Why is scrambling necessary?

Question

Computer networks

Question. Why do we need scrambling that adds extra bits (and does not add) if, for example, Manchester coding eliminates the possibility of signal desynchronization?
And an additional question: are there holes between bit sequences at the data link layer, or are they only made between frames and cells?

Answer 1

Good question.
As far as I remember, the Manchester code in the world of Ethernet (as I understand it, it is about this technology) is used only in the 10BASE-T implementation. In 100BASE-TX, if I'm not confused, 4B / 5B spread coding is used, then scrambling (i.e. replacing characters with others), then MLT-3. I will assume that Manchester coding was abandoned for reasons of compactness of the spectrum and other RF parameters (when using NRZI or MLT-3, there are fewer fronts). In addition, spread coding (4B/5B) allows up to 16 characters to be used as service characters (JK, TR in the case of Ethernet 100BASE-TX). 100BASE-TX interface, for example, in the absence of user data, constantly sends service characters, which allows you to quickly detect the loss of a link.
Scrambling (100BASE-TX and above), i.e. replacing characters with others, as far as I know, is done when twisted pair is used to normalize such characteristics as cross-talk.
By the way, I found an interesting picture . And an article .