Mosaic: An Architecture for Linking Databases and Scalable Interactive Visualizations

Real-time interaction and visualization over large data volumes requires careful coordination of data queries and visual updates. Mosaic is an architecture for optimizing scalable and interoperable visualizations backed by a database, providing a platform for developing and deploying optimizations that span both visualization clients and backing databases. Mosaic applications consist of data-consuming clients that publish data needs as declarative queries, parameterized by shared filtering selections. These queries are managed and automatically optimized by a coordinator that proxies access to a scalable data store. For example, by analyzing selection predicates and client queries, the coordinator automatically constructs materialized views to perform selection updates over pre-aggregated data at interactive rates. Given only a high-level specification, Mosaic automatically enables orders-of-magnitude performance improvements over standard update queries.