[RFC] Potential SQL deprecation

[mpolson64]

In #2974 we added a message that reads "ATTENTION: The Ax team is considering deprecating SQLAlchemy storage. If you are currently using SQLAlchemy storage, please reach out to us via GitHub issues."

The Ax team spends a non-trivial amount of time maintaining support for saving and loading Ax Experiments and their associated structures to SQL databases via SQLAlchemy. However, its not easy for us to tell if this feature is being used by our open source community. In order to help us make the right decisions for Ax please let us know if you use this feature by commenting on this post.

Thank you to everyone in the Ax community for helping to make Ax a successful open source platform.

[pjspereira]

I am attempting to use the SQLAlchemy backend right now, migrating from custom botorch to your service api. This was primarily based on what I estimated would be most efficient for many concurrent experiments with 0.1-3600 seconds sample latency (limited by black box function eval) and 10-100k samples, and some level of persistence as experiments may run for many days. I don’t have any metrics or experience to back the selection though. We use json file storage today, though it is not optimal. Can you compare your backend options for typical experiments? Perhaps just periodically saving json is fine?

[sgbaird]

In my circles, I'm not aware of people using the SQLAlchemy storage.

Perhaps goes without saying, but being able to save to JSON is critical.

Related:

• MySQL vs. MongoDB #1124

[yannik-osphim]

Dear Ax-Team!
We were actually trying to migrate from JSON to SQL (Postgres, in our case), but that seems to be quite difficult. At least we couldn't get it working out of the box. Maintaining the SQL Storage (and maybe even with an emphasis on Postgres) would be greatly appreciated.

[nemozor]

Dear Ax-Team,
We are using the SQL storage feature on one of our machines weekly and plan to expand its use to other machines in the future.
Maintaining this feature would be greatly appreciated on our side as well.

[yannik-osphim]

@nemozor are you using Postgres as well? If so, we would appreciate to get in touch about it!

[lena-kashtelyan]

Hi folks, this is very useful to hear! We will keep the SQL storage around. We were just trying to get a sense of whether anybody needs it : )