Fluid solutions of the EFE are very useful in GR and pervade all of the theory's applications be it cosmological (FRW dusts for matter-dominated models, radiation fluids for radiation-dominated ones...) or astrophysical (stars interior models, etc). Now, I know there is a component of idealization in all these models, but it must have some base on the physical universe we observe. I mean our universe either behaves globally as a perfect fluid or it doesn't, right? So for GR fluid solutions, to what extent is being a perfect fluid a physical feature of our universe as opposed to an unphysical one, i.e. coordinate-dependent?