A growing number of applications require the ability to analyse massive amounts of streaming data in real time. Examples of such applications are: market data processing, anti-spam and anti-virus filters for e-mail, network security systems for incoming IP traffic in organisation-wide networks, automatic trading, fraud detection for cellular telephony to analyse and correlate phone calls, fraud detection for credit cards, and e-services for verifying SLAs. Typically, such applications require strong analysis and processing capabilities, i.e., data mining, to discover facts of interest. Data analysis happens today on clusters of workstations using specialized middleware and applications. Although solutions for real-time processing of information flows already exist, current platforms and infrastructures phase three main limitations:

(a) scalability, (b) autonomy, and (c) performance.

STREAM aims at scaling system size by an order of magnitude, to 100s of nodes, achieving real-time processing of information flows, and providing unsupervised and autonomous operation. This will allow for much broader deployment of such products and services to new areas that need to manipulate large information flows in a cost-effective manner, and in particular, the Telecom, Financial, and E-services sectors.