TechTakes

Everything after this is so pointlessly condescending and confusing. Even if someone knows what monotonic ids are it doesn't automatically mean they're going to have any clue about what that means with regards to index performance. In the spirit of not being an asshole, I'll write it out here based on my research since everyone else just seems interested in putting others down rather than being helpful.

• "Monotonic" implies something that is always increasing (or decreasing). You'll never get a result that's lower than one you've gotten before (or higher if you're dealing with monotonically decreasing stuff).
• Random UUIDs are not monotonic because they're random.
• Even time based UUIDs are not monotonic because of the format. Rather than being store high, medium, low, they're stored low, medium, high. Think of it like storing numbers like "1 20 300" for 321. 322 would be "2 20 300". To make it worse, the end of them is "random" (a MAC address). So, not monotonic at all because MAC addresses can change. (See here for proposed new formats, where they mention this as a problem https://www.ietf.org/archive/id/draft-peabody-dispatch-new-uuid-format-04.html)
• Monotonic primary keys are useful because they're more easily inserted into an index because you're always inserting into one specific part of the index rather.