LSM Tree

Stop Buying Expensive Databases. S3 Already Is One.

📅 August 18, 2026 📂 Tech Industry

S3 is object storage — dumb, slow, eventually consistent. Yet engineers built a fully transactional, crash-safe key-value database on top of it using LSM trees, conditional writes, and fencing tokens. The real lesson isn’t about S3. It’s that the line between “storage” and “database” is a convention you build, not a law you obey. If you understand your invariants, the cheapest substrate is often enough.

Compaction Conditional Writes Database Architecture Distributed Systems Go LSM Tree Object Storage S3

Most Job Queues Are Over-Engineered. Here’s the Proof.

📅 August 3, 2026 📂 AI & Machine Learning

Most job queues are over-engineered for the tasks they actually perform. Zizq is a single-binary, zero-dependency job queue that challenges the assumption that you need distributed infrastructure. Built on LSM-trees and lock-free data structures, it’s fast, durable, and language-agnostic. The real question: have we been scaling for the wrong problem?

Architecture Background Processing Developer Tools Job Queue LSM Tree Open Source Redis Rust Single Binary Zero-Dependency