B-trees too but also bucketing for formats like delta lake or iceberg, where having ids that cluster will reduce the number of files you need to update.