Databricks ODBC driver tooks long time to list columns

[Youngwb]

I'm testing the performance of databricks, when I use ODBC driver to submit query, I found it slower than notebook,and it looks like odbc driver will send "list columns" request to databricks, 

1. Is there a way to prevent the ODBC driver from sending these requests?

2. Are these requests necessary?

[Kaniz_Fatma]

Hi @Youngwb,  

• Unfortunately, there isn’t a direct way to prevent the ODBC driver from sending “list columns” requests. These requests are typically part of the driver’s metadata retrieval process, which helps ensure that the query execution environment is correctly set up.
• These “list columns” requests are generally necessary for the ODBC driver to understand the schema of the tables you’re querying. This helps optimize the query execution and ensure that the results are correctly formatted. However, they can add some overhead, especially if the schema is large or complex.

Although you can't directly prevent these requests, they help optimize query execution and format results accurately, albeit with some overhead. To boost performance, ensure you're using the latest ODBC driver version, simplify your queries, optimize table indexing and partitioning, and consider alternative querying methods like Databricks Notebooks or the JDBC driver if performance issues persist.