Scheduling multiple scientific workflows using containers on IaaS cloud
- PDF / 2,149,571 Bytes
- 16 Pages / 595.276 x 790.866 pts Page_size
- 20 Downloads / 217 Views
ORIGINAL RESEARCH
Scheduling multiple scientific workflows using containers on IaaS cloud P. Rajasekar1 · Yogesh Palanichamy1 Received: 4 May 2020 / Accepted: 14 August 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract With the adaptable paradigm of cloud computing and obtainable of data accumulated from largely high-powered scientific devices, workflows have turn into an occurring aim to execute considerable scientific advances at an enhanced speed. Occurring Workflow as a Service (WaaS) frameworks provide scientists an effortless, simply accessible and cost-efficient manner of using their applications from anywhere and at anytime in the cloud. They are multitenant platforms and are developed to handle the execution of heterogeneous workflows continuous workload. To fulfill this, they utilize the compute, network and storage services provided by Infrastructure as a Service (IaaS) vendors. Therefore, at any considerable particular moment, a WaaS framework should be proficient of effectively schedule these continuous workload of workflows with various features and quality of service (QoS). Therefore, we propose a strategy of scheduling and resource provisioning planned particularly for WaaS platforms. The algorithm is dynamic and scalable to adjust to improve in the workload and platform. It supports containers to deal the inefficiency of resource utilization and targets to reduce the overall execution cost of infrastructure resources as fulfilling each single workflow deadline constraint. To our information, this approach that explicitly deals VM sharing in the subject of WaaS by devising the utilization of containers in the heuristics of scheduling and resource provisioning. Our experimental results shows its responsiveness to the uncertainties of the environment, its potential to achieve deadlines, and its cost-effectiveness when compared to other recent algorithms. Keywords Cloud computing · Workflow as a service · Scheduling · Resource provisioning · Deadline and cost minimization
1 Introduction A group of computational tasks with dependencies are described as workflows. Workflows are a familiar application model is used in the field of computational science. They allow the examination of data in an organized and distributed way and have been used successfully to produce extraordinary scientific improvement in different specialization such as physics, biology, astronomy and medicine (Gil et al. 2007). Their significance is attracted in current big data span as they provide a productive way of execution and * P. Rajasekar [email protected] Yogesh Palanichamy [email protected] 1
Department of Information Science and Technology, College of Engineering, Anna University, Guindy, Chennai, Tamilnadu, India
knowledge extraction from the data given by more powerful equipment’s such as gravitational wave detectors, telescopes and particle accelerators. Therefore, it is usual for workflows to be a wide-range data and resource intensive model that are used on distributed platforms wi
Data Loading...
