A Dynamic Overwrite Protocol for Multiversion Concurrency Control Algorithms
The main difficulty with multiversion database systems is that of storage space, i.e. in a multiversion database we would need to keep a potentially unlimited number of versions of data items. As storage space is always limited, the number data item versions maintained in the database is also limited. This means that some write operations must overwrite old versions of data items in order to create new ones. This paper presents the novel dynamic overwrite protocol which compared to conventional natural overwrite protocol used in practice minimize the number of transactions abortions caused by the limited storage space. The main idea of this new overwrite protocol lies on finding data item versions which can be safely overwritten because no active transaction accesses them in the future.
KeywordsData Item Concurrency Control Read Operation Version Node Database Operation
Unable to display preview. Download preview PDF.
- D. Agrawal, and S. Sengupta. Modular Synchronization in Multiversion Databases: Version Control and Concurrency Control, Proc. of the ACM-SIGMOD Conf. on Management of Data, pp. 74–83, (1989).Google Scholar
- P. A. Bernstein, N. Goodman, and V. Hadzilacos. Concurrency Control and Recovery in Database Systems, Addison-Wesley Publ. Co. (1987).Google Scholar
- T. Hadzilacos, and C. H. Papadimitriou. Algorithmic aspects of multiversion concurrency control, Journal of Computer and System Sciences 3,3, pp. 297–310, (1986).Google Scholar
- Morzy, T., On the use of the timestamp-ordering algorithm for concurrency control in K-version database systems, Proc. Int. Congress on New Technologies for Software and Supercomputers Development, Caracas, (1989).Google Scholar
- Oracle Corporation, Oracle Database Administrator’s Giude,(1983).Google Scholar
- C. H. Papadimitriou. The theory of database concurrency control, Computer Science Press, (1986).Google Scholar
- A. Silberschatz. A multi-version concurrency scheme with no rollbacks, Proc. ACM SIGACT-SIGOPS Symp. on Principles of Distributed Computing, Ottawa, pp. 216–223, (1982).Google Scholar
- Son, S.H., and Haghighi, N., Performance evaluation of multiversion database systems, Proc. 6th IEEE Data Engineering Conference, pp. 129–136, (1990).Google Scholar
- R. Sun, and G. Thomas. Performance results on multiversion timestamp concurrency control with predeclared writesets,Proc. 6th ACM SIGACT-SIGMOD-SIGART Symp. on Principles of Database Systems, San Diego, pp. 177–184Google Scholar