Towards Security in an Open Systems Federation

This paper argues that security design for Open Distributed Processing (ODP) would benefit from a shift of focus from the infrastructure to individual servers as the owners and enforcers of security policy. It debates the policy nuances, mechanisms, and protocol design consequences, that would follow from such a change of emphasis. In ODP, physically separate systems federate into heterogeneous networks of unlimited scale, so there can be no central authority, nor ubiquitous security infrastructure. Servers that offer, trade, supply and consume services must maintain their own security policies and defend themselves. For servers to take security policy and enforcement decisions, design is concerned with how they might seek advice and guidance from higher authority. This contrasts with an administrator imposed policy on a closed homogeneous network, where an infrastructure enforces administrator declared access rights to potential clients, including rights to delegate rights.