Limits of Pushdown

There are limiting factors to the ability to push down queries. If there are multiple data sources connected on CData Virtuality, a query spanning multiple databases can not be pushed down completely, so some operations will have to be executed in CData Virtuality’s memory (or, to be more exact, in the buffer space, which will be discussed below)

As an example, two different databases of the same Microsoft SQL server can be joined when querying it directly. If a different connection represents each database, the performance will be mitigated because CData Virtuality interprets them as different sources. A way to work around this behaviour is to create a view on the Microsoft SQL Server that joins them already on the Microsoft SQL Server side.

For databases like Snowflake and Microsoft SQL Server, CData Virtuality supports having different catalogs and different databases in the same datasource. For more details, please refer to the usage of the importer.catalog property for Snowflake and Microsoft SQL Server.