Databricks offers a variety of ways to help you load data into a lakehouse backed by Delta Lake. Databricks recommends using Auto Loader for incremental data ingestion from cloud object storage. The add data UI provides a number of options for quickly uploading local files or connecting to external data sources.
If you haven’t used Auto Loader on Databricks, start with a tutorial. See Run your first ETL workload on Databricks.
Auto Loader incrementally and efficiently processes new data files as they arrive in cloud storage without additional setup. Auto Loader provides a Structured Streaming source called
cloudFiles. Given an input directory path on the cloud file storage, the
cloudFiles source automatically processes new files as they arrive, with the option of also processing existing files in that directory.
You can simplify deployment of scalable, incremental ingestion infrastructure with Auto Loader and Delta Live Tables. Note that Delta Live Tables does not use the standard interactive execution found in notebooks, instead emphasizing deployment of infrastructure ready for production.
You can securely upload local data files or ingest data from external sources to create tables. See Load data using the add data UI.
Databricks partner integrations enable you to load data into Databricks. These integrations enable low-code, scalable data ingestion from a variety of sources into Databricks. See Databricks integrations.
COPY INTO allows SQL users to idempotently and incrementally load data from cloud object storage into Delta Lake tables. It can be used in Databricks SQL, notebooks, and Databricks Jobs.
Databricks provides a single command to convert Parquet or Iceberg tables to Delta Lake and unlock the full functionality of the lakehouse; see Convert to Delta Lake.
Here are a few things to consider when choosing between Auto Loader and COPY INTO:
If you’re going to ingest files in the order of thousands, you can use
COPY INTO. If you are expecting files in the order of millions or more over time, use Auto Loader. Auto Loader requires fewer total operations to discover files compared to COPY INTO and can split the processing into multiple batches, meaning that Auto Loader is less expensive and more efficient at scale.
If your data schema is going to evolve frequently, Auto Loader provides better primitives around schema inference and evolution. See Configure schema inference and evolution in Auto Loader for more details.
Loading a subset of re-uploaded files can be a bit easier to manage with COPY INTO. With Auto Loader, it’s harder to reprocess a select subset of files. However, you can use COPY INTO to reload the subset of files while an Auto Loader stream is running simultaneously.
For a brief overview and demonstration of Auto Loader, as well as COPY INTO, watch this YouTube video (2 minutes).
The Data Science & Engineering workspace Data tab allows you to use the UI to load small files to create tables; see Explore and create tables in DBFS.
You can connect to a variety of data sources using Apache Spark. See Interact with external data on Databricks for a list of options and examples for connecting.
Apache Spark automatically captures data about source files during data loading. Databricks lets you access this data with the File metadata column.