Advances in device technology and dense wavelength division multiplexing have enabled modern optical backbones to provide high data rate services in a cost-effective manner. At the same time, applications are emerging with the potential to leverage these capabilities for short time spans: load migration for Internet clouds, integrated computational grids, and remote surgery. Until such applications mature, investment in real networks to support them is unattractive, but without such investment, the applications mature slowly. Current networks thus remain dominated by lightpaths provisioned by hand, over the space of days or weeks. Three problems must be solved to enable a rapid transition to dynamic optical networks that integrate well with demands at the Internet Protocol (IP) level: traffic characterization, resource dimensioning, and resource allocation. Traffic characterization provides models of traffic demand variation and evolution, which can only be predicted at this stage. In this paper, we provide an overview of our past and current work on the resource dimensioning and allocation subproblems. We describe our methodology for dimensioning networks based on expected traffic behavior. We next offer a few observations about dealing with statistical variations in network behavior. We develop analytic models of opportunity cost and a congestion-aware routing algorithm based on the models. A near-optimal threshold value for this algorithm can be chosen based on analysis, and simulations demonstrate that this choice is robust to variations in both topology and capacity.