Chapter 4. Snapshot management

A snapshot is a read-only copy of the state of an image at a particular point in time. One of the advanced features of Ceph block devices is that you can create snapshots of the images to retain a history of an image’s state. Ceph also supports snapshot layering, which allows you to clone images quickly and easily, for example a virtual machine image. Ceph supports block device snapshots using the rbd command and many higher level interfaces, including QEMU, libvirt, OpenStack and CloudStack.

When cephx is enabled, you must specify a user name or ID and a path to the keyring containing the corresponding key for the user.

You might also add the CEPH_ARGS environment variable to avoid re-entry of the following parameters:

rbd --id USER_ID --keyring=/path/to/secret [commands]
rbd --name USERNAME --keyring=/path/to/secret [commands]

[root@rbd-client ~]# rbd --id admin --keyring=/etc/ceph/ceph.keyring [commands]
[root@rbd-client ~]# rbd --name client.admin --keyring=/etc/ceph/ceph.keyring [commands]

Create a snapshot of a Ceph block device.

List the block device snapshots.

Rollback a block device snapshot.

Delete a snapshot for Ceph block devices.

Purge block device snapshots.

Rename a block device snapshot.

Ceph supports the ability to create many copy-on-write (COW) or copy-on-read (COR) clones of a block device snapshot. Snapshot layering enables Ceph block device clients to create images very quickly. For example, you might create a block device image with a Linux VM written to it. Then, snapshot the image, protect the snapshot, and create as many clones as you like. A snapshot is read-only, so cloning a snapshot simplifies semantics—​making it possible to create clones rapidly.

Each cloned image, the child, stores a reference to its parent image, which enables the cloned image to open the parent snapshot and read it. This reference is removed when the clone is flattened that is, when information from the snapshot is completely copied to the clone.

A clone of a snapshot behaves exactly like any other Ceph block device image. You can read to, write from, clone, and resize the cloned images. There are no special restrictions with cloned images. However, the clone of a snapshot refers to the snapshot, so you MUST protect the snapshot before you clone it.

A clone of a snapshot can be a copy-on-write (COW) or copy-on-read (COR) clone. Copy-on-write (COW) is always enabled for clones while copy-on-read (COR) has to be enabled explicitly. Copy-on-write (COW) copies data from the parent to the clone when it writes to an unallocated object within the clone. Copy-on-read (COR) copies data from the parent to the clone when it reads from an unallocated object within the clone. Reading data from a clone will only read data from the parent if the object does not yet exist in the clone. Rados block device breaks up large images into multiple objects. The default is set to 4 MB and all copy-on-write (COW) and copy-on-read (COR) operations occur on a full object, that is writing 1 byte to a clone will result in a 4 MB object being read from the parent and written to the clone if the destination object does not already exist in the clone from a previous COW/COR operation.

Whether or not copy-on-read (COR) is enabled, any reads that cannot be satisfied by reading an underlying object from the clone will be rerouted to the parent. Since there is practically no limit to the number of parents, meaning that you can clone a clone, this reroute continues until an object is found or you hit the base parent image. If copy-on-read (COR) is enabled, any reads that fail to be satisfied directly from the clone result in a full object read from the parent and writing that data to the clone so that future reads of the same extent can be satisfied from the clone itself without the need of reading from the parent.

This is essentially an on-demand, object-by-object flatten operation. This is specially useful when the clone is in a high-latency connection away from it’s parent, that is the parent in a different pool, in another geographical location. Copy-on-read (COR) reduces the amortized latency of reads. The first few reads will have high latency because it will result in extra data being read from the parent, for example, you read 1 byte from the clone but now 4 MB has to be read from the parent and written to the clone, but all future reads will be served from the clone itself.

To create copy-on-read (COR) clones from snapshot you have to explicitly enable this feature by adding rbd_clone_copy_on_read = true under [global] or [client] section in the ceph.conf file.

Clones access the parent snapshots. All clones would break if a user inadvertently deleted the parent snapshot. To prevent data loss, by default, you MUST protect the snapshot before you can clone it.

Clone a block device snapshot to create a read or write child image of the snapshot within the same pool or in another pool. One use case would be to maintain read-only images and snapshots as templates in one pool, and writable clones in another pool.

Before you can delete a snapshot, you must unprotect it first. Additionally, you may NOT delete snapshots that have references from clones. You must flatten each clone of a snapshot, before you can delete the snapshot.

List the children of a snapshot.

Cloned images retain a reference to the parent snapshot. When you remove the reference from the child clone to the parent snapshot, you effectively "flatten" the image by copying the information from the snapshot to the clone. The time it takes to flatten a clone increases with the size of the snapshot. Because a flattened image contains all the information from the snapshot, a flattened image will use more storage space than a layered clone.