What type of node is recommended to combine two data sources that have a similar set of dimensions?

The recommendation to use a union node when combining two data sources that share a similar set of dimensions is based on how unions operate in data management. A union node merges the rows from both datasets into a single output while retaining all unique rows from each source. This is particularly advantageous when the datasets contain the same dimensional structure but possibly different measures or non-overlapping entries.

When using a union node, you can efficiently consolidate data, allowing for comprehensive analyses that take advantage of the complete dataset without duplicating or omitting any relevant information. It's an ideal approach in cases where the same attributes (dimensions) are relevant across different datasets and you want to analyze a combined dataset as though it were one cohesive unit.

In contrast, other options like inner join, star join, or simply joining would typically be utilized to merge datasets based on common attributes or for specific analytical requirements. However, they are not specifically designed for scenarios where you aim to aggregate similar dimensional data comprehensively without the need for matching rows from each dataset. Therefore, in the context of combining two similar data sources, the union node is the most appropriate choice due to its functionality and flexibility in handling rows from multiple sources.