DocumentCode
3696586
Title
Elastic virtual network function placement
Author
Milad Ghaznavi;Aimal Khan;Nashid Shahriar;Khalid Alsubhi;Reaz Ahmed;Raouf Boutaba
Author_Institution
David R. Cheriton School of Computer Science, University of Waterloo, ON, Canada
fYear
2015
Firstpage
255
Lastpage
260
Abstract
Nowadays, many cloud providers offer Virtual Network Function (VNF) services that are dynamically scaled according to the workload. Enterprises enjoy these services by only paying for the actual consumed resources. From a cloud provider´s standpoint, the cost of these services must be kept as low as possible, while QoS is maintained and service downtime is minimized. In this paper, we introduce Elastic Virtual Network Function Placement (EVNFP) problem and present a model for minimizing operational costs in providing VNF services. In this model, the elasticity overhead and the trade-off between bandwidth and host resource consumption are considered together, while the previous works ignored this perspective of the problem. We propose a solution called Simple Lazy Facility Location (SLFL) that optimizes the placement of VNF instances in response to on-demand workload. Our experiments suggest that SLFL can accept two times more workload while incurring similar operational cost compared to first-fit and random placements.
Keywords
"Bandwidth","Elasticity","Cloud computing","Conferences","Manganese","Entropy","Engines"
Publisher
ieee
Conference_Titel
Cloud Networking (CloudNet), 2015 IEEE 4th International Conference on
Type
conf
DOI
10.1109/CloudNet.2015.7335318
Filename
7335318
Link To Document