Adaptive data push/pull for the DejaVu pattern matching system
Patrick Lau · Repository for Publications and Research Data (ETH Zurich) · 2009
In traditional database systems where the query processor operates by "pulling" data, a different approach of input handling is needed to deal with real-time requirements complex event processing (CEP) applications do request.Thus CEP Systems, which are built over a traditional database framework such as DejaVu [11], need to support a "pushing" mechanism where incoming data is being pushed to and processed by the query processor on-the-fly.Since the data rate may be extremely high or bursty, switching from "Push" to "Pull" reduces the "pressure" for the query processor.The goal of this thesis is to provide such a "pushing" mechanism for DejaVu and apply the switching between those two modes in an adaptive way. Pull vs.