Auto-scaling Compute and Network Resources in a Data-center.

This paper focusses on auto-scaling network and compute resources using an intermediary enterprise that uses resources from a public cloud to provide a virtual private cloud to a single client enterprise. The primary goal of the auto-scaling technique is to achieve a profit for the intermediary enterprise while satisfying service level agreements associated with client requests as well as maintaining the desired grade of service for the client enterprise. A secondary goal is to minimize the east-west traffic in the Data Center by reducing the number of network links utilized. Network bandwidth and compute resources are reserved to service the requests from a broker. This paper presents an auto-scaling algorithm and includes a discussion of system design. A simulation-based performance analysis is presented to demonstrate the effectiveness of the proposed technique.