Towards an Efficient Policy Evaluation Process in Multi-Tenancy Cloud Environments

Abstract

Cloud computing offers most of its services under multi-tenancy environments. To satisfy security requirements among collaborating tenants, each tenant may define a set of access control policies to secure access to shared data. Several cloud solutions make use of XACML to specify such policies. However, existing implementations of XACML perform a brute force search to compare a request to all existing rules in a given XACML policy. This decreases the decision process (i.e., policy evaluation) performance especially for policies with a large number of rules. In this paper, we propose an automata-based approach for an efficient XACML policy evaluation. We implemented our approach in a cloud policy engine called X2Automata. The engine first converts both XACML policies and access requests to automata. Second, it combines the two automata by a synchronous product. Third, it applies an evaluation procedure to the resulting automaton to decide whether an access request is granted or not. To highlight the efficiency of X2Automata, we compare its performance, based on the OpenStack cloud environment, with the XACML implementation named Balana.