Lightpath provisioning for dynamic traffic is an important issue in WDM optical networks. Meanwhile, in order for a lightpath to survive a network failure, both a primary lightpath and a backup lightpath need to be found for each demand. A demand will be blocked if either cannot be provided. Fast restoration and efficient lightpath establishment are two capabilities sought by service providers. In this thesis, we provide our approaches to address these two issues. Current path-based proactive restoration scheme pre-computes a backup lightpath when a primary lightpath is setup. Better spare capacity utilization can be achieved using backup multiplexing technique; however, long restoration time is inevitable because of the end-to-end signaling required to setup the backup lightpath upon failure. We present a new proactive lightpath restoration method that computes one primary lightpath and two backup lightpath segments for each traffic demand. Simulation results showed that the new method could significantly reduce the restoration time with only minor increase in capacity requirement. We also propose an efficient establishment scheme of restorable lightpaths. Three heuristic ideas are proposed to exploit the wavelength usage information and to make link channels across the network used more evenly, leading to lower blocking probability. Simulation shows that, the load balancing routing algorithm (LBA), with our heuristic cost functions for primary path selection and backup path selection, achieves comparable performance as the centralized algorithm does. However, LBA asks for much less information to be disseminated, and therefore is more scalable.