Simple, you just allocate a bigger contiguous chunk of memory and copy the entire vector over when the current chunk maxes out.