Merge data use case

This article outlines a use case in which you merge data from two sources together to create a final output.

A very common use case is merging data from a data stream (A) with a reference data source (B). This requires you to use a field to lookup (and merge) data from the reference data source (B) with the data stream (A) and then save the resulting data in the output.

This can be achieved by adding a lookup to the output and specifying the matching key. Any data brought in from the data source (B) comes in via an aggregation in the lookup.

For example, you can use the LAST aggregation function to select the last value of a field within a time window. This can also be used to look up user information that is stored in a user data table—that is, given the USER_ID, look up the USER_DETAIL, USER_DETAIL2, and so on.

Required Level of Expertise

• Intermediate

Add a lookup

1. On the Outputs page, select the desired output and click Edit (or create a new output).

2. Click Add Lookup and select the lookup source (e.g. an existing data source or lookup table). Then click Next.

3. In Join Fields, select the field to join to in data source (B) (e.g.USER_ID). The matching field in data stream (A) is selected automatically.

Note: The field on the left is the field in data source (B) that we are looking up from. The field on the right is the key from data stream (A).

4. Add aggregations; for example:

• LAST(USER_ID)

• LAST(USER_DETAIL1)

• LAST(USER_DETAIL2)

5. Set the Time Frame to infinite, by entering -10000 in the From field. This ensures you can access all the data irrespective of when it was written.

6. Select Use Full History to treats the data source (B) as a reference data table. This is useful if the latest view of the data is required, and the actual time that the data was written is irrelevant.

7. Enter an Alias for the lookup and click Save.

Toggle from the UI to SQL view to see the JOIN statement (it can appear as part of a WITH statement or inline).

8. Click Run and fill in the required details to deploy the output. See: Run an output