The B-tree and its variants have been found to be highly useful (both theoretically and in practice) for storing large amounts of information, especially on secondary storage devices. We examine the problem of overcoming the inherent difficulty of concurrent operations on such structures, using a practical storage model. A single additional “link” pointer in each node allows a process to easily recover from tree modifications performed by other concurrent processes. Our solution compares favorably with earlier solutions in that the locking scheme is simpler (no read-locks are used) and only a (small) constant number of nodes are locked by any update process at any given time. An informal correctness proof for our system is given.

This paper proposes the use of VLSI technology to perform relational database operations directly in hardware. It is shown that relational compulations, such as intersection, remove-duplicates, union, join, and division, can all be pipelined elegantly and efficiently on networks of processors having an array structure. These (systolic) processor arrays are readily and cost-effectively implementable with present technology, due to the extreme simplicity of their processors, and the high regularity of their interconnection structures.

The concurrent manipulation of a binary search tree is considered in this paper. The systems presented can support any number of concurrent processes which perform searching, insertion, deletion, and rotation (reorganization) on the tree, but allow any process to lock only a constant number of nodes at any time. Also, in the systems, searches are essentially never blocked. The concurrency control techniques introduced in the paper include the use of special nodes and pointers to redirect searches, and the use of copies of sections of the tree to introduce many changes simultaneously and therefore avoid unpredictable interleaving. Methods developed in this paper may provide new insights into other problems in the area of concurrent database manipulation.