PI Server is taking a lot of resources to interpolate a value before the first archived value

In a rescent issue we faced, we discoreved that when you try to interpolated a value before the first archived value on a system with a lot of archive files, the system must read all the archive files trying to find a previous value for interpolation. For a small amount of values, the speed stay reasonable but for a PI Vision calculation that tries to compute a trend for a month, it consume so much resources that it could freeze the PI server. As we get new values, the problem tends to go away, but new PI points are all vulnerable to this issue that can break a PI system if not handle.

See Case 01263765