Chapter 16


16.1 Introduction

The concept of a transaction has been very useful in allowing concurrent processing of data with consistent results. A transaction is a sequence of operations such that that entire sequence appears as one indivisible operation. For any observer, it appears as if the entire sequence has been executed or none of the operations in the sequence have been executed. This property of indivisibility is preserved by a transaction despite the presence of concurrency and failures. By concurrency, we mean that multiple transactions may be going on at the same time. We are guaranteed that the transactions do not interfere with each other. The concurrent execution of multiple transactions is equivalent to a serial execution of those transactions. Further, if the transaction has not committed and there is failure, then everything should be restored to appear as if none of the operations of the transaction happened. If the transaction has committed, then the results of the transaction must become permanent even when there are failures.

As an example of a transaction, consider transfer of money from account A to account B. The transaction, say T1 can be written as

begin-transaction               withdraw x from account A;               deposit x to account B;end_transaction

Once the two operations of withdrawing and depositing have been grouped as a transaction, they become atomic to the rest of the world. Assume, for example, that another transaction ...

Get Concurrent and Distributed Computing in Java now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.