Generations-based map vs. sparse map

In ClickHouse, we name it ClearableHashMap: https://github.com/ClickHouse/ClickHouse/blob/master/src/Com...

My colleague Michael Kolupaev, who is an ACM ICPC winner, taught me about this technique like it is some sort of common knowledge. Since then, I've been talking about it to everyone :)

HN discussion for the article this is a reply to: https://news.ycombinator.com/item?id=37360314

Does anyone know what data structure is that? I never seen anything like it, so I'm interested in learning how to call it. :)

Isn't it "just" a sequence counter for detecting stale data? Interesting that you got such a huge speedup by reading from fewer memory locations!