A common problem with ultra-high vacuum viewport windows is that often the window material deforms, into a saddle or Pringles chip shape. .
An engineer from a viewport manufacturer explained to me that this can happen when the hoop stress around the window is not fully symmetric. If it were symmetric, the window would simply bulge or cup in the middle. That seems intuitive, but why would an uneven stress cause a saddle-like strain in particular? What is a simple physical explanation for this? So far, the fact that uneven stress in x/y directions cause different curvatures seems intuitive. What is also important is that glass is a very stiff material and cannot stretch (so volume/thickness is assumed to be conserved).
EDIT: the brazing pad (thick black) is where the metal sleeve (thin black lines) is brazed to the braze adapter (red lines representing another cylinder). The glass is hermetically sealed to the bottom of the metal sleeve.