MENTOR (“Middleware for Enterprise-Wide Workflow Management”) is a joint project of the University of the Saarland, the Union Bank of Switzerland, and ETH Zurich [1, 2, 3]. The focus of the project is on enterprise-wide workflow management. Workflows in this category may span multiple organizational units each unit having its own workflow server, involve a variety of heterogeneous information systems, and require many thousands of clients to interact with the workflow management system (WFMS). The project aims to develop a scalable and highly available environment for the execution and monitoring of workflows, seamlessly integrated with a specification and verification environment.

Enterprise-wide workflow management where workflows may span multiple organizational units require particular consideration of scalability, heterogeneity, and availability issues. The Mentor project, introduced in this paper, aims to reconcile a rigorous workflow specification method with a distributed middleware architecture as a step towards enterprise-wide solutions. The project uses the formalism of state and activity charts and a commercial tool, Statemate, for workflow specification. A first prototype of Mentor has been built which allows executing specifications in a distributed manner. A major contribution of this paper is the method for transforming a centralized state chart specification into a form that is amenable to a distributed execution and to incorporate the necessary synchronization between different processing entities. Fault tolerance issues are addressed by coupling Mentor with the Tuxedo TP monitor.

The paper analyzes how financial researchers manage large numbers of time series and how they work with these data. The authors show that replication services are a central facility of a time series management system and define the requirements for such replication services. An evaluation of current time series management systems shows that they do not support replication. Neither replication systems of other database management systems nor other kinds of currently available replication systems cover their requirements. Thus, they present their design of replication services that are adapted to the needs of time series management. It is based on a publish-and-subscribe mechanism, a sophisticated scheduling facility, dynamic replication, and integrated directory services.

No abstract available.

The analysis of time series is a central issue in economic research and many other scientific applications. However, the data management functionality for this field is not provided by general-purpose DBMSs. Therefore, we propose a data model of a specialized time series management system (TSMS) which accounts for these needs. The model is centered around an object-oriented kernel that offers the classes and value types needed for the target applications. The model provides base classes for multivariate time series and for groups as a means to hierarchically partition the time series space. The system offers a computationally complete data manipulation language including capabilities to query time series and groups. An elaborate array model is supported to account for the functional needs of statistical computations. Furthermore, a customizable calendar system providing a variety of predefined calendars is included.>