Cinch is in public alpha.
IoT & Device Fleets
10,000 devices. Most of them offline right now. Each one needs its own data store. Traditional databases make this wildly expensive. Cinch makes it trivial.
One database per device
The usual approach: one big TimescaleDB or InfluxDB instance. Every device writes to the same tables. Queries need device_id filters everywhere. Compliance means row-level access control. And when one device starts blasting telemetry at 10x the normal rate, reads for every other device slow down.
With Cinch, each device gets its own isolated database for telemetry, config, and state. Active devices run on NVMe. Offline devices archive to cloud storage automatically. There are no shared tables to filter against because the data is physically separate.
Fleet economics
Most IoT fleets follow a pattern: 10-20% of devices are active at any given time. The rest are sleeping, powered off, or out of range. You shouldn't pay active prices for the whole fleet.
| Fleet size | Shared time-series DB | Cinch |
|---|---|---|
| 1,000 devices | ~$200/mo | ~$25/mo |
| 10,000 devices | ~$1,500/mo | ~$245/mo |
| 100,000 devices | ~$10,000+/mo | ~$2,450/mo |
Assumes 100MB per device, 15% active. Active = $0.50/GB NVMe. Offline = $0.20/GB cloud storage.
Matches the device lifecycle
Database wakes from cloud storage. Milliseconds to ready. Device writes telemetry and reads config.
After idle timeout, database archives to cloud storage. Cost drops to $0.20/GB. No manual intervention.
Delete the database. All device data gone. No orphaned rows in shared tables to clean up.
Create a database via API during device setup. Milliseconds. Part of your provisioning flow.