Back to browse
My Fable 5 Project Was a Multiraft Database and Blob Store

My Fable 5 Project Was a Multiraft Database and Blob Store

by SamInTheShell·Jul 19, 2026·1 point·0 comments

AI Analysis

●●●BangerWizardryRabbit HoleEye Candy

Live browser simulation of Raft consensus and sharding you can crash and recover in real-time.

Strengths
  • Visualizing leader elections and quorum loss in real-time makes abstract consensus concepts concrete.
  • Interactive controls let users crash nodes and watch the cluster self-heal or fail visibly.
  • Backed by a real Go binary implementation, not just a theoretical diagram or animation.
Weaknesses
  • Documentation is sparse; the visualizer assumes prior knowledge of Raft and distributed storage.
  • The underlying database is a custom implementation competing with battle-tested systems like etcd.
Target Audience

Distributed systems engineers and students learning consensus algorithms

Similar To

etcd · Consul · The Raft Consensus Algorithm Visualizer

Post Description

Linked from this post is the visualizer and simulator I had Fable 5 create to demonstrate the

In a thread the other day, I asserted it was nice that code is no longer a moat. Getting Raft or Paxos consensus working is one of the most challenging tasks out there, even which etcd.io/raft and hashicorp/raft.

The database took about a day for Fable 5 to create. It was updated as tons of work was done on a sandbox app running on top of it.

What it delivered was a distributed unified KV and blob store. It automatically does sharding and erasure coding.

You'll find something named Private Cloud Platform in the `sandbox/` directory. Will be making a video on it soon going over the ridiculous amount of features it has packed into it. There's some things not fully fleshed out like the docs and spreadsheets (no import/export or cross format support yet), but has other features like a Markdown document editor, Kanbans, and Diagrams.

Most of the work since Fable 5 service was restored has been in the Private Cloud Platform app. I done cursory tests to just make sure the features seem like they work (except the home camera/doorbell stuff, that was just added yesterday).

All the testing so far has been done in a KinD cluster (with the service binaries deployed on budget clouds). I'm preparing for the next phase of testing, where I just run this and do some chaos monkey style validation over the next couple of months. While running this, I'm also going to live out the lifecycle of the system and fix any gaps missed during the development process so far.

It is genuinely interesting. The code moat has been reduced to "can you describe the problem and what you want?" (and do you have some disposable income for an AI subscription?).

This project still has a lot left to do. Testing-wise, I'm going to busy for months. Also, you'll notice there's an API for most things in the Private Cloud Platform. I haven't even started working on the mobile apps, but Opus 4.8 and maybe even Sonnet 5 can knock those out.

My prompting to get the database mostly involved discussing the features, architecture, and intentions for the software. The past decade for me, I've been reading about consensus and distributed data patterns, so describing what I wanted would have been non-trivial without the high level domain knowledge. I also had a couple of days before Fable 5 went down to see what it could do if I'm very vague (it wasn't good, but no clue if Fable would have fixed it since I started over). I also seeded directories structures and maintained open source libraries I was leveraging like pebble and etcd/raft in a REFERENCES/ directory so it could be pulled into context as needed. The AI did not choose much for me on how the project was structured or how the architecture was supposed to be. I do not use MCP servers or skills (except whatever skills the model might have automatically used).

Overall, I'm having a blast and am enjoying the temperature of this code moat.

Happy prompting everyone!

Similar Projects

Other●●Solid

2D Coulomb Gas Simulator

WebGPU-accelerated random matrix visualizer; deep research appeal, narrow audience.

Big BrainRabbit Hole
swesnow
4495mo ago
Developer Tools●●Solid

I built an interactive LSM Tree simulator

This actually simulates the full write path — WAL to memtable to immutable flush — and animates cascading leveled compactions so you can watch key movement and file counts in real time. The live bloom-filter checks and amplification metrics are the parts that will teach you something immediately; it's clearly built as a learning/debugging sandbox rather than a production profiler.

Niche GemWizardry
saiprakashreddy
105mo ago