Techniques are disclosed for discovery and tracking of locations of devices attached to a computing system network. By way of example, an automated method for determining a location of a target element in a distributed network includes the following steps. One or more locations of one or more sample topology elements in the distributed network are learned. Location mappings are generated, the location mappings representing associations between the learned locations of the one or more sample topology elements and one or more network attributes. A location of a target element is estimated using the location mappings.