how did you benchmark this against much less convoluted solutions, like "a text file"?

how much better was this to justify all that extra complexity?