Three storage lines on dedicated hardware: backup targets that other machines write to, NAS servers that people and applications mount, and bulk chassis for datasets too large to rent by the terabyte. Up to 192 TB raw, RAID 6 or JBOD, 10 Gbps unmetered. From $89/mo.
The difference is not capacity — it is who reads the data and how often.
Object storage is priced to make retrieval expensive. A storage server is priced per month, and reading your own data is free.
At $1.99/TB/month on RAID 6 usable capacity, bulk storage is roughly ten times cheaper than S3 Standard — and reading it back costs nothing at all, against $20 to $90 per TB in retrieval fees on cold cloud tiers.
Cold cloud tiers restore in 3 to 12 hours. A storage server reads at 10 Gbps immediately, which is the difference between a restore drill you can actually run and one that exists on paper.
Full root and IPMI. Run ZFS, Ceph, MinIO or plain ext4; set your own snapshot and encryption policy rather than accepting a provider's.
RAID 6 survives any two simultaneous drive failures, and we replace failed drives at no charge, 24/7. Or take JBOD and let ZFS or Ceph handle redundancy the way you prefer.
Backup if the server is a target that other machines write to and nobody reads from day to day. NAS if people and applications mount it as a share. Bulk if the dataset is simply too large to be worth paying per-TB cloud pricing for.
RAID 6 spends two drives' worth of capacity on parity so the array survives any two simultaneous failures. On a 192 TB raw chassis that leaves about 160 TB usable. JBOD presents all of it raw if you would rather manage redundancy in ZFS or Ceph.
No. Every storage server ships with a 10 Gbps unmetered uplink, which is the point — retrieval fees are what make object storage expensive at scale.
Yes. Storage servers sit in the same datacenters as our dedicated servers, so a compute node and its backup target can share a facility and a private network segment. Ask us and we will place them together.