Interestingly, we've never gone back to tier 1. Once we move from t1 to t2, it stays there. This is only because of our pipeline. Obviously, there's good reason to go from t2 back to t1. I'm wondering if you would really move all of t1 to t2, then back again? I could see moving older data (based on last access time) over to t2... then move from t2 back to t1 if the data is accessed, but you'd only be moving that which is accessed.