Some cases the DNA Doe Project have taken on have been put on indefinite hold because they are unable to get sufficient samples. Some of these are not old cases. Of the 8 UID cases from Snohomish County, two were with DDP before being put on hold. I believe DDP mentioned for both that the reason they couldn't get samples was because the bones had been boiled or had some other technique done to them to preserve them/clean them which had actually made DNA extraction very difficult. I believe the ones in question are #UP2886 and #UP6599? Found in 2007 and 2009 respectively, so they're not old cases (though one is a partial cranium only and the other is also skeletal and had been there up to 5 years).
A lot depends on how the remains are treated and stored, as well as how old they are.
Again all MOO from what I've researched, and could be wrong.
This article talks a bit more in detail about Othram Labs and how they're able to use far less DNA than most to make a workable sample for forensic genealogy:
Another ID through DNA: Who killed Rodney Johnson? | HeraldNet.com
It specifically mentions that most of these tests need 750 nanograms of DNA which can be difficult to get from a skeleton, but Othram are able to work with just one fifth of a billion of a gram.