Karin Kailing, Alexander Löser, Volker Markl
Challenges and Trends in Information Management
Abstract
From a customer perspective, three main dimensions are relevant when evaluating and procuring database systems: functionality, performance, and total cost of ownership. Traditionally, database research has focused a lot on performance improvements of database systems, but less on new functionality and reducing the total cost of ownership. In this paper, we give our perspective on these three dimensions based on our experience in an industrial research laboratory. The paper is not intended to give a comprehensive overview of all activities, nor is it intended to provide an in-depth discussion of the research work we illustrate. Instead it highlights a set of activities at IBM’s Almaden Research Center and outlines open research challenges that could be tackled by universities in Germany and elsewhere.
With respect to the performance dimension, systems are being designed to be more scalable, by utilizing hardware support to evaluate queries close to the storage subsystem (Netezza), and by massively parallel systems like Google’s map/reduce, which go beyond classical shared-nothing or shared-disk parallelism. That said performance seems to play a less important role in modern database systems as users are willing to trade performance for a reduction of total cost of ownership and – to a lesser extent – for an increase in functionality.
The functionality dimension of database systems has evolved from simple SQL queries to recursive queries, to data mining, and to support for semi-structured data (for example, XML). However, several areas remain insufficiently supported.
We see a future trend to enable support of unstructured information, and for combined storage and querying of semistructured and unstructured data. In Section 1 we introduce AVATAR, a project aimed at discovering and extracting structured information buried in volumes of unstructured text such as emails and blogs. This information can then be exploited to drive the next generation of search and business intelligence applications.
Another emerging trend is the demand for legislative compliance technology. In Section 2 we discuss Hippocratic Database technology as IBM’s approach to tackle the enforcement of privacy regulations or confidentiality policies. Increasing legal regulations regarding privacy of personal data require support for compliance without impeding the flow of information. Whenever business-critical data is involved, companies want to protect the confidentiality of their data and share only on a need-to-know basis according to carefully crafted sharing policies.
Total cost of ownership has become an active research area. As computer systems become cheaper, manpower to maintain those systems is becoming the largest cost factor. In Section 3 we discuss some of the work carried out at IBM in order to make database systems more self-tuning. As a future manifestation of that trend we see that DBMSs will be offered as appliances, with little or no configuration required.
Artikel als Volltext herunterladen*
BibTeX-Eintrag für diesen Artikel
* Zum Ansehen benötigen Sie mindestens den Acrobat-Reader 5.0. Den aktuellen Acrobat-Reader bekommen Sie unter www.adobe.de/products/acrobat/
