

Explore our range of reading and writing tools

Your file is being prepared and should be ready in a few minutes. If it's a large file, it might take a bit longer. You can close this window, and we'll email you the file when it's done.

You have reached a maximum limit of <strong>{limit}</strong> columns in the table. Remove at least <strong>1</strong> column to add or create another one.

Wei Yan

Author Tools

Chat about Author

Papers

Energy minimization for embedded systems with discrete voltage levels

An algorithm used to eliminate intra-iteration data dependencies

High-Dimensional Approximate Nearest Neighbor Query Based on Fuzzy Control and Genetic Algorithm

Algorithms for Random Adjacency Matrixes Generation Used for Scheduling Algorithms Test