Guide and Best Practices: Moving from Community Edition to Free Edition
Introduction
This guide is for educators moving their course materials from Databricks Community Edition (CE) to the new Databricks Free Edition (FE). It covers the key differences, known limitations, and practical workarounds to give your students a smooth upgrade experience.
Kindly note that CE will be discontinued over the coming months!
If you encounter issues not listed, please contact free_edition_help@databricks.com
What is new in Free Edition?
- Completely free: no AWS costs or paying for your own compute, and no credit card required
- Modern feature set: includes MLflow, ingestion tools, Jobs, Delta Live Tables, dashboards, model serving, evaluation, semantic search, Unity Catalog, Assistant and more
- Serverless compute: Free Edition manages clusters for students so they can focus on learning, not managing infrastructure
- World-class reliability: the same infrastructure trusted by leading companies for critical workloads
Common issues and resolutions
|
Issue (symptom)
|
Why it happens
|
How to resolve / workaround
|
Docs
|
|
Cannot create or read files in /dbfs/
|
Public access to the DBFS root file is disabled in modern Databricks workspaces for security
|
Store datasets in a Unity Catalog Volume
|
|
|
dbutils.fs.mount commands fail
|
Modern Databricks workspaces do not support legacy DBFS mounts
|
Register an external location in Unity Catalog, then read or write through a volume path (/Volumes/<catalog>/<schema>/<volume>/…)
|
External Locations
|
|
RDD-based code does not work
|
The Apache Spark team recommends using DataFrames, a more modern and performant abstraction than RDDs. As such, Free Edition only supports DataFrames & Datasets
|
Refactor notebooks to use DataFrame APIs. Databricks Assistant can auto-suggest replacements for common RDD patterns
|
Apache Spark Docs,
Databricks Assistant
|
|
Scala or R code does not work
|
Serverless runtime only runs Spark, Python and SQL. Scala is on the roadmap but not currently supported
|
If Scala is essential, contact us at free_edition_help@databricks.com . We are aiming to support Scala on Free Edition at a later date
|
Serverless Compute Limitations
|
|
Unable to choose classic cluster sizes or instance types
|
Free Edition supplies serverless compute only; BYOC and classic clusters are removed to keep the tier completely free and let students focus on learning, not managing infrastructure
|
Let students run notebooks on serverless (auto-scales with no set-up). If any issues contact us at free_edition_help@databricks.com
|
Connect to Serverless Compute
|
|
GPUs do not work
|
GPUs are not available in Free Edition’s serverless fleet
|
If GPU workloads are essential contact us at free_edition_help@databricks.com for paid options
|
Free Edition Limitations
|
|
Job or Lakeflow pipeline count exceeded
|
FE caps at 5 concurrent job tasks and one active Lakeflow pipeline of each type
|
Combine tasks where possible, or stagger pipeline runs. Pause jobs and pipelines when not in use
|
Free Edition Limitations
|
|
Model Serving / Vector Search endpoints scale poorly
|
FE limits each account to two concurrent endpoints, with no provisioned throughput
|
Keep demos lightweight (e.g. one RAG application). For production-like serving contact us at free_edition_help@databricks.com for paid options
|
Free Edition Limitations
|
|
Enterprise admin features (SSO, SCIM, private networking) missing
|
FE is designed for learning, not enterprise operations; account-level APIs and SLAs are therefore not supported
|
For enterprise projects, contact us at free_edition_help@databricks.com for paid option
|
Free Edition Limitations
|
We strongly recommend moving your course to Free Edition as soon as possible as Community Edition will be discontinued over the coming months. Please contact us at free_edition_help@databricks.com if you experience any issues.
Migration Tips
- Copy notebooks: Export from Community Edition and upload them to Free Edition
- Adjust data paths: Use volume paths and external storage instead of dbfs:/ or mounts
- Test and iterate: Run all your sample code early to catch incompatibilities
- Use Databricks Assistant: Use the coding co-pilot to rewrite broken code like RDDs in bulk
- Individual Accounts: Students should work within their own account where possible. Users can share an account, but this should only be done for collaborating on a group project
- Register with an academic email address: Please ensure that students register for a Free Edition account using their academic/student email address for full access (e.g. xxx.edu; xxx.ac.yy etc.) .
We’re excited to support you teaching on the new Databricks Free Edition. Please contact us at free_edition_help@databricks.com if you experience any issues not addressed in the documentations provided/referenced.
