The Cost of GUIDs as Primary Keys (2002)

It's fascinating to me how some solutions in programming become 'trendy' and get used everywhere, even when it may not be the right solution.

And then other solutions, like the "COMB GUID" from 2002 have stayed in relative obscurity despite neatly solving a small, concrete problem so many of us encounter.

The author's 20 year old solution is amazingly similar to the recent UUIDv7 draft, right down to assigning 48 bits to the timestamp component; buried on page 6 of an otherwise unassuming article.

This should have been a revelation for everyone building database backed applications, but instead has either been outright ignored or reinvented in various ways over the years instead.

Makes me wonder what other solved-problems our profession has largely overlooked.