When creating a Node Key constraint does the order of the fields matter?

I need to create a composite constraint using 3 fields:

  • One with completely random strings
  • Another one that's numeric and monotonically increasing
  • A third one that's skewed towards low valued integers around 0.

When defining the constraint, has the ordering of the fields have something to do with performance?

Should I put the most discriminative ones first?