1 National Science Foundation Secure and Trustworthy Cyberspace (SaTC) Social, Behavioral, and Economic (SBE) Science Research High Assurance Cloud: Maintaining Trust Barbara Endicott-Popovsky, Director Center for Information Assurance and Cyber Security University of Washington, Seattle, WA 98195 Jan Whittington, Assistant Professor Urban Design and Planning Department College of Built Environments University of Washington, Seattle, WA 98195 Sam Chung, Associate Professor Computing & Software Systems Institute of Technology University of Washington, Tacoma, WA 98402 The advent of cloud computing allows many organizations to significantly reduce capital expense for IT infrastructure. The obvious financial advantage from cloud computing of trading in capital costs for operational costs has lured many small or startup businesses and non-profit organizations to the cloud1. We can reasonably expect the top cloud computing vendors to lock their doors and secure their infrastructure but customers regularly leave their application doors unguarded, and the combined challenges of security, privacy, and trust in computing intersect with legal and economic concerns that bring in many other forms of organizations with many different interests. High information assurance in the cloud is a significant and growing concern. Mike Howard, Principal Security Program Manager at Microsoft, notes that due to ever evolving code and the constant vigilance of hackers, “security is a never ending battle” and stresses the importance of providing ongoing training in secure programming.2 Kirk Bailey, a founding leader of a regional association of information systems security professionals called Agora, mentioned during a PBS interview that technology cannot be secured, we should instead think of it as “risk-managed.”3 The development of high assurance clouds – cloud services with high levels of assurance that information will be secure, often associated with the extreme demands of defense, health, finance, and other critical functions [cites] – will only be possible when vulnerabilities within and across technologies, service providers, and users are addressed in a systematic way. A systematic 1 2 3 Hofman, P., Woods, D., Cloud Computing: The Limits of Public Clouds for Business Applications, IEEE Internet Computing, Nov./Dec. 2010, p. 90-93 Howard, Michael. “Lessons Learned from Five Years of Building More Secure Software”. MSDN Magazine. November 2007. http://msdn.microsoft.com/en-us/magazine/cc163310.aspx Bailey, Kirk, PBS Interview, http://www.pbs.org/wgbh/pages/frontline/shows/hackers/interviews/bailey.html 2 understanding recognizes interdependencies between elements in the system and establishes mechanisms of feedback that identify vulnerabilities, assess the risks that may accrue from vulnerabilities, and bring about effective action in response. Effective actions constitute safeguards against existing and future risks, effective when they reduce the incidence and scope of vulnerabilities. As the weakest links in the system are shored up, every user benefits from higher levels of information assurance. These aims are met when the system meets the management or business needs of the most extremely sensitive forms of information. [cites] Approaching the Problem from a Systems Perspective Figure 1 presents a generalized schematic overview of cloud systems. The use of the cloud is contingent upon the flow of data between service providers and users of those services, which exist in an ecosystem of firms, organizations, and individuals producing and providing services to users of their applications. This collection of users increasingly includes the organizations that operate the other forms of critical infrastructures, such as health, energy, transport, water, food, and other means of communication. Cloud services, like the critical infrastructure systems on which all of our economies and societies depend, are a global phenomenon, leading and following users across the jurisdictional boundaries of nation-states, and thus through the full range of variation in the contexts that are possible for understanding, using, and protecting information. The demands placed on this system from extreme security needs are evident in the features that define the interdependencies of the elements with each other, yet form the most striking contrasts in terms of technologies and models of operations. Figure 1: High Assurance Clouds: Demands of the System 3 Some demands of the system are part of the regular business activities of cloud providers, as members in the CSA, for instance, or as service providers in the marketplace. Jurisdictional Boundaries: With a business model dependent on operations across jurisdictional boundaries, cloud services depend on the formation of standards and effective enforcement for compliance with those standards. Cloud systems demand standards and compliance because this is an efficient way to provide a basis for users – the people and organizations who comprise the market for cloud services – to determine the reputation of providers, within the social and economic contexts set, in part, by nation-states. Data Flows: Similarly, though the flow of data must pass obvious thresholds for performance in service level agreements, the willingness of parties such as users to engage in voluntary agreements for services from cloud providers depends on the mechanisms of governance for that exchange and the ability of either party to audit at will – thus revealing the provenance of data and giving parties the means to separate trustworthy from questionable data at its origin. Another set of demands are raised by the customers or users of the cloud. Cloud Users with App Services: Cloud users, when in the business of developing or providing services through software applications, create substantial vulnerabilities for the multi-tenancy architectures of cloud service providers if the applications they develop do not embed security into their source code. Attacks to or through applications may be destined for either neighboring cloud tenants or fellow application users. Until applications contain engines acting in real time, with accurate and meaningful responses to attacks, this vulnerability brings risks for all parties to the system. Users of applications, whether they are individuals or organizations, increase the demand for security through the drive toward increasingly convenient – and widely varied – devices for generating, accessing, transmitting, and transforming data. App Users: At the same time, the identity of each user is perhaps the smallest, yet single-most important unit of analysis for determining whether or not any given cloud sits in a trustworthy computing system. Despite the dramatically public nature of massive amounts of personal information, people and organizations desire and deserve the protection of this information from misuse and abuse. To disregard this fundamental component of the market for cloud services is to risk losing the trust of the sources of finance that keep the entire system afloat. Critical Infrastructures; And last but not least, those responsible for operating the systems of infrastructure that made modern industry possible, such as our health, energy, transport, water, food, and communication systems, are rapidly deploying technologies up to the cloud with questionable influence on the efficiency and security of these legacy systems. Each infrastructure has its own deep structures of policies, business, operations, maintenance, and capital, some with technologies that have remained in place for over a century. Trustworthy cloud computing for critical infrastructures is entirely dependent upon the context created by these aged systems. 4