“We‘re not just going cheap and deep. We don’t want a lot of gimmicks to deliver QLC, such as the use of storage-class memory as a cache. We get rid of the gimmicks to deliver both ...