I do not speak Spanish, but I can use google translate. Hope the translation honestly reflects what you mean to say.

There are few good AVL implementations, but there are a lot of good Red-black tree libraries. In my benchmark (“Another look at my old benchmark”), AVL and RB tree have quite similar performance in terms of both speed and memory. And therefore, if you like to use binary search tree (BST), you might want to use RB tree instead. For C++ programmers, STL’s implemention is nearly the best. For C programmers, I recommend tree.h from FreeBSD. See my post (“Comparison of binary search trees (BSTs)”) for more details.

Although in my benchmark, B-tree marginally outperforms RB tree, in other applications, RB/AVL tree may be faster. More advanced implementation of RB may also use less memory than my B-tree. But this all depends on the type of data in the BST/B-tree.

I have not tested other B-tree (B*/B+) implementations. But if you want to use an on-disk B-tree for large data set, I would highly recommend SQLite. It is faster than another B-tree library and is much more flexible.