Launching today
Thunder
A blazing-fast, persistent, key-value AI workload database
33 followers
A blazing-fast, persistent, key-value AI workload database
33 followers
Thunder is a blazing-fast, persistent, and scalable key-value database built in Rust for modern, data-intensive applications. Designed from the ground up for high throughput, low latency, and efficient resource utilization, Thunder uses a shard-per-core architecture to scale with CPU cores while minimizing contention. Unlike traditional in-memory databases, Thunder combines the performance of modern NVMe storage with intelligent memory management.
Free
Launch Team
Render Sandboxes: a computer for every agentIsolated compute, persistent state and a real filesystem
Promoted
The shard-per-core plus NVMe combo is what caught my eye — most fast KV stores I've used make you choose between in-memory speed and actually keeping the data. For AI workloads, where you're storing chat history and large payloads across runs, that tradeoff hurts. Curious how it holds up on large values in practice.
@jcbuilds You can test it and I will post the benchmark. it is currently in alpha and it is more than 2x faster than redis in 8mb - 512mb+ data size.