I Built a Storage Engine from Scratch in C++ — WAL, Raft, and Object Storage

I wanted to understand one thing: how does data actually survive a crash?

Not what the documentation says. Not what the abstraction promises. What actually happens at the byte level when a process dies mid-write, and how a storage engine recovers from it.

So I built VeriStore — a correctness-first key-value storage engine in C++, built from first principles, evolving from a simple in-memory store to a Raft-replicated distributed system with a mini S3-style object storage layer.

Why Build a Storage Engine?

Every database, stream processor, and distributed system you've ever used is built on top of primitives like:

Write-Ahead Logging (WAL)

Crash-consistent recovery

Group commit batching

Consensus replication Understanding these primitives doesn't just make you a better infrastructure engineer — it makes you better at using these systems because you understand what guarantees they actually provide.

How It Works

v0.1 —…