We have our own implementation for compare the keys lexicographically. To make the code more maintainable, we should consider to use the standard memcmp function instead as it should have already contained most of the optimizations that we hand roll out ourselves in the standard implementations.