| In a multidatabase system, a view is a virtual global relation defined as a query on global schema. Views provide a way to make complex queries easier to compose. When a view is seldom referenced by global queries, efficient view materialization is important. In this dissertation, we first provided a multi-level algorithm for efficient view materialization. When views are frequently referenced, storing materialized views is effective for reducing the query response time. A materialized view is a derived relation from base relations. Any update to the corresponding base relations needs to be propagated to the materialized view to ensure that the data in the materialized view conform with the data in the base relations. This is known as view maintenance. Different maintenances to the same materialized view for different local updates may interfere with each other, and result in incorrect maintenance. This is known as view maintenance anomaly. To correct the problem, we often need to precisely detect the occurrence of each anomaly. However, in a distributed environment, the departure order of a set of messages from one site may be different from the arrival order of these messages at another site due to unpredictable network routings and delays. In such an environment, precise detection of the occurrence of each anomaly is a difficult problem. In this dissertation, we proposed a synchronization method for precisely detecting the occurrence of each maintenance anomaly with the consideration of the possibility that messages may be delivered and received in different orders. This method was then applied to our maintenance algorithm for views defined by SPJ queries. We also proposed a maintenance algorithm to handle views involving outerjoin and data inconsistency. This maintenance algorithm minimizes the number of the queries that need to be sent to the local database systems for view maintenance. Minimizing the number of the queries that need to be sent to local database systems not only reduces the cost of view maintenance but also reduces the chance of encountering anomalies. |
