Theory of serializablity for a parallel model of transactions

Authors:
Ravindran Krishnamurthy;Umeshwar Dayal
Affiliations:
IBM Thomas J. Watson Research Center, Yorktown Heights, New York;Computer Corporation of America, Cambridge, Mass.
Venue:
PODS '82 Proceedings of the 1st ACM SIGACT-SIGMOD symposium on Principles of database systems
Year:
1982

Citing 7
Cited 1

Concurrency control in a system for distributed databases (SDD-1)

ACM Transactions on Database Systems (TODS)
Parallel Program Schemata and Maximal Parallelism II: Construction of Closures

Journal of the ACM (JACM)
The serializability of concurrent database updates

Journal of the ACM (JACM)
An anomaly in disk scheduling: a comparison of FCFS and SSTF seek scheduling using an empirical model for disk accesses

Communications of the ACM
The notions of consistency and predicate locks in a database system

Communications of the ACM
An optimality theory of concurrency control for databases

SIGMOD '79 Proceedings of the 1979 ACM SIGMOD international conference on Management of data
DIRECT - a multiprocessor organization for supporting relational data base management systems

ISCA '78 Proceedings of the 5th annual symposium on Computer architecture

Foundations of data-aware process analysis: a database theory perspective

Proceedings of the 32nd symposium on Principles of database systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we present a parallel program schema model of a transaction system and generalize the concept of serializability from the sequential two-step model to a parallel multi-step model. We define two classes of serializable executions, and for each class we discuss two problems: recognition and scheduling. It is shown that the results for the recognition and online scheduling problems for the sequential model generalize to the parallel model. But it is argued that online scheduling is not suitable for a parallel execution environment. Therefore, batch schedulers are defined and a minimal set of precedence constraints is derived. Finally, it is shown that any optimal batch scheduler that uses syntactic information alone cannot be efficient.