"Junk DNA" was a very unfortunate term coined in the early days of genetic sequencing when it was thought that DNA --> Protein was the major player. It is now known that it is not anywhere near that simplistic. We now suspect that the majority of "non-coding" sequences actually have a function. Even spacer DNA that is 1000's of bases long serves an important function even though they don't "do" anything other than keep two segments seperated. The term "junk DNA" really needs to completely fall out of usage.
This is an extremely minority view among actual working Geneticists. It essentially springs from the extreme overreach in the selling of the ENCODE project in which they absurdly equated 'transcribed' with 'has a function'. In fact, we
know that a large chunk of DNA is junk because we know exactly what it is, and the majority of the rest is almost certainly junk. There are bits and pieces of signal among the noise but the majority of it has no function at all. The differential rates of mutation between Junk DNA and coding DNA across evolutionary time should be sufficient to convince you of that.