qianzii2

qianzii2 / rockduck

Public

HTAP embedded database: transactional DeltaStore + columnar Vortex storage + DuckDB SQL engine

20
0
89% credibility
Found May 28, 2026 at 20 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

RockDuck is an embedded database that combines fast transactional writes with powerful analytical queries, supporting time-travel to see past data states, SQL access through DuckDB, and export to the Iceberg format.

How It Works

1
💾 Store your data in one place

Open or create a database with a single folder path, and your data lives there safely with built-in crash recovery.

2
✍️ Add records with simple commands

Insert rows by specifying a table name, a unique key, and your column values — just like filling out a form.

3
🔍 Find what you need instantly

Look up records by their key or scan a range of records with optional filters to narrow down results.

4
⏪ See your data at any point in time

Query the database as it existed at any past moment — perfect for auditing, debugging, or comparing changes over time.

5
Choose how to work with your data
🦆
Query with SQL

Connect DuckDB to your database and write SQL queries to analyze your data with full SQL power.

📤
Export to Iceberg

Export frozen data segments to the standard Iceberg format so other tools can read your data.

🎉 Your data is always available and consistent

Whether you need fast transactions, powerful analytics, or time-travel queries — everything works together seamlessly in one embedded database.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 20 to 20 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is rockduck?

RockDuck is an HTAP embedded database written in Rust that handles both transactional and analytical workloads in a single engine. It combines row-based delta storage for writes with columnar storage for fast analytics, powered by DuckDB's SQL execution layer. You get MVCC snapshot isolation, time-travel queries, and native Iceberg v2 export -- all in a library you embed directly in your application without running a separate server.

Why is it gaining traction?

The architecture borrows proven ideas from production systems: Shadow Column MVCC from Apache Iceberg, segment organization from ClickHouse MergeTree, and zone map predicate pushdown from Snowflake. The adaptive compaction scheduler uses query feedback to automatically prioritize which segments to merge based on actual workload patterns. Iceberg v2 export means your data can be read by DuckDB, Spark, or any Iceberg-compatible tool without vendor lock-in.

Who should use this?

Rust developers building analytics features into edge devices, desktop apps, or embedded systems who want SQL queries without deploying a database server. Teams evaluating embedded columnar stores like DuckDB or SQLite extensions will find the HTAP routing interesting -- point queries skip the columnar overhead entirely while scans leverage vectorized execution. If you need time-travel queries or audit trails built into your storage layer, the MVCC design is worth evaluating.

Verdict

RockDuck is architecturally ambitious for a project with 20 stars -- the feature set (HTAP routing, adaptive compaction, Iceberg export, query feedback) suggests significant engineering investment. The credibility score of 0.9% reflects this early stage: documentation is extensive but test coverage and community feedback are unknown quantities. Worth watching, but wait for a 1.0 release or production case studies before betting on it for critical systems.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.