Cinder Multiple Backends

Cinder supports multiple storage backends simultaneously. You can offer different storage tiers—fast SSD, bulk HDD, Ceph distributed—in the same OpenStack deployment and let users choose via volume types. This guide configures Cinder with LVM, Ceph, and NFS backends on OpenStack 2024.2 Dalmatian.

How It Works

Cinder uses volume types to map user requests to backends. When a user creates a volume with a specific type, the Cinder scheduler places it on the matching backend.

User creates volume (type=ssd) → Scheduler → LVM-SSD backend
User creates volume (type=ceph) → Scheduler → Ceph RBD backend
User creates volume (type=nfs)  → Scheduler → NFS backend

Prerequisites

Step 1: Configure Multiple Backends in cinder.conf

Edit /etc/cinder/cinder.conf on the storage node:

[DEFAULT]
enabled_backends = lvm-ssd,ceph-rbd,nfs-bulk
default_volume_type = ceph

# ---- LVM SSD Backend ----
[lvm-ssd]
volume_driver = cinder.volume.drivers.lvm.LVMVolumeDriver
volume_group = cinder-volumes
target_protocol = iscsi
target_helper = lioadm
volume_backend_name = ssd-storage

# ---- Ceph RBD Backend ----
[ceph-rbd]
volume_driver = cinder.volume.drivers.rbd.RBDDriver
rbd_pool = volumes
rbd_ceph_conf = /etc/ceph/ceph.conf
rbd_user = cinder
rbd_secret_uuid = 457eb676-33da-42ec-9a8c-9293d545c337
rbd_flatten_volume_from_snapshot = false
volume_backend_name = ceph-storage

# ---- NFS Backend ----
[nfs-bulk]
volume_driver = cinder.volume.drivers.nfs.NfsDriver
nfs_shares_config = /etc/cinder/nfs_shares
nfs_mount_point_base = /var/lib/cinder/nfs
nas_secure_file_operations = false
nas_secure_file_permissions = false
volume_backend_name = nfs-storage

Create the NFS shares file:

echo "10.0.0.50:/export/cinder" | sudo tee /etc/cinder/nfs_shares
sudo chown cinder:cinder /etc/cinder/nfs_shares

Step 2: Create Volume Types

Volume types are how users select a backend:

source openrc admin admin

# Create volume types
openstack volume type create ssd
openstack volume type create ceph
openstack volume type create nfs

# Associate each type with a backend name
openstack volume type set ssd --property volume_backend_name=ssd-storage
openstack volume type set ceph --property volume_backend_name=ceph-storage
openstack volume type set nfs --property volume_backend_name=nfs-storage

Step 3: Restart Cinder

sudo systemctl restart cinder-volume cinder-scheduler

Verify all backends are reported:

cinder service-list
openstack volume service list

You should see three cinder-volume entries, one for each backend.

Step 4: Test Volume Creation

Create a volume on each backend:

openstack volume create --size 10 --type ssd fast-vol
openstack volume create --size 50 --type ceph distributed-vol
openstack volume create --size 100 --type nfs bulk-vol

openstack volume list

Step 5: Set a Default Volume Type

When users do not specify a type, Cinder uses the default:

openstack volume type set --property is_public=true ceph
cinder type-default ceph

Or in cinder.conf:

[DEFAULT]
default_volume_type = ceph

Backend Capabilities Comparison

Using QoS Policies

You can apply I/O limits per volume type:

# Create a QoS policy
openstack volume qos create high-iops \
  --consumer front-end \
  --property total_iops_sec=5000 \
  --property total_bytes_sec=209715200

# Associate with the SSD type
openstack volume qos associate high-iops ssd

Troubleshooting

Summary

Multiple Cinder backends let you offer storage tiers—fast SSD, distributed Ceph, and bulk NFS—within one OpenStack cloud. Volume types map user requests to the right backend, and QoS policies add I/O controls.