It’s often less about engineering effort and more about taking some small risks to try less mainstream (but still relatively mature) alternatives by reasoning from first principles and doing a bit of homework.
From our experience, you can actually end up in a situation that requires less engineering effort and be more stable, while saving on costs, if you dare to go to a bit lower abstraction layers. Sometimes being closer to the metal is simpler, not more complex. And in practice complexity is much more often the cause of outages rather than hardware reliability.
From our experience, you can actually end up in a situation that requires less engineering effort and be more stable, while saving on costs, if you dare to go to a bit lower abstraction layers. Sometimes being closer to the metal is simpler, not more complex. And in practice complexity is much more often the cause of outages rather than hardware reliability.