topic Advice for generic file processing for ingestion of multiple data formats in Data Engineering

Advice for generic file processing for ingestion of multiple data formats

Lea — Tue, 07 May 2024 13:51:43 GMT

Hello,

We are using delta live tables to ingest data from multiple business groups, each with different input file formats and parsing requirements. The input files are ingested from azure blob storage. Right now, we are only servicing three business groups, and for PoC purposes have three different parsing scripts that are tailored to the specific requirements of each group. In the future, we could be servicing hundreds of business groups and do not want to maintain separate parsing scripts for each business group that opts in. Does databricks offer a solution for generic file processing? Additionally, over time the input file for any particular business group may change and their business processes change. Can databricks adapt automatically to changing file formats?

Re: Advice for generic file processing for ingestion of multiple data formats

raphaelblg — Tue, 21 May 2024 17:18:55 GMT

Hello @Lea ,

I'd like to inform you that our platform does not currently provide a built-in feature for ingesting multiple or interchangeable file formats. However, we highly value your input and encourage you to share your ideas through Databricks' Ideas Portal.

Autoloader can discover files to process based on a Glob Pattern: https://docs.databricks.com/en/ingestion/auto-loader/patterns.html#filtering-directories-or-files-using-glob-patterns. Although this is not exactly generic file processing, this feature can help in some cases.

At the moment, Autoloader supports the following file formats: File Format Options
Common patterns: Common Data Loading Patterns
Apache Spark DataFrames and SQL supported formats:Data Format Options

If you have any further questions, please don't hesitate to reach out.