Databricks Community

wcohen · Wednesday

Hi there - I'm a professor at CMU teaching a course on scalable ML, and the last few semesters we have run assignments using the DataBricks Community Edition, which is no longer open for new accounts. The free accounts aren't a great substitute for us since the assignments are based on on RDD API, which isn't supported by the free accounts. The deprecation of community accounts took us a bit by surprise - is there any way to open these up for a hundred-odd students for a few more weeks?

szymon_dybczak · Wednesday

Hi @wcohen ,

There was kind of similar question some time ago. It seems that access to existing Community Edition accounts will remain available for the rest of the year, but new users attempting to sign up for Community Edition will be redirected to the Free Edition instead.

Solved: Community (Legacy) Edition Question - Databricks Community - 123948

wcohen · Thursday

Thanks for the answer! But unfortunately that doesn't solve our problem - the students in this fall's don't have community accounts, so they would be sent to the free accounts, which don't support RDDs.

ilir_nuredini · Thursday

Hello @wcohen ,

As mentioned by @szymon_dybczak , the community edition one will be available only for those that are already signed up. What I would suggest is to reach out directly to the Databricks support/representative to see if there is any workaround, as I think as community we would be out of options.

I hope you will get this sorted out.

Best, Ilir

szymon_dybczak · Thursday

So you can try to talk with databricks representatives (but since the end of support is planned for quite some time, I doubt they will agree to extended this) or as a workaround you can just use docker images with preinstalled pyspark. The setup is super quick and you will basically have the same ability as with Databricks Community edition.

BS_THE_ANALYST · Thursday

That's a great idea with Docker @szymon_dybczak although, I suppose from a uni student's perspective, if they have to configure anything, that could end in a disaster 😆. I did a maths degree at uni and thought the command prompt was part of the matrix 🤣. Could just be me projecting haha.

@wcohen whenever I've used Labs, provided by Databricks, they've been hosted through Vocareum. I wonder if that's a potential route for you but I imagine there's some cost involved. I mention this as it'd require less governance than an actual databricks environment (I'm sure your IT department would appreciate that) and I'd have thoughts there's caps/restrictions in place to reduce costs (or potentially fixed prices).

@wcohen please keep us in the loop, I'm interested to hear about the resolution. I guess the RDD restriction is based on Serverless compute, if the free edition makes classic compute available at some stage, that could solve your problem. I'm not sure if that's on the roadmap for the free edition, though.

All the best,
BS