Object storage is only useful if data can be retrieved and accounted for under failure. Storj's distributed model should be evaluated with your actual object sizes, access pattern, and security requirements.
Upload a representative mix of small and large files through the interface you would operate. Verify checksums after download, test listing and deletion, and measure latency from the regions your users occupy. Calculate cost for storage, retrieval, and requests—not just the headline price. Document who holds access credentials and encryption keys, and what happens when they are lost. Consult the Storj documentation for current S3 compatibility and lifecycle behavior.
Rehearse a restore of an entire sample dataset with another tool. Keep an independent copy until retrieval, access control, retention, and billing are proven. A successful upload is the beginning of the test, not its conclusion.
Akousa