I got super lucky and came up with the question an hour before I first administered the interview. Memcached was the first codebase I went for, because someone had asked me a question involving it a few months ago, and it was conceptually a "simpler" version of MemSQL.

Later on, we tried to diversify the question to other codebases (because it leaked, although less publicly). I remember trying it out on Redis, which was too easy because the codebase is so nice. We never found a great alternative while I was there (through 2016). Although, the team may have found something since.

At my current company, Impira, we have a similar question on a different codebase :)