Research Area:  Cloud Computing
Cloud computing infrastructure has in recent times gained significant popularity for addressing the ever growing processing, storage and network requirements of scientific applications. In public cloud infrastructure predicting bandwidth availability on intra cloud network links play a pivotal role in efficiently scheduling and executing large scale data intensive workflows requiring vast amounts of network bandwidth. However, the majority of existing research focuses solely on scheduling approaches which reduce cost and makespan without considering the impact of bandwidth variability and network delays on execution performance. This work presents a time series network-aware scheduling approach to predict network conditions over time in order to improve performance by avoiding data transfers at network congested times for a more efficient execution.
Keywords:  
Author(s) Name:   Rachael Shaw,Enda Howley,Enda Barrett
Journal name:  
Conferrence name:  International Conference on Service-Oriented Computing
Publisher name:  Springer, Cham
DOI:  10.1007/978-3-319-69035-3_15
Volume Information:  pp 221-228
Paper Link:   https://link.springer.com/chapter/10.1007%2F978-3-319-69035-3_15