Additions/suggestions to improve the Superset DB query table

[GerbenvdHuizen]

The Superset DB contains a table with records of historic query runs simply called query. The query table currently contains the following columns relevant to query runtime:

• status
• start_time
• end_time
• start_running_time
• end_results_backend_time

Although there already are many columns that help infer the runtime of a query, I found that some small additions could improve the analytical value of this table greatly. I therefore suggest implementing the following additions:

• Although query duration can technically be calculated with the already available columns I think this would be a simple improvement that provides immediate value. I therefore suggest adding a duration column to the query table.
• Currently when a query fails the end_time field will contain a null value. I suggest adding the timestamp of when the query ended/failed instead of the null value so this can be used for analytical/debug purposes. I'm not sure if the null value for end_time on failure is a bug.
• I suggest adding differentiation between failed and timed-out queries within the status column. A timed-out status would already provide a lot more insight on why a query failed while still being different from a query failure status due to for example a parsing exception.