Meaningful benchmarks are hard. Having said that, in this particular context it is a good metric to measure the latency for the simplest possible function invoc

Meaningful benchmarks are hard. Having said that, in this particular context it is a good metric to measure the latency for the simplest possible function invocation as that is important for serverless at the edge (and why you put those workers/functions as close as possible to the clients in the first place)