Would a heap onto a SQL DB be an elegant appoach for big data?

If a programming language allocated dynamic memory on a (clustered) DB (through SQL queries) instead of computer RAM, would it be able to solve big data problems directly?