Train, the 20th version of the cloud infrastructure software OpenStack, is now available with improved security and resource management.
Some of the most highlighted features of the release can be found in the Nova component which helps to manage and automate computer resource pools. The cloud computing fabric controller now for example sports a framework for hardware-based encryption which can be useful in scenarios where a system’s hardware is publicly accessible.
Nova also supports live migration for servers with a NUMA topology, pinned CPUs, or huge pages as well as those with SR-IOV ports attached when the libvirt compute driver is in use. Like this, hypervisors are able to run pinned and unpinned workloads in parallel.
The separation of Placement and Nova has been completed in the Train release, meaning Placement is a standalone service now that has to be installed separately if needed. Users are promised “massive performance enhancements in the placement service, especially in environments with large numbers of resource providers and high concurrency” by the new incarnation.
Data and metadata protection component Karbor now has notifications for plan, checkpoint, restore, scheduled, and trigger operations available. It also allows users to backup image boot servers with newly added data from the root disk.
Other security related changes can be found in the Keystone identity service for example, which has been fitted with access rules support for application credentials, and ways to make keystone roles, projects, and domains immutable. Due to some changes in its API, a look into the module’s documentation might be worth a minute of your time.
To make sure OpenStack is future-proof, the community also looked into ways of helping those parts of the user base, that have to work on machine learning use cases. The result allows the Cyborg project to create blueprints for launching and managing virtual machines with accelerators via a Cyborg-Nova interaction spec.
A more detailed list of the changes in OpenStack Train can be found in the project’s release notes.
While OpenStack seems to be alive and well with supporters in a variety of large enterprises, SUSE made the news last week by ditching their OpenStack Cloud and pretty much leaving the managed field to Red Hat to focus energies on its Cloud Application Platform, and container management product CaaS Platform.