This paper presents two families of non-locking concurrency control methods for database systems, which are "optimistic" in nature, relying primarily on transaction backup to avoid conflicts between transactions. The authors argue that these methods can be more efficient than locking mechanisms in certain scenarios, such as query-dominant systems and large tree-structured indexes. The optimistic approach involves restricting writes to a read phase, a validation phase, and a possible write phase, with the validation phase ensuring that the transaction does not cause integrity issues. The paper discusses the implementation details of these methods, including the assignment of transaction numbers and the handling of validation failures. It also analyzes the performance of these methods in the context of B-trees, showing that they can achieve high concurrency and avoid locking overhead. The authors conclude by highlighting the potential advantages of optimistic methods over locking approaches, particularly in systems where transaction conflicts are rare.This paper presents two families of non-locking concurrency control methods for database systems, which are "optimistic" in nature, relying primarily on transaction backup to avoid conflicts between transactions. The authors argue that these methods can be more efficient than locking mechanisms in certain scenarios, such as query-dominant systems and large tree-structured indexes. The optimistic approach involves restricting writes to a read phase, a validation phase, and a possible write phase, with the validation phase ensuring that the transaction does not cause integrity issues. The paper discusses the implementation details of these methods, including the assignment of transaction numbers and the handling of validation failures. It also analyzes the performance of these methods in the context of B-trees, showing that they can achieve high concurrency and avoid locking overhead. The authors conclude by highlighting the potential advantages of optimistic methods over locking approaches, particularly in systems where transaction conflicts are rare.