Policity: An experimental evaluation of policy-based administration in a city simulation

Abstract

The administration cost of today's complex IT systems is a major inhibitor of growth for many companies in the burgeoning services economy. One proposed solution is policy-based system administration - a type of autonomic computing system that can manage itself with minimum human intervention. To explore the cost-benefit tradeoffs of policy-based system administration, we conducted an experiment in which participants managed a simulated city (Policity) using either a policy-based interface or a manual GUI-based interface. Results showed that for novices, performance was 50% higher using the policy interface compared to using the manual interface, but the interaction workload was significantly higher with the policy interface than with the manual interface. Overall, our results suggest that policy-based system administration can provide substantial productivity improvement at the cost of some additional interaction workload. © 2005 IEEE.