Welcome to Apache OpenDAL | Apache OpenDAL

OpenDAL represents Open Data Access Layer. Our imaginative and prescient is to entry knowledge freely.

What OpenDAL does?​

Getting began​

See the web page for fast begin with a number of languages: Quickstart.

Why OpenDAL?​

The imaginative and prescient of OpenDAL is entry knowledge freely, the place “free” refers to 4 important features:

1. Free from providers​

OpenDAL should allow customers to entry numerous storage providers starting from s3 to dropbox through its personal native API. It ought to present a unified API for accessing all these providers.

For instance, we DO

whereas we DO NOT

• Add help for Google Cloud Storage(GCS) through XML API: GCS has native JSON API which extra highly effective
• Add help for structural knowledge in MySQL/PostgreSQL: We are able to deal with a database as a easy key-value retailer, however we won’t help unified entry of structural knowledge.

2. Free from implementations​

OpenDAL must separate the assorted implementation particulars of providers and allows customers to put in writing equivalent logic for various providers.

For instance, we DO

• Add a brand new functionality to point whether or not presign is supported: Customers can now write logic based mostly on the can_presign possibility.
• Add a default_storage_class configuration for the S3 service: Configuration is particular to the S3 service.
• Add an possibility for content_type within the write operation: It aligns with HTTP requirements.

whereas we DO NOT

• Add a brand new possibility in learn for storage_class: As totally different providers might have various values for this parameter.

3. Free to combine​

OpenDAL must be built-in with totally different methods.

For instance, we DO

• Add Python binding: Python programmers can use OpenDAL.
• Add object_store integration: object_store customers can undertake OpenDAL.

4. Free to zero value​

OpenDAL must implement options in zero value manner which suggests:

• Customers want to not pay value for unused options.
• Customers can not write higher implementation for used options.

For instance, we DO

• Add layer help: Customers can add logging/metrics/tracing in zero value manner.
• Implement search for Reader: Customers can not write higher search help, all of them must pay the identical value.

we DO NOT

• Add Arc for metadata: Customers could solely want to make use of metadata as soon as and by no means clone it. For individuals who do need this characteristic, they will add Arc themselves.