According to me Data blending comes into picture in few cases only like
1. Granularity of data is different in two data sources.
2. Crossjoins r not supported.
3. Joins causes duplicates instead we can use blending and blend on a dimension.
4. You can go for this we want to do data cleaning,
5. Volume of data is huge
Crossjoins:
Crossjoins allows you to join data from two different data sources as if they are
in same data base.
2. Blends are a post aggregate left join. That means that the grouping and
aggregation is done on the rift side befor it is combined with the left side. Where
are joins r not.
I hope this helps