Aggregations in SQL, also known as "GROUP BY queries," allow users to summarize large datasets efficiently. This capability is essential for generating reports and identifying trends in data. Cloudflare's recent enhancement to R2 SQL, their serverless SQL query engine, introduces support for these aggregations, enabling users to analyze data stored in the R2 Data Catalog more effectively. With this feature, users can spot anomalies, generate reports, and gain insights from extensive datasets, which is particularly useful for tasks like analyzing sales across various departments. The article explains two main approaches to executing aggregation queries: computing aggregates in advance or on-the-fly. The former requires knowing the aggregate values before executing additional computations, while the latter allows for incremental result building. Cloudflare's R2 SQL uses a technique called "scatter-gather aggregations," where worker nodes compute pre-aggregates and send results to a coordinator node. This method scales well, especially for simple aggregates. However, scatter-gather becomes inefficient for queries that involve sorting or filtering based on aggregation results. For instance, finding the top departments by sales volume requires knowing total sales across all workers, which can lead to missed results if a department's sales are spread thinly across nodes. To address this, the article hints at a more advanced method—shuffling aggregations—that ensures comprehensive data is considered when sorting or filtering, thus improving the accuracy of such queries.