Sortide

Sortide is a term used in information science to describe a family of sorting techniques that handle multi-key or multi-criteria ordering of items. The central idea is to partition items by one or more keys and then sort within partitions, with an emphasis on locality of reference and parallelism to improve performance on modern hardware.

Origin and scope: Sortide emerged in discussions of data processing and database indexing to address performance bottlenecks in multi-attribute sorting. It is not a single algorithm but a class of strategies that share the principle of combining coarse partitioning with fine-grained sorts.

Characteristics: The approach can be stable, preserving input order among equal keys, and can be designed for external sorting when data do not fit in memory. It often uses a two-phase process: partitioning by one or more keys into buckets, followed by sorting within each bucket. Some implementations leverage cache-friendly layouts and parallel processing to reduce memory latency.

Complexity and performance: Average-case time complexity depends on the specific method; typical in-memory variants achieve O(n log n) or near-linear time with multikey radix-like variants, while external variants focus on minimizing disk I/O. The practical performance is highly data-dependent.

Applications: Sortide techniques are used in databases for multi-attribute ordering, data warehousing, catalog sorting in e-commerce, and analytics pipelines where items have multiple sortable attributes. See also sorting algorithm, multikey sort, and external sort for related concepts.