VMatch: A Matching Theory Based VDC Reconfiguration Strategy
Published:
Authors
Anurag Satpathy, Manmath Narayan Sahoo, Lucky Behera, Chittaranjan Swain, Ashutosh Mishra.
Conference
2020 IEEE 13th International Conference on Cloud Computing (CLOUD) (Core Rank - B), Acceptance Rate - 17%
Received the YESC Student Paper Participation Award Link
Abstract
A virtual data center (VDC) mostly encapsulates multiple virtual machines (VMs) with communication dependencies. These VDC requests are dynamic in nature and often experience fluctuating demands across different resources. In this paper, we propose a dynamic resource reconfiguration strategy called VMatch that generates an efficient relocation/remapping plan for already assigned virtual links (VLs) facing bandwidth expansion. The overall problem is formulated as a one-to-one matching game that aims to minimize the relocation cost from the users perspective and at the same time improves resource utilization from a service providers (SPs) perspective. By using the concept of preferences in the matching game, different stakeholders, i.e., end-users and SPs express their priorities. Thorough simulation analysis of the proposed approach shows that the model on an average can reduce the remapping cost by 19% and improve server utilization by 21% in comparison with the baselines.