IDEBench

IDEBench是一个专为评估数据库系统在交互式数据探索（IDE）工作负载下的性能而设计的基准。它不同于传统的分析数据库系统基准，如TPC-DS和TPC-H，后者主要关注静态报告场景。IDEBench旨在提供更接近实际数据探索工作流程的负载和数据集，特别关注查询性能与结果质量之间的权衡。该基准包括一个数据生成器，能够将任何种子数据集扩展到任意大小，并保持数据分布和属性间的关系。IDEBench主要关注大型数据集上的聚合查询，这些查询是增量构建的，并且结果在可视化前端进行交叉过滤。此外，IDEBench还模拟了用户在IDE前端常见的交互模式，如独立浏览、顺序链接、1:N链接和N:1链接，以更真实地反映用户行为。

IDEBench is a benchmark specifically designed for evaluating the performance of database systems under interactive data exploration (IDE) workloads. In contrast to traditional analytical database benchmarks such as TPC-DS and TPC-H, which primarily focus on static reporting scenarios, IDEBench aims to provide workloads and datasets that align more closely with real-world data exploration workflows, with a particular emphasis on the trade-off between query performance and result quality. This benchmark includes a data generator capable of scaling any seed dataset to arbitrary sizes while preserving data distributions and inter-attribute relationships. IDEBench primarily focuses on aggregation queries over large datasets, which are incrementally constructed and whose results are cross-filtered in visualization frontends. Additionally, IDEBench simulates common user interaction patterns in IDE frontends, including independent browsing, sequential linking, 1:N linking, and N:1 linking, to more authentically reflect actual user behaviors.