Oct. 13, 2011
Whoever orders books on the Internet, withdraws money from a cash machine or uses a navigation system to arrive at a destination is (usually without realizing it) using companies’ very large databases. These are accessed and managed by computer programs which — depending on the type of application or search request — work quite differently. Saarbrucken computer scientists have recently developed a concept for a database system that automatically adapts to different requirements and thus combines features of previously different systems.
In order to implement this idea in future industrial practice, the Saarbrucken scientists are being sponsored by the ‘Bundesministerium für Bildung und Forschung, BMBF’ (Federal Ministry of Education and Science) with 1.1 million Euros.
Jens Dittrich, professor for information systems at Saarland University, describes the current situation: “Large enterprises spend a lot of money on database systems, since these are the backbone of nearly every company’s applications. Often, many different systems have to be acquired at the same time, due to their different areas of application.” If an enterprise manages only customer data or a depository, it frequently draws on a line-oriented database system, a so-called ‘row store’. For instance, product sales or employee payrolls are registered and managed by this system. “But, if the CEO wants to see the development of an increase in turnover during the last quarter, or his employees’ age distribution, a line-oriented system often is not sufficient. More complex requests require a column-based system, so-called ‘column stores’,” Jens Dittrich explains. Even more challenging demands are imposed on database systems which, for example, control airports or rail traffic. Those systems have to synchronize data permanently and react within milliseconds, in order to prevent crashes. These are the situations in which data stream systems are used.
“Our vision is to develop a database system that combines all different systems used simultaneously on today’s market,” as professor Dittrich characterizes the challenge. His project has attracted broad interest at the Federal Ministry of Education and Science. Within the scope of the ‘VIP-Programm — Validierung des Innovationspotenzials wissenschaftlicher Forschung’ (Validation of the Innovative Potential in Scientific Research), scientists receive support for in-depth use of scientific results, as well as their economic utilization. With the 1.1 million Euros in funding, Saarbrucken scientists are now able to build a prototype of the new database system. To represent the ‘Octopus’ research project, Jens Dittrich chose an extremely versatile animal that adapts its appearance to match its environment. “Like the octopus, our software should find out independently what kind of requirements the environment demands and what adjustments are necessary in return. In the future, we want to offer everything that you now diligently store in different databases, in only one database system.” For companies, this would offer the advantage that the data within the systems wouldn’t have to be painstakingly synchronized. Furthermore, corner cases that, so far, haven’t fit in one of the existing systems could be treated more efficiently. “However, some research remains to be done in order to make this system work in practice,” the Saarbrucken computer scientist says.
At the beginning of this year, the scientists won the first prize for their visionary database concept in one of the top worldwide database conferences, the Conference on Innovative Data Systems Research (CIDR) in California. The category ‘Outrageous Ideas’ recognizes exceptional ideas that normally wouldn’t receive much attention at regular conferences, either because of their visionary character or due to the fact, that they present not yet completed solutions. The prize is intended to promote groundbreaking ideas in database research.
Note: ScienceDaily reserves the right to edit materials for content and length. For original reprint permissions, please contact the source cited above.