I would be interested in seeing what your actual proposed solution to the example problem is here. You seem to be implying that we should pull back the full list of posts and order & limit in memory. This might make sense for some scales and workloads but it definitely wouldn't be my first or second solution for the given example.
today, lambda (speed layer + batch layer) or kappa architecture (kafka streams) -- ymmv. not sure if I would build that infra out just for this single problem, but generally once that infra is in place it can be applied to many problems including this one.
edit: (assuming i can't get it working with a jury rigged cronjob and the usual dumb solutions that will still use the ruby code. i'd try a bunch of those types of solutions first obv.)
