How does mysql-python compare to pymssql and pyodbc?

[cdeil]

So far I was using pymssql and pyodbc to connect to MSSQL and Azure SQL from Python. Both mostly work OK with some issues, and it was never clear to me which one was better or recommended.

Now with mysql-python you create a new one which I presume will be different and better than the existing two solutions in some ways. Could you please add a FAQ or docs entry explaining briefly from end-user perspective how it compares to the existing packages?

[perrysk-msft]

Thank you for your feedback! We understand that choosing the right SQL driver can be confusing, especially with multiple options available. Our goal with the new mssql-python driver is to provide a modern, well-supported, and performant solution for connecting to Microsoft SQL Server and Azure SQL from Python.

Specifically, we aim to bring improved reliability, better async support, and a more streamlined developer experience. Here is our roadmap: https://github.com/microsoft/mssql-python/blob/alpha/ROADMAP.md

In the meantime, if there are specific aspects you’re curious about, feel free to ask—we’d love to hear what’s most important to you!

[gustavoleo-art]

new article about the benchmark
https://devblogs.microsoft.com/python/mssql-python-vs-pyodbc-benchmarking-sql-server-performance/

Im curious to see the benchmark with connector-x on select
https://github.com/sfu-db/connector-x

[cdeil]

Great. Thanks devs who work on mysql-python and for putting out the blog post. Congratulations!

The first benchmark with SELECT having such a large variance with mysql-python is curious (benchmark artifact or something that can be improved?).

And yes, would be interesting to extend the benchmarks to cover analytics use cases where the goal is usually to get DB <-> Arrow memory for many MB or even few GB of data which is what polars, pandas, duckdb etc use.