dingodb/dingo

A multi-modal vector database that supports upserts and vector queries using unified SQL (MySQL-Compatible) on structured and unstructured data, while meeting the requirements of high concurrency and ultra-low latency.

What it solves

DingoDB is a distributed multi-modal vector database designed to handle massive datasets while maintaining strong consistency and relational semantics. It solves the problem of needing a unified platform that can handle both traditional relational data (scalars) and vector data for AI applications, eliminating the need to manage separate databases for these two different data types.

How it works

DingoDB integrates vector semantics into a distributed system that is compatible with the MySQL protocol. It uses a distributed storage system based on the Raft protocol to ensure data high availability and strong consistency. The system supports automatic elastic data sharding, which allows it to dynamically split and merge shards to allocate resources efficiently as the business grows.

Who it’s for

Developers and enterprises building modern data-driven applications that require a high-availability, scalable vector database with the same familiarity of SQL-based access and MySQL compatibility.

Highlights

  • Hybrid Retrieval: Supports seamless fusion and retrieval of both scalar and vector data.
  • MySQL Compatibility: Offers seamless compatibility with the MySQL protocol for easier integration.
  • Disk-based Search: Provides cold-hot tiered retrieval to minimize memory consumption for massive datasets.
  • Automatic Scaling: Features fully automatic elastic data sharding for efficient resource management.
  • Built-in High Availability: Includes built-in configurations for high availability without requiring external components.

Related

  • Project
    tenstorrent/tt-metal

    A comprehensive software stack for Tenstorrent AI hardware, providing both high-level neural network libraries and low-level kernel programming tools to optimize and run large-scale models.

  • Project
    mlrun/mlrun

    An open-source AI orchestration platform that automates the delivery of ML and generative AI pipelines, from data management and training to serverless deployment and monitoring.

  • Project
    kubeflow/katib

    A Kubernetes-native AutoML project that provides automated hyperparameter tuning, neural architecture search, and early stopping for any ML framework.

  • Project
    BindsNET/bindsnet

    A PyTorch-based library for simulating spiking neural networks (SNNs) to develop biologically inspired machine learning and reinforcement learning algorithms.