SPECIFICATION AND MANAGEMENT OF INTERDEPENDENT DATA IN OPERATIONAL SYSTEMS AND DATA WAREHOUSES

(Inter)Dependent objects include data replicated or cached in multiple database systems, data collected and summarized in data warehouses fo r analysis, planning, and decision support, as well as any other categ ory of objects whose states are related and they are maintained in dif ferent information systems. In this paper we discuss dependencies betw een objects in an environment consisting of operational systems and a data warehouse, and describe their specification and enforcement. To s pecify object dependencies we introduce Object Dependency Descriptors (ObjectDDs). These describe the relationships between dependent object s, and define how much inconsistency between original objects and thei r replicas/collections/summaries can be tolerated before it is necessa ry to restore their consistency. Object dependencies are enforced by e xtended transactions designed specifically for evaluating if dependent objects satisfy their specified relationships, evaluating whether pos sible inconsistencies can be tolerated, and (if not) restoring consist ency. To describe the transactional behavior of such consistency evalu ation and restoration transactions we use Transaction Dependency Descr iptors (TransactionDDs). TransactionDDs define the transactional relat ionships between consistency evaluation and restoration (asynchronous) transactions, as well as the relationships between such asynchronous transactions and regular (synchronous) transactions executed directly by applications. To automatically maintain the consistency of dependen t objects, we propose the concept of a Dependency Management System (D MS). A DMS monitors dependent objects, evaluates object consistency, a nd schedules and controls consistency restoration transactions to keep dependent objects within acceptable consistency levels. We describe k ey components in the DMS architecture, and a relatively simple impleme ntation involving straightforward extensions in a relational DBMS.