http://blogs.msdn.com/b/axinthefield/archive/2012/08/01/database-maintenance-strategies-for-dynamics-ax.aspx
http://www.brentozar.com/archive/2009/02/index-fragmentation-findings-part-1-the-basics/
That reminded me of an article which I found a couple of years back that described index fragmentation in this manner:
"Index fragmentation is a lot like cholesterol. The bad kind, not the good kind. It builds up slowly. Some deletes occur, leaving empty space in a data page here. Inserts occur, but the target page is packed, so a page split occurs so the record can be inserted in the correct order, yet the other page is now mostly empty. Updates are a double whammy. Over time, your index pages continue to be less and less full, meaning you have to perform that many more reads per query.
Just like
cholesterol, it’s not perceptible. Sure, if you compared yourself now to 10
years ago, you’d be able to instantly recognize that you feel tired all the
time, or occasionally dizzy spells or have blurred vision. Then all of a sudden
— HEART ATTACK!"
