Dynamic Resource Allocation Through Workload Prediction for Energy Efficient Computing

• Published in: Advances in Computational Intelligence Systems Vol. 513; pp. 35 - 44
• Main Authors: Ahmed, Adeel, Brown, David J., Gegov, Alexander
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2016; Springer International Publishing
• Series: Advances in Intelligent Systems and Computing
• Subjects: Artificial intelligence; Automatic control engineering; Dynamic resource allocation; Energy efficient computing; Image processing; NARX neural network; Workload prediction
• ISBN: 3319465619; 9783319465616
• ISSN: 2194-5357; 2194-5365
• DOI: 10.1007/978-3-319-46562-3_3

Rapid and continuous increase in online information exchange and data based services has led to an increase in enterprise data centres. Energy efficient computing is key to a cost effective operation for all such enterprise IT systems. In this paper we propose dynamic resource allocation in server based IT systems through workload prediction for energy efficient computing. We use CPU core as a dynamic resource that can be allocated and deallocated based on predicted workload. We use online workload prediction as opposed to offline statistical analysis of workload characteristics. We use online learning and workload prediction using neural network for online dynamic resource allocation for energy efficient computing. We also analyse the effect of dynamic resource allocation on clients by measuring the request response time to clients for variable number of cores in operation. We show that dynamic resource allocation through workload prediction in server based IT systems can provide a cost effective, energy efficient and reliable operation without effecting quality of experience for clients.