Can we simplify concurrent access to B-trees for database efficiency? This paper explores overcoming the difficulties of concurrent operations on B-tree data structures, crucial for storing large amounts of information. The research introduces a practical storage model using a single additional “link” pointer in each node to enable easy recovery from modifications by other concurrent processes. The proposed solution offers a locking scheme that is simpler than previous methods, eliminating the need for read-locks and limiting the number of locked nodes to a small constant. This approach enhances the efficiency of update processes and reduces contention in B-tree operations. An informal correctness proof is provided for the system, suggesting its reliability and practicality for real-world database applications. This work is pertinent for applications that involve computer science and software engineering.
Published in ACM Transactions on Database Systems, this article is well-suited for the journal's focus on theoretical and practical aspects of database management systems. The paper's exploration of efficient locking mechanisms for concurrent B-tree operations aligns with the journal's emphasis on improving database performance. References within the article likely connect it to other works on database concurrency and data structures.