Relational databases enable fast, efficient queries by sharing indexes across queries, although fine-grained coordination is necessary to ensure the correctness of updates. In large-scale, massively parallel settings such as stream-processing systems, this fine-grained coordination is currently enabled by replicating indexed state; each query independently maintains computation and operator state. McSherry et al. eliminate this redundancy and demonstrate increased query efficiency through “shared arrangements,” in which indexed views of the maintained state are shared across queries. Shared arrangements expose different versions of the indexed state to different operators based on timestamp, effectively emulating the separate indexes. This approach is particularly applicable in the context of interactive analysis of real-time data.