Is the red-black tree implementation at the bottom of zset better than the skip table?

[褚润发]

Hi, I wonder if the redis team has considered using the red-black tree as the underlying implementation of zset?
I used a red-black tree to implement all the commands of zset, and the test found that the efficiency of both memory usage and random addition and deletion is better than that of skip tables.

If there is an idea in this regard, I can contribute the code as a reference.