For the Rz, by its definition, you do not need to measure all peaks and values, just the 5 biggest of each, in absolute terms. See:
If your measured profile (or, more precisely, your upper cut-off length Lc, which is taken as the basis of roughness calculations) is not long enough to contain 5 peaks or valleys, then it is unfit for purpose and should be increased.
Depending on where they came from, metallic samples are not necessarily homogenous, so hardness will vary. Microhardness testing, which tests the material at a much smaller scale than its macro counterpart, will register greater differences. Usually there is a steep gradient near the surface, due to work-hardening or even explicit case-hardening processing. The hardened zone varies in depth and is wider for forged or drawn surfaces than machined ones. You can check this if you measure very close to the edge of the transverse section of your sample.