Volumes
Volumes
Extra disks on our software-defined storage, read-optimised or write-optimised, with backups and snapshots.
What a volume is
A volume is an extra disk you attach to a server. The server sees it as a new disk, you format it and mount it, and it can grow while the server runs.
Data on a volume is kept apart from the system disk. Reinstalling the server keeps its attached volumes, and a volume detached from one server attaches to another.
Why ours
Apart from the host
A volume lives on our software-defined storage, not on the machine the server runs on. Every class except Solo keeps replicas, so a failed disk or host does not take your data with it.
Low latency
A 4k read answers in about 160 microseconds and a replicated write in about 150, measured inside an ordinary customer server.
Read-optimised or write-optimised
Basic disks fit almost everything. Database disks run on drives with power-loss protection, so a database commits at full speed and an acknowledged write survives a power cut.
Change it in place
Grow a volume while the server runs, or move it to another speed tier of the same class. Only the speed limits change, the data stays where it is.
Four disk classes
Basic is the default, replicated, with five speed tiers. Database has the same tiers on drives with power-loss protection, for databases and queues.
Solo keeps a single copy and runs as fast as the host goes, for caches, build workspaces and scratch data. Storage is replicated bulk space for archives, media and cold data, made for large sequential copies.
Speed limits are exact: a tier gives the number shown on it. A Basic server cannot drive the top tiers, so the wizard shows what it really reaches, and High Frequency servers get the full tiers.
Snapshots and backups
A snapshot is a quick copy of the disk kept next to it. Use it to roll back after a bad change or to make a new disk from it.
A backup is an encrypted copy of the disk stored in another data centre. It survives the loss of the disk and the cluster.
Both can run on a schedule set for the volume, or follow the account default. A restore makes a new disk with the same filesystem and data.
How to start
- Create a volume in the Cloud Panel and pick a class, a speed and a size.
- Attach it to a running server from the Volumes list.
- Format it and mount it. A volume made from a snapshot or a backup already has its data, mount it as it is.
Example usage
# find the new disk
lsblk
# format a new volume and mount it
sudo mkfs.xfs /dev/sdb
sudo mkdir -p /mnt/data
sudo mount /dev/sdb /mnt/data
# mount it on every boot
echo '/dev/sdb /mnt/data xfs defaults,nofail 0 2' | sudo tee -a /etc/fstabQuestions
Which class should I take?
Basic fits almost everything. Take Database for a busy transactional database, Solo for data you can make again, and Storage when you need terabytes more than speed.
Can a volume grow?
Yes, while it is attached. It cannot shrink. The system may need a rescan to see the new size.
Can one volume be attached to two servers?
A volume attaches to one server at a time. Detach it and it can go to another server.
How large can a volume be?
Up to 10 TB. Each speed tier has its own range of sizes, shown when you create the volume.
Do I pay for a volume that is not attached?
Yes. A volume is billed for its size by the hour, attached or not.
What happens to a Solo volume when the hardware fails?
Its data is lost. Solo keeps a single copy, so put only data on it that you can make again.