01
Why does this matter to us?
S3 compatibility differs between providers. Large data sets, versioning, private endpoints, bandwidth limits and reliable retries turn a simple copy operation into an ongoing operational process.
← All experiments
Workshop · In technical testing
s3s.sh: Object storage sync
A control plane for incremental synchronization and recovery between S3-compatible storage systems. It can run as a managed service or through an agent in the customer’s own infrastructure.
- Experiment status
- In technical testing
- Area of inquiry
- Cross-provider S3 backups
- Outcome
- Potential product path
What are we trying to find out?
How can large data sets be backed up repeatedly across different S3 providers without unnecessarily exposing private endpoints or credentials?
02
What did we build?
The control plane and data path are separated. Sync pairs describe the desired state; jobs run through isolated managed workers or self-hosted agents, receive secrets only when needed and report progress and results.
03
Where does the experiment stand?
Teams, schedules, billing, agents, the managed fleet, Vault-backed secrets, browsing and restore workflows are implemented. The remaining work mainly concerns real end-to-end acceptance and production infrastructure for the managed fleet.
04
What happens next?
Complete testing against external systems, compare the managed runtimes on real host hardware and then publish the project through a dedicated product page.
Technologies and topics
- Laravel
- S3
- rclone
- Vault
- MicroVMs
- Billing
