A series of experiments were conducted using both CloudLab and a dedicated laboratory testbed to test the performance of the Cognitive Space Gateways (CSG)-an approach to DTN routing that exploits a spiking neural network as a learning element to discover and dynamically forward data bundles. The experiments focus on evaluating the average response time of bundles and throughput of a data flow sent at given traffic intensity over an emulated space network. The same topology and identical, repeatable tests were run on both facilities. The results help to validate the performance of the CSG routing approach compared to the Contact Graph Routing algorithm using independent experimental settings. Furthermore, the experiments reveal some of the challenges associated to the use of network virtualization, as done in CloudLab, when applied to network performance measurements.