KiviDB Open-Sourced: High-Performance Redis-Compatible In-Memory Database and Vector Cache Engine Released
By Mr.Xu Community Post
Published:
Summary:KiviDB is an open-source in-memory database developed by KiviBio, designed to offer a high-performance, low-latency Redis-compatible solution. Built with Rust, it features a multi-threaded architecture and leverages io_uring for optimal performance on Linux. KiviDB supports both RESP2 and RESP3 protocols and is compatible with most Redis 8.10 commands. The core engine has been open-sourced on GitHub under the AGPLv3 license. In benchmarks, KiviDB outperforms Dragonfly and Redis in pipelined GET
Background and Motivation
KiviDB is developed by KiviBio to address the limitations of existing in-memory databases like Dragonfly and Redis, particularly in areas such as GC pauses, memory bloat, and OOM (Out Of Memory) issues. The project began in February 2026 with the goal of helping developers better understand the Rust language and incorporating the best features of other databases like Garnet and Dragonfly.
Technical Architecture and Features
- Multi-threaded and Sharded Architecture: KiviDB employs a multi-threaded architecture with a sharded keyspace across threads, leveraging multi-core processors effectively.
- io_uring Support: On Linux, KiviDB utilizes io_uring for efficient asynchronous I/O operations, significantly boosting performance.
- Cross-Platform Compatibility: With the tokio_uring crate, KiviDB excels on Linux while providing a tokio fallback for other operating systems.
- Protocol Support: It supports both RESP2 and RESP3 protocols, maintaining compatibility with most Redis 8.10 commands.
Performance
In benchmarks, KiviDB demonstrates superior performance in pipelined GET operations, achieving 21.2M ops/sec compared to Dragonfly 2.0's 9.1M ops/sec and Redis 8.10's 2.4M ops/sec. In non-pipelined scenarios, Dragonfly performs slightly better, but KiviDB maintains a stable 3.07M ops/sec. In terms of memory usage, KiviDB is comparable to Redis and slightly worse than Dragonfly.
Vector Data Processing
KiviDB supports efficient vector data querying, outperforming most alternatives at lower ef, but lagging behind Dragonfly at high ef.
Open-Sourcing and Community Building
The core engine of KiviDB has been open-sourced on GitHub under the AGPLv3 license. Developers can quickly deploy and test KiviDB using Docker. The author expresses a desire to build a community around KiviDB and provide better Rust-based tools.
Future Outlook
The release of KiviDB provides developers with a high-performance Redis-compatible in-memory database option, particularly in applications requiring low latency and high throughput. KiviBio plans to continue optimizing KiviDB and welcomes community contributions.
— END —Source: GitHub Projects via Hacker News (2026-10-19)
Tags: #KiviDB #Open Source AI #In-Memory Database #Vector Cache #Rust
Editorial & Fact-Checking Note: This article is compiled from primary research, official release documentation, and source papers by the ZICQ Newsroom pipeline with automated entity verification and human editorial review. If you notice any technical inaccuracy, please submit a correction via our corrections policy or email our editorial desk directly.
Community Comments