Snowflake Query Tags

urn:js:virtue:aspire:proposal:7.1

TL;DR

Query Tags in Snowflake MUST conform to the Snowflake Query Tag Schema.

Rational

Query Tags (QUERY_TAG) in Snowflake, set using ALTER SESSION SET QUERY_TAG='' MUST conform to the Snowflake Query Tag Schema Snowflake Query Tag Schema. Tags MUST NOT not be used that don’t conform to the schema.

• Snowflake Schemas
• Snowflake Query Tag Schema

Background

Snowflake provides the ability to apply a string in the ‘QUERY_TAG’ column to all queries executed within a Snowflake session. By formatting this string as a JSON blob, it is possible to more easily analyse the Snowflake Query History.

Benefits

• A common format for the QUERY_TAG allows it to be easily processed and identified. Where different JSON schemas are utilised processing of the column becomes more difficult.
• Queries issued in different session can be easily linked together to help with debugging e.g.

    select
        CHECK_JSON(query_tag) as json_error,
        parse_json(query_tag):correlation_id::string as correlation_id,
        *
    from
        snowflake.account_usage.query_history
    where
        (query_tag != '') and
        (json_error is null) and
        (user_name = 'USER_TO_FILTER_ON') and
        (correlation_id = '7b3128bd-8d89-d1e5-174a-469634a64002')
    ;
  

• Easy to determine whether teams are conforming to the standard by validating QUERY_TAG column entries against the schema.

Impact / Downsides

• Teams already using the QUERY_TAG feature will need to alter their code to conform to the new standard.
  • Few users are currently using the QUERY_TAG field.
• It is not possible to prevent users from apply invalid tags, it is only possible to monitor for compliance and engage with those that do not comply.

Implications

None.