The topic of catching performance regressions came up again recently, which reminded me of this excellent work by David Daly and the team at MongoDB