11 February 2020 Reading time: 3 minutes

Ceph support in VMmanager: create high-availability clusters

We have added support of Ceph network storage to VMmanager. Now a high-availability cluster can be created directly from the panel. Integration with Ceph is available for versions 3.7.1 or higher, starting from February 11. Below we will review the advantages of Ceph and how integration with this storage is implemented in VMmanager.

How Ceph works

Ceph is a distributed network storage. Its features and advantages are most evident in comparison with the standard cluster architecture.

Standard cluster architecture. Virtual machines are stored and executed on the same node.

Distributed storage. The disks of the virtual machines are stored in network storage, while the execution point is on the node.

In a cluster with standard architecture, the management server has an exceptional role. During maintenance, the administrator has to move the execution point along with the disk file of the virtual machine to another server on the network. In case of failure, you can only restore virtual machine data manually from backups. Both transfer and recovery take a long time.
Standard cluster architecture. When migrating virtual machine No.6, all its data are copied from server to server. The operation is resource-intensive and time-consuming.
In a cluster with distributed storage, the execution point is not bound to a single server. The transfer is in one click: the domain of the virtual machine is changed and the drive is simply reconnected. In case of failure, the virtual machine domain is generated and its execution is started on a running server. The VM drive is automatically connected to a new execution point.
Distributed storage. When you migrate a virtual machine, only its settings are copied. The hard drive of the virtual machine is simply reconnected. The operation takes a few seconds.

Advantages of Ceph for business

There are several advantages to using Ceph to organize your infrastructure.
  1. You are no longer bound to the space constraints of a particular node. You no longer need to try to guess how much space is available at the node for backup, since all drives are now in network storage.
  2. The service speed of the node increases by many times and does not affect the customer’s services. The time when the system remains unavailable (RTO) is reduced. Downtime is excluded.
  3. Ceph ensures preservation of customer’s data.

Integration to VMmanager

Ceph provides both file and block access interface. VMmanager supports RBD - distributed block device with kernel client and QEMU/KVM driver.
When using RBD, virtual disks are distributed to several objects and stored in this form in a distributed Ceph - RADOS storage.

How to start using Ceph in VMmanager

  1. Create a new cluster and select "Ceph" storage type.
  2. Connect Ceph storage, specify IP address, login and password.
  3. Connect the nodes.