MobileEVE: A Distributed System for Data Warehouse Maintenance under Schema Changes

摘要

In a world more and more dependant of all kind of networks connecting all kind of information sources, data warehousing techniques seem to be a guarantee for informational resources availability. Data warehouses are constituted by informational resources collected from distributed information sources. They have to be regularly updated and maintained, when information sources change their contents and/or their schema, in order to ensure the integrity, the accessibility, the availability and the consistency of the afforded information. Changes detection is considered as the front line of the Data Warehouse Maintenance System (DWMS) under Schema Changes. This task is so important that the accuracy of the whole system depends on it. In previous research, the change detector is supposed to be downloaded into the information source location and is well thought-out as a static object. This kind of detector is no longer efficient due to the spread of information through networks and the need to integrate intelligence in it. In fact information source can not be only implemented on a single machine but also on a local or wide area network. For this reason mobile agent technology imposes its self as not only as a new solution to schema change detection but also for the whole maintenance system. This technology is able to avoid network saturation, to reduce communication costs, to integrate intelligence and take advantage of and improve DWMS components autonomy. In this paper, we propose to study the issues of using mobile agents to achieve data warehouse maintenance under schema changes. We focus on the detector functioning and its collaboration with the other DWMS agents leading to the availability optimization of informational resources. Our solution is modeled using UML which will be a great help, not only for determining mobile agents' representation, interaction and functioning, but also for identifying DWMS parallel processes. The proposed solution permits not only a clear design allowing system evolving and comprehension, but also to decrease the maintenance process time, to avoid network saturation and to increase data warehouse maintenance system components availability.