UpStream: Storagecentric Load Management for Data Stream Processing Systems
Processing fast updating data streams in real-time must reflect the most recent data. A number of technologies including data stream management systems have emerged to respond to this challenge. While running their queries in a continuous fashion on high-volume push-based data streams (e.g. sensor data, GPS coordinates and stock quotes), one of the most important optimization problems that these systems face is load management based on Quality of Service (QoS). In stream processing, QoS may depend on a number of factors like latency, freshness, quality and completeness of the results.