Introduction to Transaction Management

Abstract

Up to this point the basic access primitive that we have considered has been a query. Our focus has been on retrieve-only (or read-only) queries that read data from a distributed database. We have not yet considered what happens if, for example, two queries attempt to update the same data item, or if a system failure occurs during execution of a query. For retrieve-only queries, neither of these conditions is a problem. One can have two queries reading the value of the same data item concurrently. Similarly, a read-only query can simply be restarted after a system failure is handled. On the other hand, it is not difficult to see that for update queries, these conditions can have disastrous effects on the database. We cannot, for example, simply restart the execution of an update query following a system failure since certain data item values may already have been updated prior to the failure and should not be updated again when the query is restarted. Otherwise, the database would contain incorrect data.