Adaptive Commitment for Real-Time Distributed Transactions

• Authors: Nandit Soparkar;Eliezer Levy;Henry F. Korth;Avi Silberschatz
• Venue: Adaptive Commitment for Real-Time Distributed Transactions
• Year: 1992

Real-time distributed transaction management systems are useful for both real-time and high-performance database applications. Guaranteeing response times in such environments is difficult to achieve mainly due to the inherent asynchrony present. The standard approach to distributed transaction management is to employ the two-phase locking scheme in each of the participating sites, and to coordinate the executions of the various subtransactions through the use of the two-phase commit protocol. Such an approach ensures the atomicity and serializability properties of the transactions. Unfortunately, the unpredictability, the cost and the fault-tolerance properties of the two-phase commit protocol render it unsuitable for real-time applications. The approach taken in this paper is to identify ways in which a commit protocol can be made adaptive in the sense that under situations that demand it, such as a transient local overload, the system can dynamically change to a different commitment strategy. The decision to do so can be taken autonomously at any site. The different commitment strategies exploit a trade-off between the cost of commitment and the obtained degree of atomicity. The inexpensive protocols incur a reduced cost as they are based on the optimistic assumption that transactions failures are the exception rather than the rule. When transactions do fail, these protocols rely on local compensatory actions to recover from non-atomic executions. We provide the necessary framework to ensure the logical and temporal correctness criteria, and describe examples to illustrate the use of our strategies.