I find this approach strange. Surely if you're sorting data, don't do it naively by raw bytes. Use your knowledge of the data type and sort using the appropriate comparison operations of the data type. E.g. sorting by 64 bit ints little endian, do the comparison based on that. If using packed numbers, they must be unpacked first.
Generic data stores often don’t have this luxury — if you’re designing a system in which data is relatively opaque, you’re often forced to work with bytes. (e.g. rocksdb, etc)
