In terms of redis being single threaded, we put a lot of work into making the search concurrent, see here for details: https://redislabs.com/blog/making-redis-concurrent-with-modu.... Basically we rotate between long queries and short queries, making long queries run slower but not block short ones. Right now rate limiting is left to the user and not solved at the engine level.
Yeah, I was also thinking about adding execution cap to a single query, that's not hard to add as there is a simple scheduler that checks their progress constantly. So just terminating a long running query is simple.
