Two-layered IP-over-wavelength-division-multi plexing networks supporting IP and wavelength services are becoming increasingly popular. IP services are traditional data services such as virtual private networks, web browsing, etc., whereas wavelength services are new bandwidth-intensive services with strict quality-of-service (QoS) requirements, such as telemedicine, remote visualization, etc. A generic wavelength service (circuit) may require protection from a dedicated backup circuit to meet strict QoS requirements. With the upcoming huge transmission speeds of 100 Gbps and beyond, these dedicated backup circuits will induce significant underutilization of network capacity and high operational expenditure, which can be minimized by supporting IP traffic over idle backup circuits. However, during failure of backup circuits, the IP topology must always remain connected, and this problem has been preliminarily addressed by means of a suboptimal two-step approach [Opt. Switching Networking, vol. 7, p. 196, 2010]. In this paper, we discuss the opportunities and the challenges of a one-step integrated design, and we propose new methodologies for the integrated design of wavelength and IP services, with backup capacity sharing and ensuring survivability of both the services. We consider two levels of survivability for IP services, i.e., ensuring (a) connectivity of the IP topology and (b) fully reroutable capacity for the IP topology under all single physical link failures. For both the scenarios we propose an integer linear program solution and a computationally efficient heuristic. We observe that integrated provisioning with backup capacity sharing enables much higher resource utilization allowing up to a 60% decrease in wavelength channel usage in supporting IP traffic and up to 35% in the total number of wavelength channels needed to support both IP and wavelength services compared with no backup capacity sharing in our experiments.