Your two mentioned cases also have a turnover of new equipment. There are lots of enterprises that do system refreshes based on depreciation schedules (3-5 years) and when they do, they buy what is current h/w and put the (then) current versions of O/S and app on top of it. And there it sits for the next 3-5 year cycle. Updates (really fixes) are being applied to a static base. Those kinds of issues are easy to handle. If, however, you have *lots* of systems, think 5-6 digit counts, you are in the position that you are *always* buying new gear. You can't help it simply because of the scale. If you only have resources to rack&stack 2000 systems a year but you have 5-10 times that number in service, you are always churning. You now face the problem of either backport hell to get h/w enablement or you move closer to mainline where that enablement was based. This is a painful place to be which means that there is a scale to "enterprise". RH's sweet spot is the former (smaller case), not the later. And if you are that big, you need to apply the inhouse resources to get closer to mainline. A lot of shops do this but not really understanding the scale issues, do it badly.