Mitigating Cloud Computing security risks using a self-monitoring defensive scheme

Cloud Computing (CC) promises to offer seamless provisioning of on-demand services and data through the network. These services are made discoverable to the user in the form of simple abstractions via virtualized resources. These services offer unprecedented dynamic, primarily hardware dependent, scalability. Although CC promises to make life much easier for the user, it comes with significant security issues. Because on-demand service provisioning for applications and data will be used by hundreds of thousands (if not millions) of users simultaneously, a successful intrusion would not only expose sensitive data, but it could also completely cut users off from both applications and/or data. This paper examines the underlying security risks inherent to the CC paradigm, compares approaches to mitigate known security risks, and offers a solution that leverages intelligent multi-agent systems and network data ontologies to provide automated defense for both known and unknown malware security risks. We describe a mechanism whereby a dynamic ontology can be self-enriched over time to provide for some protection against unknown security risks.

I. Introduction

Cloud Computing (CC) has been described a number of ways: network virtualization, the re-birth of application service providers (ASP), or a lot of hype for a rebirth of the services that mainframe computers have already been providing for many years. For the purposes of this paper, we will use CC as described in the recently released Draft Special Publication 800–146, “DRAFT Cloud Computing Synopsis and Recommendations”[1].