Storage backends¶

AutoCRUD separates metadata and resource data storage via IStorage / IStorageFactory.

A storage factory is configured at AutoCRUD(...) construction time or via:

crud.configure(storage_factory=...)

Per model, you can also pass an explicit storage backend to add_model().

This allows different resources to use different storage systems.

Options¶

DiskStorageFactory (minimal viable production)¶

Stores everything locally.

Component	Backend
meta	SQLite file
revision	filesystem files
blob	filesystem

Example:

from autocrud import AutoCRUD
from autocrud.storage.disk import DiskStorageFactory

crud = AutoCRUD(
    storage_factory=DiskStorageFactory("./data")
)

Best for:

local development
single-node deployments
small to medium systems

Pros:

zero infrastructure
simple backups
easy debugging

Cons:

not horizontally scalable
limited concurrent writers

S3StorageFactory¶

Stores revisions and blobs in object storage (S3 compatible).

Component	Backend
meta	SQLite stored in S3
revision	S3 objects
blob	S3

Example:

from autocrud.storage.s3 import S3StorageFactory

crud = AutoCRUD(
    storage_factory=S3StorageFactory(
        bucket="my-bucket",
        endpoint_url="https://s3.amazonaws.com"
    )
)

Best for:

cloud deployments
large datasets
multi-node services

Pros:

highly scalable
cheap storage
cloud-native

Cons:

metadata queries slower than SQL
requires object storage infrastructure

PostgresStorageFactory¶

Stores metadata in PostgreSQL while keeping large data in S3.

Component	Backend
meta	PostgreSQL table
revision	S3 objects
blob	S3

Example:

from autocrud.storage.postgres import PostgresStorageFactory

crud = AutoCRUD(
    storage_factory=PostgresStorageFactory(
        postgres_url="postgresql://user:pass@host/db",
        bucket="my-bucket"
    )
)

Best for:

production systems
heavy query workloads
large datasets with indexing

Pros:

fast queries
scalable storage
powerful indexing

Cons:

requires database infrastructure
slightly more complex setup

MemoryStorageFactory¶

Stores everything in memory.

Component	Backend
meta	in-memory dict
revision	in-memory dict
blob	memory

Example:

from autocrud.storage.memory import MemoryStorageFactory

crud = AutoCRUD(
    storage_factory=MemoryStorageFactory()
)

Best for:

testing
unit tests
demos

⚠️ Data is lost when the process exits.

Choosing a backend¶

Use case	Recommended backend
unit tests	`MemoryStorageFactory`
local development	`DiskStorageFactory`
simple production	`DiskStorageFactory`
cloud storage	`S3StorageFactory`
large-scale production	`PostgresStorageFactory`

Rule of thumb:

small project → Disk
cloud system → S3
large system → Postgres + S3

Per-model override¶

Different resources can use different storage backends.

Example:

crud = AutoCRUD(
    storage_factory=DiskStorageFactory("./data")
)

crud.add_model(User)

crud.add_model(
    Image,
    storage=S3StorageFactory(bucket="image-bucket")
)

This allows:

hot data in Postgres
large data in S3
local data on disk

All resources still share the same ResourceManager API.