Case study
What Spindle is — without the hype.
Founded by Max McCutcheon. This page is the honest product brief for companies, recruiters, and anyone Googling the name.
Why teams evaluate Spindle vs Dynamo alone
DynamoDB and peers win on raw hyperscale. Spindle aims at convenience teams can defend: an MIT engine you can open when storage misbehaves, portability out of one cloud, and founder-reachable support — plus the idea of one flat Cloud bill (no RCU/WCU spreadsheets) if hosted durability ships. That is positioning, not a claim that businesses already rely on Spindle Cloud. See /enterprise/.
Does the product actually work?
The engine: yes. Spindle is a real Rust LSM you can clone, cargo test, crash with kill -9, and walk in DESIGN.md. WAL, memtable, SSTables, leveled compaction, blooms, scans, MVCC — implemented and tested. Directional benches (mem put ~1.9µs, durable put ~238µs, flushed get ~8µs) are documented there.
Spindle Cloud hosting: not built. There is no public multi-tenant API endpoint, no hosted PITR / dashboards / uptime SLA, and no way to pay. The site, Agent (local brain on Pages; GPT needs a server + key), and playground are public. Cloud has a waitlist so demand decides whether it gets built.
Hasn’t this been done before?
Yes — LSMs aren’t new. LevelDB, RocksDB, Badger, and a hundred papers exist. Spindle isn’t a research breakthrough. It’s a founder-built engine + product surface: crash stories, compaction amp, honest benches, and a Cloud path companies can subscribe to as it grows.