In Unity 2019+, you can create world-space canvases and place UI objects on them. The Canvas comes with a Canvas Scalar component, which has two parameters: Dynamic Pixels Per Unit and Reference Pixels Per Unit. Changing these will scale the way text and images are displayed in the canvas.
However, the canvas transform has a scale component you can use to scale the canvas as a whole.
My question: is there any reason (performance, future-proofing etc) to chose one approach over the other?