Below is a table explaining some specifics around how FIDO interacts with each data source it connects to.
Data Source Type | Persistent Connections Supported? | Other Notes |
---|---|---|
Athena | N/A - Athena has no persistent connection support. | All features supported |
BigQuery | N/A - BigQuery has no persistent connection support. | All features supported |
Clickhouse | Yes | All features supported |
Databricks | N/A - Pooled connections maintained but no keepalives used. Warehouse is allowed to sleep when no Explo activity occurs. | All features supported |
Motherduck | Yes | All features supported |
MySQL | Yes | Only 5.7+ supported; percentile calculations only fully supported in 8.0+ |
Postgres | Yes | All features supported (recommend Postgres 12+) |
Redshift | Yes - but possible to disable keepalives for serverless warehouse. Contact support if interested. | All features supported |
Snowflake | N/A - Pooled connections maintained but no keepalives used. Warehouse is allowed to sleep when no Explo activity occurs. | All features supported |
SQL Server | Yes | All features supported |
Supabase | Yes | All features supported; note must use IPv4 to connect or Supavisor Session Mode |
For data source types that support persistent connections, we periodically validate that each connection is still valid.
After one minute since the last successful query on that connection, we will run SELECT 1 as result
as validation. If
this query fails, we will replace that specific connection. Keeping these connections alive and valid helps decrease
query latency, providing a better end-user experience.
For data source types that offer a serverless variety (such as Redshift Serverless), keep alive queries can cost you money because they keep the database awake. We offer the ability to disable keepalives for cases like that. Please reach out to support if you’d like to disable keep alive queries.