A demonstration of the MaxStream federated stream processing system

Federated Stream Processing Support for Real-Time Business Intelligence Applications

“…If they do, the new rows are returned, otherwise the task is again suspended. Of course, there are several complexities, including possibilities for concurrency problems which must be handled; details are given in our technical report [5].…”

“…SPEs, by contrast, are push-based. We explored several mechanisms for getting the results back to the clients, including continual polling, periodic selects, database triggers, and adding a subscription mechanism to the client interface, but found none of them adequate for our purposes [5]. All either scaled badly as the number of streams and queries on them grew, or were inefficient.…”

“…In this case, MaxStream's ability to bridge across heterogeneous SPEs and data sources will be invaluable. In [5] we give an example of a simple sales order scenario in which the corporate headquarters gathers statistics for a map leveraging the SPEs at each of the worldwide sites. We expect that as more businesses adopt stream processing engines, this case will become increasingly common, and MaxStream is prepared to handle it.…”

“…It executes those plans on each stream job in the queue, sending them the relevant tuples out of the newly inserted set. For a more detailed description of our ISTREAM implementation, please see our technical report [5].…”

“…MaxStream is designed to provide the necessary support for real-time (or near real-time) business intelligence applications [5,6]. MaxStream combines capabilities for data management, data federation, and stream processing in a single system.…”