Observe Performance Cookbook: Type Data Columns¶
Problem¶
There are slow queries related to data with a column extracted from a JSON column.
Solution¶
Use the Column header to convert variant columns to specific types such as string or int64. In OPAL, examples of applicable functions are string
, int64
, and parse_timestamp
.
Explanation¶
A field extracted from a JSON column has the type of variant
. Processing variant
values can be significantly more expensive in query execution and transform execution. Observe recommends casting fields that are extracted from a JSON column to an explicit type (for instance, use make_col cluster_uid:string(FIELDS.cluster_uid)
instead of make_col cluster_uid:FIELDS.cluster_uid
).