Junk DNA is repeating codons, or codons that occur in areas that are outside of the "start/stop" codon triplicate pairs. A DNA transcribing protein will read the genetic code from a start signal, until it gets to a stop signal. Then it clips itself off the chain and re-binds the chain together for the next transcriber to use. Sometimes there are extra codons between a stop signal and the next start signal, sometimes there are hundreds of thousands of extra codons. They aren't there for structural reasons, all DNA is the same 4 codons linked together over and over, all the different chromosomes are different sizes. All of this DNA is reported when the cells divide, that's the only time those regions between the stops and starts actually come into play. This is very easily proven, we know the structure of the reading proteins down to the molecule (indeed there are starts and stops and triplicate base pairs that design these transcribing proteins). The "important" junk DNA that has significance while not being in a "start->stop" zone are the codons that occur before the first start codon on either side of a DNA strand, when DNA is replicated the protein that starts replicating it has to start at 1 end of 1 side of the DNA in order to be able to read it, except it needs to find the end first, and to make sure it's all the end it "clips" the first 6 (? Maybe more maybe less, it's been decades since I've studied this) codons from the strand of DNA, this is lost for all future replications of the cell, your DNA actually gets shorter every time your cells reproduce (except your miosis division cells, they have a special replication process that keeps the full length of every chromosome).
Sorry for the wall of text, but there's plenty of examples of blatantly junk DNA, and there are known methods of how it occurs. Anyone who says every codon pair has a purpose has a screw loose and is ignorant to the mechanics of evolution.