Durable volumes
Your database and application disks survive a machine failing, without a restore.
All six parts of Forge
Citadel · Storage
Storage that survives a machine failing
Citadel is where data on Forge lives - the disks under applications and databases, and S3-compatible buckets for files. On the main tier, every piece of data is kept on three different machines.
Powered by Ceph
Why it matters
A disk in a single server is a single point of failure. When it goes, so does everything on it, and the restore starts.
Citadel spreads data across machines instead. Losing one machine means Ceph quietly makes a new third copy, rather than someone restoring from last night.
- copies on the main tier
- 3
- storage machines
- 4
- storage drives in the cluster
- 14
- raw space in the cluster
- 107 TB
How it works
Three copies, three machines
The main tier runs on fast NVMe drives. Each piece of data is written three times, and Ceph places each copy on a different machine.
Disks for applications
Virtual machines and Kubernetes workloads get block volumes from Citadel, so an application can move between machines and keep its data.
S3-compatible buckets
Files, media and backups go in S3-compatible buckets. Each service gets its own bucket and its own credentials, kept in Vault.
A bulk tier for big, cold data
Large media and archives sit on a separate hard-disk tier, with copies spread across drives in one machine. It is cheaper per terabyte and does not carry the three-machine promise.
Live today
- Three copies on separate machines for the main tier.
- Block volumes for virtual machines and Kubernetes.
- S3-compatible buckets, one per service, with separate credentials.
- A bulk hard-disk tier for media and archives.
In progress
- Encryption at rest for stored data.
- Backups of buckets and volumes kept off-site, with tested restores.
- A more space-efficient layout for cold data.
What you get
Your own bucket
An S3-compatible bucket with its own credentials, usable from any S3 library or tool.
Standard formats
S3 and plain volumes: nothing proprietary to export from if you move.
Built on
Ceph
The open-source distributed storage system behind a large share of the world's private clouds.
Ceph CSI
Connects Kubernetes workloads to Ceph volumes.
Proxmox VE
Our open-source virtualisation platform, which runs Ceph on the cluster's machines.
Related parts
Keystone
How our services find each other, and how browsers and services know they are really talking to us.
Graft
Every setting lives in Git. Change the file and the platform follows - so there is a record of everything.
Flash
One "Sign in with Forge" account across the apps you use with us, defined as code. In testing now.