We propose to demonstrate a practical alternative approach to the current state-of-the-art query processing techniques, called the “Query Mesh” (or QM, for short). The main idea of QM is to compute multiple routes (i.e., query plans)1, each designed for a particular subset of data with distinct statistical properties. Based on the execution routes and the data characteristics, a classifier model is induced and is used to partition new data tuples to assign the best routes for their processing. We propose to demonstrate the QM framework in the streaming context using our demo application, called the “Ubi-City”. We will illustrate the innovative features of QM, including: the QM optimization with the integrated machine learning component, the QM execution using the efficient “Self-Routing Fabric” infrastructure, and finally, the QM adaptive component that performs the online adaptation of QM with near-zero runtime overhead.
Query Mesh, QM, query processing techniques, multiple routes, classifier model, streaming context
Date of this Version