Providing integrated access to information stored at multiple, distributed, autonomous data sources has received much interest in recent years.

Due to the large volume of data involved, and taking the distance of the data sources and the site where the integrated information is provided into consideration, materializing this integrated information as a view is a preferred approach. However, maintaining this view when the information at the data sources changes poses some challenges due to the problems of interfering updates, misordering of messages and the lost of these messages.

Most existing solutions for view maintenance place the assumption that messages sent between the data sources and view site are delivered in a first-sent-first-received manner, and these messages never fail to reach their destination.

We proposed the use of the concept of version numbers in our algorithm that ensures that the materialized view is still maintained correctly when any of the above two assumptions fail. By taking the join graph into consideration when querying the base relations, we avoid the execution of a Cartesian product, which otherwise degrades the performance of the algorithm. And together with the use of the version numbers, multiple base relations within the same data source can be accessed via a single query, and different data sources can be queried in parallel in our proposed algorithm, while existing solutions can only query one base relation at a time.