Comment by jeffbee

Wasn't it a conclusion of the Google hard drive reliability study that models based on SMART were not useful? I.e. drives with sector reallocations are much more likely to fail than those without, but their failure rate is still something like 15% per year, so what useful thing can you do with that signal?