This builds on top of this issue to capture additional workload insights and details in our telemetry pipeline. The previous format for query events is outlined in the following spreadsheet.

Additional information to capture:

• HasIndexJoin (we have HasFullTableScan today) (not available)
• • index_join_count field
• Contention (wait time) (not available)
• Statement Fingerprint ID (measuring uniqueness of a workload) (not available)
• Database ID (not available)
• • complications with adding Database ID, documented in: sql: Add database ID to sampled query log #84195
• Database name
• Session ID
• Transaction ID (measure how large/conversational transactions are)
• Statement ID (Avoid having to filter based on the query text)
• Plan from the plan gist (determine the complexity and how optimal the workload is)
• Whether the query touched all cluster nodes (related issue) (not available)
• Index recommendation (not available)

Jira issue: CRDB-10514

Epic CRDB-13538