While the domain, shared resources, and access concepts provide the ability to focus on where security properties are enforced in a CDS, these concepts by themselves do not provide sufficient ability to construct complex architectures. Domain decomposition allows a domain to be refined into sub-domains, shared resources, and accesses between them. These entities are called children of the domain.
All processes in a domain have essentially unlimited access to objects in that domain. With decomposition, access is refined within a domain by dividing the parent domain into sub-domains and shared resources in order to explicitly define the accesses that are allowed within the domain. Sub-domains obey the same rules as domains and may only interact via shared resources.
For example, consider a CDS security architecture with three domains: the low information domain, the guard application domain, and the high information domain. Using decomposition, the guard domain can be populated with several sub-domains representing individual guard stages or applications, each of which has a subset of the security permissions of the parent guard domain. This process allows for stepwise introduction of sub-domains, while keeping the focus on the security goals and least privilege, thereby increasing the assurance of the CDS security architecture.
Decomposed domains result in some overall security architecture constraints:
Decomposition may be successively applied to subdomains to provide the level of granularity needed in the security architecture. With decomposition, the CDS Framework allows a developer to produce a detailed and precise security policy, while maintaining the security goals of the architecture and preserving the high-level simplicity of the original design.
©2005 - 2008 Tresys Technology, LLC
Patent Pending