Of course analog systems must be treated carefully, but I think we shouldn't rush to do away with the practice entirely. Naturally, we do want to study black holes directly (with experiments like LIGO, LISA, etc.), but this is very expensive, difficult, and time consuming. If a small lab wants to try out some gravitationally analogous system, why not?
that is exactly the point of analogue systems - identical equations have identical solutions, and thus analogue system allows to observe and explore the solutions (which in many cases aren't that easy/possible to obtain mathematically/numerically)
Also, a modern version of the ether is pretty well accepted - quantum fields that occupy all of space. Particles etc.. are disturbances in this field. Ether is something of a tainted term these days so quantum fields are not referred to as ether.
To my mind, we should take this as a lesson that the notion of analogue systems (e.g., using results from fluid dynamics to make statements about gravitation) is bogus. The ether episode seems to indicate that if we want to make statements about gravitational Hawking radiation, we need to do tests on a system where the gravitational dynamics are important (e.g., a black hole), not an unrelated system that just happens to be governed by an identical set of equations in some regime.
Another example of people incorrectly conflating properties of an analogue system with the real system of interest is the Nature monopole paper: http://www.nature.com/nature/journal/v505/n7485/full/nature1... The actual result contained in the paper is that you can make a BEC whose vorticity field has regions of nonzero divergence. It is mathematically similar to (but fundamentally unrelated to) actual magnetic monopoles. Nonetheless, the paper contains the phrase "Dirac monopoles" in the title and then spends half the abstract talking about actual magnetic monopoles, as if the result is somehow relevant to actual magnetic monopole searches (which it isn't).