I am strongly for it. I wanted to raise the same proposal after we see the results of the survey (which I hope we will run after the Summit) - in order to get more data points on how many users use it. But I agree the datapoints we have support the "low usage, high trouble" status of it. MSSQL creates a lot of troubles, it has bitten us more than once.
One comment though: I think an important point of this removal should be some viable way for our users to move out to (likely) Postgres. We should provide at least a description, ideally some scripts that will be able to move data from MSSQL to Postgres as part of this "support dropping". I hate to say to anyone "sorry, no more support, deal with it yourself". But saying: "sorry, no more support, here is a way how you can migrate to what we support" is a much better message . J. J. On Fri, Aug 18, 2023 at 2:06 PM Elad Kalif <elad...@apache.org> wrote: > Hello everyone, > > In May 2021 we added MsSQL as an experimental backend. PR > https://github.com/apache/airflow/pull/9973 > Mailing list announcement: > https://lists.apache.org/thread/byx4czfyowwm3qy8c8p2o65vd53fyvsd > > During this time we hoped it would become stable and widely adopted. > To my taste MsSQL a backend has left a niche and is *not* worth the > maintenance of it in our CI. > > Some data points: > > 1. Only 5 issues > < > https://github.com/apache/airflow/issues?q=is%3Aissue+label%3Abackend-mssql-experimintal+is%3Aopen > > > were > reported on MsSQL as a backend which suggests it's not widely adopted. > These issues also don't have much traffic. > 2. None of the Airflow as service cloud vendors use it as a backend or > expressed interest in it (to the best of my knowledge) > 3. We had many PRs trying to fix issues related to MsSQL. To name a few: > https://github.com/apache/airflow/pull/33265 > https://github.com/apache/airflow/pull/32433 > https://github.com/apache/airflow/pull/31815 > each one required diagnostic and effort to address the problems (in some > cases required also rebasing of all other PRs after fix was merged) > > > I think that after 2+ years of experiment we can terminate it. > Since this is experimental we can drop support for this feature as this is > not considered a breaking change. > > *My Proposal:* > Drop support for MsSQL as official/supported backend. > > WDYT? > > Elad Kalif >
