Vector Databases · myscale

MyScaleDB

MyScaleDB is an Apache 2.0 licensed vector database built on ClickHouse that adds high-performance vector search and full-text search capabilities while maintaining full SQL compatibility. It targets developers building AI/RAG applications who want familiar SQL syntax instead of custom APIs.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/myscale/MyScaleDB

GitHub stars

Forks

C++

Primary language

Apache-2.0

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	myscale/MyScaleDB
Owner	myscale
Primary language	C++
License	Apache-2.0 — OSI-approved
Stars	1k
Forks	72
Open issues	10
Latest release	myscaledb-v1.8.0 (2024-10-21)
Last updated	2025-02-05
Source	https://github.com/myscale/MyScaleDB

What MyScaleDB is

A C++20 ClickHouse fork optimizing columnar OLAP architecture for vector operations, metadata filtering, and hybrid text-vector queries. Supports structured, unstructured, vector, and JSON data in a single system with millisecond-scale latency on billion-scale vectors claimed.

Quickstart

Get the MyScaleDB source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/myscale/MyScaleDB.git2cd MyScaleDB3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Production RAG Systems with Rich Metadata

Combine vector similarity search with structured filtering, full-text search, and SQL joins in a single query. Eliminates separate systems for vector DB, text search, and relational data.

Large-Scale AI Application Backends

Single unified platform for managing embeddings, operational data, audit logs, and analytics. Columnar storage reduces resource consumption vs. row-based databases for analytical workloads.

Hybrid Search and Multi-Modal Search

Leverage native support for text-vector hybrid search, JSON metadata filtering, and image-search workflows without custom engineering across multiple systems.

Implementation considerations

Project created March 2024, latest release October 2024, last push February 2025: relatively young but actively maintained. Verify v1.8.0 stability for production before committing.
Requires operational expertise in ClickHouse (memory, disk, replication, sharding). Self-hosted deployment is non-trivial; consider managed MyScale Cloud unless you have ClickHouse operations expertise.
SQL compatibility is claimed but MyScaleDB is a fork with custom vector functions; test application migration and query compatibility thoroughly, especially vector-specific operations.
Performance benchmarks are provided on myscale.github.io/benchmark but are self-published; independent validation recommended before performance-critical decisions.
Data consistency guarantees, backup/recovery strategies, and disaster recovery SLAs should be validated for your durability requirements, especially if self-hosted.

When to avoid it — and what to weigh

High-Frequency Transactional Workloads — ClickHouse is an OLAP database optimized for analytical queries, not OLTP. Row-oriented databases (PostgreSQL, MySQL) are better suited for transaction-heavy, low-latency writes.
Specialized Vector DB Requirements — If you need proprietary vector algorithms (HNSW variants, custom distance metrics) not exposed by MyScaleDB, or require specialized vector DB tuning that ClickHouse does not expose.
Lightweight Embedded Vector Search — Overhead of running ClickHouse infrastructure may be unnecessary for small-scale vector workloads. Lighter alternatives (Qdrant, Weaviate, local embedding models) may be more cost-effective.
Minimal DevOps/Infrastructure Expertise — Self-hosted deployment requires managing ClickHouse infrastructure, configuration, tuning, and operational complexity. Managed cloud service (MyScale Cloud) available but introduces vendor dependency.

License & commercial use

Apache License 2.0 is a permissive OSI-approved license allowing commercial use, modification, and distribution with minimal restrictions (requires license notice and attribution).

Apache 2.0 permits commercial use without warranty or liability. However, verify: (1) whether you will use self-hosted (license applies fully) or MyScale Cloud (separate commercial terms apply); (2) whether ClickHouse upstream code carries any additional restrictions; (3) support and SLA requirements are not covered by the license alone.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Adequate
License clarity	Clear
Deployment complexity	Moderate
DEV.co fit	Good
Assessment confidence	Medium

Security considerations

Apache 2.0 code audit responsibility is yours. ClickHouse has a security history (review CVEs); MyScaleDB inherits this plus custom vector code. Network security (default config allows localhost only) and access control (users.d XML configuration) require proper hardening. Encryption-at-rest, encryption-in-transit, and secret management are not described; verify before production deployment. No mention of security certifications or penetration testing results.

Alternatives to consider

PostgreSQL + pgvector

Row-oriented transactional database with native vector extension. Simpler operational overhead, strong transaction semantics, but slower on large-scale analytical and filtered vector queries vs. MyScaleDB's columnar approach.

Pinecone / Weaviate / Qdrant

Purpose-built vector DBs with simpler APIs and managed cloud options. Specialized performance for vector-only workloads, but lack unified SQL + structured data capabilities and higher cost at scale.

Elasticsearch + Vector Search

Unified full-text and vector search with inverted index strength. Less efficient for structured metadata filtering and SQL joins vs. MyScaleDB's columnar model; different language and tooling ecosystem.

Software development agency

Build on MyScaleDB with DEV.co software developers

Request a technical deep-dive on deployment architecture, performance validation, and production readiness before committing. Compare with PostgreSQL+pgvector, Pinecone, and Weaviate based on your use case.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

dingo lancedb qdrant

Related on DEV.co

Explore the category and the services that help you build with it.

Vector Databases AI Application Development API Development Cloud Deployment Talk to DEV.co

MyScaleDB FAQ

Can I migrate from PostgreSQL + pgvector to MyScaleDB?

Partially. SQL schema structure is portable, but vector functions and indexing strategies differ. pgvector uses HNSW; MyScaleDB adds proprietary MSTG (in Cloud) and other algorithms. Plan for query rewriting and functional testing.

What is the cost difference between self-hosted and MyScale Cloud?

Unknown from provided DATA. MyScale Cloud offers managed service, MSTG algorithm, and scaling convenience. Self-hosted is free (Apache 2.0) but requires ops investment. Compare cost models on myscale.com/pricing.

Does MyScaleDB support real-time updates?

ClickHouse supports mutations and inserts, but is optimized for analytical queries, not real-time transactional updates. Batch inserts and near-real-time latency are typical; assess your ingestion SLA requirements independently.

Are there production deployments and reference customers?

Unknown from provided DATA. Check myscale.com case studies or community Discord for deployment examples. Nascent project (1036 stars) suggests emerging production adoption; request references before committing critical workloads.

Software development & web development with DEV.co

From first prototype to production, DEV.co delivers software development services around tools like MyScaleDB. Our software development agency staffs experienced software developers and web developers for custom software development, web development, integrations, and ongoing support across vector databases and beyond.

Evaluate MyScaleDB for Your Vector Workload

Talk to DEV.co Schedule a Consultation