Optical network technologies based on a single-core and single-mode fibers have a limited capacity and cannot provide enough resources to a constant increase of bandwidth demands. One approach to overcome this is the use of Space-Division Multiplexing (SDM) which relies on sending data through multiple cores embedded into a single strand of fiber or using multiple transverse modes supported by a core. The combination of the flexibility of Elastic Optical Networks (EONs) and the high capacity of SDM is a promising solution to cope with the bandwidth demands.At the network level, when a traffic request arrives, it needs to reserve network resources to establish it. One approach to accommodate traffic demand over optical networks is the Routing, Core and Spectrum Allocation (RCSA), in which end-to-end lightpaths are offered for each individual request. In these scenarios, during the allocation process, some constraints need to be respected, such as contiguity and continuity of slots (selected in the resource selection process), and spatial crosstalk. These constraints pose extra complexity to accommodate the requests for the lightpath establishment.As one of the possible solutions, network virtualization is capable of improving the efficiency of optical networks, by allowing virtual networks to share the resources of physical networks, simplifying the management of resource and providing flexibility in resource allocation. One of the main challenges of network virtualization is to configure a virtual network efficiently which comprises allocating physical resources to accommodate incoming virtual networks.This thesis proposes solutions to the RCSA problem and the virtual network configuration problem for SDM-EON networks. The first contribution of this thesis is an algorithm to promote an equilibrium between reduction of the network energy consumption and reduction of the blocking of requests. For this purpose, we introduce a traffic grooming algorithm using batches, which takes advantage of the deadline of each request to form batches, increasing the chances of the requests to be established at another time.The second contribution of this thesis is a set of algorithms using different techniques to handle the fragmentation problem, where a small portion of available slot sequences end up scattered in a fiber link, blocking future requests, called the fragmentation problem. For this purpose, we propose proactive and reactive algorithms. Proactive algorithms use different techniques, such as multipath routing, core, and area prioritization, and metrics to use in the route selection process. The reactive algorithm uses machine learning to rearrange the spectrum and tune the RCSA algorithm to prevent the fragmentation.The third contribution of this thesis proposes a solution to improve resource sharing in network virtualization. This problem consists in configuring virtual links and nodes to physical nodes and paths. For this purpose, we propose a learning assistant control loop to handle the virtual network configurati...