Why MyVector? • Quick Start • Features • Usage • Docker • Contributing • Docs

MyVector brings the power of vector similarity search directly to your MySQL database. Built as a native plugin, it integrates seamlessly with your existing infrastructure, allowing you to build powerful semantic search, recommendation engines, and AI-powered applications without the need for external services.

It's fast, scalable, and designed for real-world use cases, from simple word embeddings to complex image and audio analysis.

flowchart LR
  App["Application / SQL Client"] --> MySQL["MySQL Server"];
  MySQL --> UDFs["MyVector UDFs"];
  MySQL --> Proc["MyVector Stored Procedures"];
  UDFs --> Index["Vector Index (HNSW/KNN)"];
  Proc --> Index;
  Binlog["MySQL Binlog"] --> Sync["Binlog Listener"];
  Sync --> Index;
  Index --> Data["MyVector Data Files"];

Get up and running with MyVector in minutes using our official Docker images.

docker run -d \
  --name myvector-db \
  -p 3306:3306 \
  -e MYSQL_ROOT_PASSWORD=myvector \
  -e MYSQL_DATABASE=vectordb \
  ghcr.io/askdba/myvector:mysql-8.4

mysql -h 127.0.0.1 -u root -pmyvector vectordb

Let's use a simple example with 50-dimensional word vectors.

-- Create a table for our word vectors
CREATE TABLE words50d (
  wordid INT PRIMARY KEY,
  word VARCHAR(50),
  wordvec VARBINARY(200) COMMENT 'MYVECTOR(type=HNSW,dim=50,size=100000,dist=L2)'
);

-- Download and insert the data (from the examples/stanford50d directory)
-- In a real-world scenario, you would generate your own vectors.
-- wget https://raw.githubusercontent.com/askdba/myvector/main/examples/stanford50d/insert50d.sql
-- mysql -h 127.0.0.1 -u root -pmyvector vectordb < insert50d.sql

CALL mysql.myvector_index_build('vectordb.words50d.wordvec', 'wordid');

Find words similar to "school":

SET @school_vec = (SELECT wordvec FROM words50d WHERE word = 'school');

SELECT word, myvector_row_distance() as distance
FROM words50d
WHERE MYVECTOR_IS_ANN('vectordb.words50d.wordvec', 'wordid', @school_vec, 10);

You should see results like "university," "student," "teacher," etc. It's that easy!

• Approximate Nearest Neighbor (ANN) Search: Blazing-fast similarity search using the HNSW algorithm.
• Exact K-Nearest Neighbor (KNN) Search: Brute-force search for 100% recall.
• Multiple Distance Metrics: L2 (Euclidean), Cosine, and Inner Product.
• Real-time Index Updates: Automatically keep your vector indexes in sync with your data using MySQL binlogs.
• Persistent Indexes: Save and load indexes to and from disk for fast restarts.
• Native MySQL Integration: Implemented as a standard MySQL plugin with User-Defined Functions (UDFs).

• Build an Index:

  CALL mysql.myvector_index_build('database.table.column', 'primary_key_column');

• Check Index Status:

  CALL mysql.myvector_index_status('database.table.column');

• Drop an Index:

  CALL mysql.myvector_index_drop('database.table.column');

• Save and Load an Index:

  -- Indexes are automatically saved. To load an index on startup:
  CALL mysql.myvector_index_load('database.table.column');

• Create a Vector:

  SET @my_vector = myvector_construct('[1.2, 3.4, 5.6]');

• Calculate Distance:

  SELECT myvector_distance(@vec1, @vec2, 'L2');

• Display a Vector:

  SELECT myvector_display(wordvec) FROM words50d LIMIT 1;

We provide official Docker images for various MySQL versions.

version: '3.8'

services:
  myvector:
    image: ghcr.io/askdba/myvector:mysql-8.4
    ports:
      - "3306:3306"
    environment:
      MYSQL_ROOT_PASSWORD: myvector
      MYSQL_DATABASE: vectordb
    volumes:
      - myvector-data:/var/lib/mysql

volumes:
  myvector-data:

We welcome contributions! Please see our Contributing Guide for details on how to get started.

For more detailed information, please see the docs directory:

• DEMO.md: A full demo with the Amazon Product Catalog dataset.
• ANN_BENCHMARKS.md: Performance benchmarks.
• DOCKER_IMAGES.md: More on using our Docker images.
• CONTRIBUTING.md: Our contribution guidelines.

• hnswlib: For the high-performance HNSW implementation.
• The MySQL Community: For creating a powerful and extensible database.

MyVector is licensed under the GNU General Public License v2.0.

For information about third-party dependencies and their licenses, see:

• NOTICE - Third-party attribution notices
• licenses/ - Full license texts for all dependencies