| DNA that does not form part of genes, that is which is never transcribed into proteins. The average beastie has an inordinate amount of DNA that seems to serve no purpose. A genome can be 90% junk. Why?
It's not that it's some mysterious stretch of DNA whose function is at present unknown. Some gene whose effect is yet to be discovered. Junk DNA can be a single base repeated hundreds or thousands of times, or a short group of bases. It's impossible that this can contain any information in the genetic code.
Possibilities: it just bulks up the chromosome physically, for some gross chemical purpose. Or it breaks it up logically, so that there is a greater likelihood that an arbitrary crossover point will be between genes rather than within them. But given that crossover within a gene is typically the same as crossover at one end, since alleles are identical to each other almost everywhere unless you happen to get on the wrong side of a SNP, it's a bit dubious how natural selection could favour increasing lengths of junk just to avoid splitting two SNPs. But maybe it does, I don't know.
Some of it is docking sites adajacent to genes, so that the transcription mechanism can identify a gene accurately and get a grip on the beginning. However, these regions may have well-defined marker sequences, so it wouldn't be right to call this junk. In any case, how much of the neighbouring DNA do you really need to identify the boundaries of a gene? Not 90%+.
Some of it might be selfish DNA, sequences that get transcribed and duplicated by RNA without contributing proteins, but that can't explain the untranscribable lengths of CCCCCCC.... | 
