Red Hat ブログ
For those of us who are involved in building and delivering Ceph, every major release of OpenStack is an exciting event that brings the opportunity to both deliver new features to existing customers and to show to new customers why Ceph remains the most popular storage option for OpenStack infrastructures according to an OpenStack Foundation user survey on enterprise OpenStack adoption that was published in November 2014.
The availability of Red Hat Enterprise Linux OpenStack Platform 6, based on OpenStack Juno, marks a particularly important milestone for Red Hat through the delivery of Ceph’s full set of advanced to ephemeral volumes as well as the ongoing support for persistent volumes and images.
The availability of Red Hat Enterprise Linux OpenStack Platform 6 represents the second successful engineering collaboration between Red Hat’s Storage, Platform and OpenStack teams since Red Hat acquired Inktank in April 2014. With today’s release, a single, integrated installer is now available for Red Hat Enterprise Linux OpenStack Platform 6 with Inktank Ceph Enterprise. This greatly simplifies and speeds deployment of the integrated solution.
Of equal importance is that Inktank Ceph Enterprise has gone through Red Hat’s extensive QA and testing for increased reliability and interoperability. Red Hat’s unique ability to deliver a full stack engineered by many of the core developers of the open source technologies means we can deliver a thoroughly tested set of best practices for deployment, performance optimization and lifecycle management. This greatly helps Red Hat’s OpenStack partners deploy and manage a single solution for Ceph OpenStack storage.
The majority of virtual machines (VM) running within OpenStack use multiple types of storage as part of their lifecycle. Ceph is used extensively to provide block devices via the Cinder API for persistent volumes, namely volumes which will survive the destruction of a VM and typically host application data. Ceph is also used to back up for Glance where OS images and snapshots of volumes can be stored. The combination of copy-on-write images and snapshotting means users are able to enjoy fast cloning of images and fast backups of volumes all under a single storage system.
However, users still need to run the actual OS itself on a local disk in the compute node. With Red Hat Enterprise Linux OpenStack Platform 6, this ephemeral volume can now also be stored and managed within Ceph, reducing the need to manage this extra layer of storage at the compute layer. Unlike in the local storage model where the OS image has to be pulled across the network to the hypervisor host, Ceph-backed ephemeral volumes enable the data to stay in situ within the Ceph cluster allowing the VM to boot more quickly without data moving across the network. This also means that snapshots of the ephemeral volume can be performed on the Ceph cluster instantaneously and then put into the Glance library, again without data migration across the network. Now VM storage functions can be delivered transparently to the user on Ceph.
For administrators, this also means that live migrations of VMs to new compute nodes to repair or decommission a node can be performed without moving any data. Only the hypervisor instance itself needs to move so both persistent and ephemeral data can be detached from the old hypervisor instance and connected to the new one.
Based on these features and foundations, Red Hat Enterprise Linux OpenStack Platform 6 with Red Hat’s Inktank Ceph Enterprise storage solution gives enterprises one of the best platforms available for deploying a production-ready OpenStack cloud for both modern applications and scale-out storage. Visit www.inktank.com/ for more information about Red Hat’s Inktank Ceph Enterprise.