Most of this random strings will "globally" enlarge, but will locally compress. For example sequences of compressible subsequences of two-three bytes are very f

Most of this random strings will "globally" enlarge, but will locally compress. For example sequences of compressible subsequences of two-three bytes are very frequent. The remaining data that can't be compressed will stress the other code that emits verbatim uncompressed sequences. To stress more the compression with valid sequences it's possible to use a very big HTML file and get ranges at random instead to get this strings from /dev/urandom