Database Providers

Selected Provider: You are currently viewing documentation targeting Sqlite. The specifications, connection string formats, and type mappings below are aligned with your selection.

AN5 ORM supports relational databases, Google Sheets serverless tables, and 6 dedicated vector database engines.


Quick Comparison

Provider Protocol Dialect Pagination Parameters Vector Search CLI Push / Migrate
PostgreSQL postgres://, postgresql:// Postgres SQL LIMIT / OFFSET %s, $1 Native via pgvector db:push
SQL Server sqlserver:// T-SQL OFFSET ... FETCH @p_0, ? Native (2025) / In-memory db:push, db:pull, db:migrate
MySQL mysql://, mariadb:// MySQL / MariaDB LIMIT / OFFSET ? In-memory fallback db:push
SQLite sqlite://, .db, .sqlite SQLite3 LIMIT / OFFSET ? Native (sqlite-vec / built-in distance functions / json_each) db:push
Google Sheets googlesheets:// In-memory query Array slicing N/A Not supported Serverless sync
Vector DBs (NBase, Qdrant, Chroma, Pinecone, Weaviate, Milvus) nbase://, qdrant://, chroma://, pinecone://, weaviate://, milvus:// Vector Engine Vector topK N/A Native Vector Store / ANN Vector indexing

1. PostgreSQL

Connection String

DATABASE_URL=postgres://user:password@localhost:5432/dbname?sslmode=disable
# or
DATABASE_URL=postgresql://user:password@localhost:5432/dbname

Features & Capabilities

  • Dialect: Standard ANSI SQL with PostgreSQL extensions.
  • Pagination: Uses LIMIT {take} OFFSET {skip}.
  • Parameters: Placeholder %s in Python, $1, $2 in native drivers.
  • Types: Supports INTEGER, BIGINT, TEXT, VARCHAR, BOOLEAN, TIMESTAMP WITH TIME ZONE, JSONB, BYTEA.
  • Vector Search: Native integration with pgvector extension. Cosine distance (<=>), L2 distance (<->), and inner product (<#>).

2. SQL Server (MSSQL)

Connection String

DATABASE_URL=sqlserver://localhost:1433;database=mydb;user=sa;password=YourPassword123!;trustServerCertificate=true

Features & Capabilities

  • Dialect: Microsoft Transact-SQL (T-SQL).
  • Pagination: Uses ORDER BY ... OFFSET {skip} ROWS FETCH NEXT {take} ROWS ONLY.
  • Parameters: Placeholders @p_0, @p_1 or ?.
  • Types: Supports INT, BIGINT, NVARCHAR, VARCHAR, BIT, DATETIME2, VARBINARY.
  • Schema Tools: Full support for db:push, db:pull, db:migrate:*, and schema catalog reflection via sys.*.
  • Vector Search: Native support in SQL Server 2025 via VECTOR_DISTANCE(), with automatic fallback to in-memory cosine ranking on earlier versions.

3. MySQL / MariaDB

Connection String

DATABASE_URL=mysql://user:password@localhost:3306/mydb
# or
DATABASE_URL=mariadb://user:password@localhost:3306/mydb

Features & Capabilities

  • Dialect: MySQL 8.0+ and MariaDB 10.5+.
  • Pagination: Uses LIMIT {take} OFFSET {skip}.
  • Identifiers: Enclosed in backticks (`table`.`column`).
  • Parameters: Positional ? placeholders.
  • Types: Supports INT, BIGINT, VARCHAR, TEXT, TINYINT(1) for boolean, DATETIME, JSON.

4. SQLite

Connection String

DATABASE_URL=sqlite://./dev.db
# or relative file path
DATABASE_URL=./dev.db

Features & Capabilities

  • Dialect: SQLite3 embedded engine.
  • Zero Configuration: No standalone database server required; operates directly on local disk or :memory:.
  • Pagination: Uses LIMIT {take} OFFSET {skip}.
  • Parameters: Positional ? placeholders.
  • Vector Search: A VECTOR(n) column stores a BLOB of little-endian float32, and the search is ranked inside the database — by sqlite-vec when it loads, otherwise by the runtime’s own an5_vec_* functions or json_each. No extension is required. See Vector Search for which strategy each runtime reaches.
  • Ideal for: Rapid local development, integration tests, and edge/desktop applications.

5. Google Sheets

Connection String

DATABASE_URL=googlesheets://spreadsheet_id?credentials=./service-account.json

Features & Capabilities

  • Serverless Storage: Uses Google Spreadsheet tabs as structured database tables.
  • Query Execution: Evaluates structured queries, filters, and projections in memory over sheet rows.
  • Use Cases: Lightweight CMS, rapid prototyping, admin data entry without running database servers.

6. Dedicated Vector Databases (NBase, Qdrant, Chroma, Pinecone, Weaviate, Milvus)

AN5 ORM supports 6 specialized vector database engines through a unified abstraction, usable either as a primary vector database or as a sidecar vector store paired with a SQL database.

Connection Strings

# NBase
DATABASE_URL=nbase://localhost:1307?token=secret&method=hnsw

# Qdrant
DATABASE_URL=qdrant://localhost:6333?collection=articles&apiKey=secret

# Chroma / ChromaDB
DATABASE_URL=chroma://localhost:8000?collection=articles&tenant=default_tenant&database=default_database

# Pinecone
DATABASE_URL=pinecone://my-index-svc.region.pinecone.io?apiKey=pcsk_secret&namespace=articles

# Weaviate
DATABASE_URL=weaviate://localhost:8080?className=Article&apiKey=secret

# Milvus / Zilliz Cloud
DATABASE_URL=milvus://localhost:19530?collection=articles&token=root:Milvus&dbName=default

Features & Capabilities

  • Unified REST Clients: Zero third-party SDK dependencies across all runtimes (TypeScript, Python, .NET, Go, Rust, Java, Kotlin, Swift).
  • Unified connectionString Configuration: Pass any vector database URI directly via connectionString (or DATABASE_URL).
  • Consistent Search API: Query nearest neighbors directly via vectorSearch() with cosine, euclidean, or dot distance metrics.