The Unsplittable Shortest Path Routing problem (USPR) is one of the optimization problems that has been well studied in the field of traffic engineering for IP networks due to its importance for improving the network's Quality of Service (QoS). Given a directed graph representing the IP network and a set of commodities depicting the demands to be sent between its nodes, the USPR problem consists on identifying a set of routing paths and the associated administrative weights such that each commodity is routed along the unique shortest path between its origin and its destination following these weights. Due to the NP-hardness of this problem, we propose a topology aggregation-based approach to solve it, which consists on efficiently aggregating the network's graph to make it more scalable before solving it. The experimental results show the efficiency of this proposal in terms of scalability and network's maximum load reduction compared to other methods from the state-of-the-art.