We demonstrate the practical implementation of a schema for Smart Grid Last Mile (SGLM) traffic management in leased Internet access networks, based on well-known queue management and traffic marking algorithms such as Random Early Drop and Token Bucket. Internet access links are often over-provisioned, with much higher installed capacity than user needs, so domestic applications do not compete for bandwidth. In this scenario the additional load posed by SGLM traffic is negligible. On some occasions, however, Internet access links are moder- ately stressed and traffic management techniques are required for SGLM traffic to coexist seamlessly with user traffic. In a previous paper, we formulated a theoretical model for SGLM traffic in general networks, and a management solution for its coexistence with user traffic in the particular case of leased Internet access networks. In the present paper, we demonstrate the feasibility of that solution in a practical scenario. We have implemented the traffic management policies on real hardware and employed SGLM traffic generators to evaluate our schema in a real network. The results indicate that our proposal achieves low- latency SGLM communications in a scenario with high domestic user traffic