Twitter moved on from having a single table of tweets a long long time ago. I imagine archived tweets (which aren't necessarily that old) are spread out in mult

Twitter moved on from having a single table of tweets a long long time ago. I imagine archived tweets (which aren't necessarily that old) are spread out in multiple archive databases and perhaps not even in traditional db formats. And I imagine that, as a rule, no user facing code is allowed to touch any of the databases, but rather goes through services with heavy caching. So that a seemingly simple feature is an engineering challenge really isn't that surprising. Twitter is optimized for very recent content.