Risk-Aware Web Service Allocation in the Cloud Using Portfolio Theory

Abstract

In this paper, we view the cloud as market place for trading instances of web services, which can be bought or leased by web applications. Applications can buy diversity by selecting web services from multiple cloud sellers in a cloud-based market. We argue that by diversifying the selection, we can improve the dependability of the application and reduce risks associated with service level agreements violations. We propose a novel dynamic adaptive search based software engineering approach, which uses portfolio theory to construct a diversify portfolio of web service instances, traded from multiple cloud providers. The approach systematically evaluates the Quality of Service and risks of the portfolio, compare it to the optimal traded portfolio at a given time. It can then dynamically decide on a new portfolio and adapt the application accordingly. We use a hypothetical scenario to demonstrate the effective use of the portfolio-based optimization.