One thing is to create a standard benchmark for your current solution, eg a dataset and some standard queries, and run it occasionally. When they propose a "be

One thing is to create a standard benchmark for your current solution, eg a dataset and some standard queries, and run it occasionally. When they propose a "better" solution, point them at the benchmark and wish them well. This will achieve the two goals of measuring raw performance and keeping them out of your hair.