This paper describes the overall architecture and design aspects of a hybrid relational and XML database system called System RX. We believe that such a system is fundamental in the evolution of enterprise data management solutions: XML and relational data will co-exist and complement each other in enterprise solutions. Furthermore, a successful XML repository requires much of the same infrastructure that already exists in a relational database management system. Finally, XML query languages have considerable conceptual and functional overlap with relational dataflow engines. System RX is the first truly hybrid system that comingles XML and relational data, giving them equal footing. The new support for XML includes native support for storage and indexing as well as query compilation and evaluation support for the latest industry-standard query languages, SQL/XML and XQuery. By building a hybrid system, we leverage more than 20 years of data management research to advance XML technology to the same standards expected from mature relational systems.

Automatic summary tables (ASTs), more commonly known as materialized views, are widely used to enhance query performance, particularly for aggregate queries. Such queries access a huge number of rows to retrieve aggregated summary data while performing multiple joins in the context of a typical data warehouse star schema. To keep ASTs consistent with their underlying base data, the ASTs are either immediately synchronized or fully recomputed. This paper proposes an optimization strategy for simultaneously refreshing multiple ASTs, thus avoiding multiple scans of a large fact table (one pass for AST computation). A query stacking strategy detects common sub-expressions using the available query matching technology of DB2. Since exact common sub-expressions are rare, the novel query sharing approach systematically generates common subexpressions for a given set of "related" queries, considering different predicates, grouping expressions, and sets of base tables. The theoretical framework, a prototype implementation of both strategies in the IBM DB2 UDB/UWO database system, and performance evaluations based on the TPC/R data schema are presented in this paper.

Materialized views (or Automatic Summary Tables—ASTs) are commonly used to improve the performance of aggregation queries by orders of magnitude. In contrast to regular tables, ASTs are synchronized by the database system. In this paper, we present techniques for maintaining cube ASTs. Our implementation is based on IBM DB2 UDB.

Incremental refresh of a materialized join view is often less expensive than a full, non-incremental refresh. However, it is still a potentially costly atomic operation. This paper presents an algorithm that performs incremental view maintenance as a series of small, asynchronous steps. The size of each step can be controlled to limit contention between the refresh process and concurrent operations that access the materialized view or the underlying relations. The algorithm supports point-in-time refresh, which allows a materialized view to be refreshed to any time between the last refresh and the present.

We investigate the problem of using materialized views to answer SQL queries. We focus on modern decision-support queries, which involve joins, arithmetic operations and other (possibly user-defined) functions, aggregation (often along multiple dimensions), and nested subqueries. Given the complexity of such queries, the vast amounts of data upon which they operate, and the requirement for interactive response times, the use of materialized views (MVs) of similar complexity is often mandatory for acceptable performance. We present a novel algorithm that is able to rewrite a user query so that it will access one or more of the available MVs instead of the base tables. The algorithm extends prior work by addressing the new sources of complexity mentioned above, that is, complex expressions, multidimensional aggregation, and nested subqueries. It does so by relying on a graphical representation of queries and a bottom-up, pair-wise matching of nodes from the query and MV graphs. This approach offers great modularity and extensibility, allowing for the rewriting of a large class of queries.

Programming languages and databases have evolved on separate foundations and with separate goals for many years. Buzzwords such as persistent data objects, objectoriented databases and software engineering databases reflect increased activity aimed at integrating the two areas at their current state of evolution. This paper suggests that we retum to basics and consider the foundation of the two areas, grammars and relations, as a basis for their integration. We present an algorithm, GeneRel, which given a grammar automatically generates a set of normalized relations in which objects derived from the grammar can be stored. We demonstrate how the algorithm applied to a meta- grammar generates relations in which grammars derived from the meta-grammar can be stored. We show how an extended relational algebra can be used to retrieve information about stored grammars and stored objects derived from the grammars. We outline an algorithm, GeneView, which given a set of non-terminals from a grammar generates a set of view definitions on the created relations.

No abstract available.