WHEN
August 25,2005
WHAT
Numerous new applications arising in finance, telecommunications, network management & security, sensor networks, RFID, military, etc. require an ability to process, filter, aggregate, and correlate large numbers of messages (from thousands to hundreds of thousands per second) with very low latency (from milliseconds to seconds). These so called data stream processing application cannot rely on traditional databases, as databases were never designed for these kinds of requirements. A new class of infrastructure, data stream processing engines, has been proposed to help solve this problem. In this talk we'll discuss some data streaming applications, why they cannot rely on traditional databases, the problems arising in processing data streams, and the different approaches to solving this problem, both from academia and from the industry.
WHO
PRESENTATION