We present a holistic approach for pedestrian routing that allows computing shortest paths that may have indoor and outdoor sections. Such routes arise, for example, when the destination is not just an address but a specific store in a large mall, or when one needs to get to a certain track at a large train station. Currently, map services as Google Maps or OSRM do not offer such functionality. We identify and overcome three main challenges for answering such complex route planning queries: (i) Pedestrian routing requires fine-grained data, as the location of stairs and elevators, building dventrances, building footprints, and elevation/level information. A single missing staircase can change the length of the computed path severely. (ii) Indoor routing has to be integrated carefully into classical path planning to allow the computation of sensible routes that may enter and exit buildings. (iii) Given the large amount of data to be considered in a query, acceleration techniques need to be applied in order to achieve interactive query times. Retrieving barrier-free routes for wheelchairs is also our important use case.