Spark Remember to execute the windowing once with Spark This is one of the things it makes sense when you stop to think about it. When you are performing the same aggregation over a window in more than one column, it is recommended to execute only once that windowing. Lets dive in with an example: I am working with
Spark Obtaining latest PK value Windows vs Joins with spark When working with change data capture data, containing sometimes just the updated cells for a given PK, it is not easy to efficiently obtain the latest value for the entry (row). Let's dive into an example. First though, we need a couple of definitions about change data capture (CDC): Change
Spark Finding latest non-null values in columns Imagine we have a table with a sort of primary key where information is added or updated partially: not all the columns for a key are updated each time, but we now want to have a consolidated view of the information, with just one value of the key containing the
