I'm working on a very large production system which needs to store many, many small segments of similar, text-like data (think alphanumeric with many similar substrings). If we gzip each segment ...