ce104b6a9c
* Add snapshot creation command * Read and start diffing snapshot * Add apply snapshot functionality * Fix cli invocation * Add log messages * Fix duplicated if check * Add (minimal) docs on schema migrations * Fix missing import * Update api/src/utils/apply-snapshot.ts Co-authored-by: Nicola Krumschmidt <nicola.krumschmidt@freenet.de> * Appease to Nicola's programming professor Co-authored-by: Nicola Krumschmidt <nicola.krumschmidt@freenet.de>
2.0 KiB
Databases
The Directus platform is primarily comprised of an API and App, working in concert to “mirror” the schema and content of your SQL database.
Database Mirroring
Instead of using a predefined “one-size-fits-all” data model to store your content, Directus “mirrors” your actual SQL database in real-time. The principle is akin to a database client (like phpMyAdmin), but includes far more advanced tools, and is safe and intuitive enough for non-technical users. This approach has many unique advantages:
- A custom (pure) SQL database schema, tailored to your exact requirements
- Significant performance improvements through optimizations and indexing
- Complete transparency, portability, and security for your data
- Direct database access and the full power of raw/complex SQL queries
- Allows importing existing databases, unaltered and without any migrations
Dynamic API
The Directus API uses Database Mirroring to dynamically generate REST endpoints and a GraphQL schema based on the connected database's architecture. It is written in Node.js and uses database abstraction to support most SQL database vendors.
Relevant Docs
Database Abstraction
Directus supports mirroring all the most widely used SQL databases. There are many different SQL database vendors, including popular choices such as MySQL, PostgreSQL, and SQLite. Each vendor has subtle (and sometimes not so subtle) differences in how they function, so Directus includes an abstraction layer that helps it avoid writing different code for each different type.
This means there is the possiblility of supporting other datastores in the future, such as NoSQL options like MongoDB, or even third-party data services like Firebase or Heroku. However these options are fundamentally different from the relational SQL databases we currently support, and so more research is needed.