My P4 is very rusty, but back when I used a vanilla install a branch was basically just a copy operation from one part of the namespace to another, and then setting up your view to point to that new spot (or setting up a separate view for it if you wanted). This was before any integration with git (or, indeed, the existence of git at all), as well.
Subversion mostly copied the way p4 did it, but without the view control stuff (where there's a server-registered and quasi-versioned description of what local file paths map to what server paths) that perforce has/had.
Hmm, that doens't sound bad.
I don't remember (nor fully understood) why it was that complicated at google. At a wild guess it's ultimately connected to google's choice to keep all company code in a single humongous repo, which required a lot of later creativity to keep P4 from collapsing under the weight...
I would think that the amount of data wouldn't be the thing that would make p4 fall over (it's pretty decent at that), but the number of users. P4, iirc, stores all client state server side, even to what files you're editing. So if you had to work around anything, it was probably that, and I can definitely see how that would break the ease of the workflow.
Subversion mostly copied the way p4 did it, but without the view control stuff (where there's a server-registered and quasi-versioned description of what local file paths map to what server paths) that perforce has/had.