The OpenNebula project is proud to announce the release of OpenNebula 4.14 ‘Great A’Tuin’ Beta2. This Beta release introduces features not present in Beta1, like like for instance better support for Qcow2 and GPU support for VMs.
To support HPC oriented infrastructures based on OpenNebula, 4.14 also enables the consumption of raw GPU devices existing on a physical host from a Virtual Machine. There is no overcommitment possible nor sharing of GPU devices among different Virtual Machines, so a new type of consumable has been defined in OpenNebula and taken into account by the scheduler. VMs can now request a GPU, and if OpenNebula finds one free resource of type GPU available, it will set up the VM with PCI passthrough access to the GPU resource, enabling applications to get the performance boost of the direct access to a GPU card.
OpenNebula users managing vCenter infrastructures will also benefit from this upgrade. The workflow of the VM importing feature has been greatly improved through Sunstone, making it easier to import your existing workload into OpenNebula. Moreover, 4.14 adds the possibility to instruct OpenNebula whether or not it should save the disks, protect your users against accidental data lost! Last, but not least, a contextualisation improvement now allows to directly pass scripts to be executed in boot time to vCenter VMs, increasing the flexibility in VM customisation from OpenNebula in vCenter.
There are many other improvements in 4.14 (check the full list of changes in the development portal, or alternatively, the post on 4.14 beta1 release):
- ceph and qcow2 disk snapshotting
- image resizing on boot time
- ability to save VMs into VM Templates for later use
- better state management of VMs
- flexible context definition of network attributes
- ability to import running VMs not launched by OpenNebula from all the supported hypervisors (including the hybrid ones, for instance now it is possible to manage through OpenNebula Azure, SoftLayer and EC2 VMs launched through their respective management portals)
- the possibility to cold attach disks and network interfaces to powered off machines (which complement the hot attach functionality)
- improvements in accounting to keep track of disk usage
- better logging in several areas
- the ability to pass scripts to VMs for guest OS customization
This OpenNebula release is named after Great A’Tuin, the Giant Star Turtle (of the fictional species Chelys galactica) who travels through the Discworld universe’s space, carrying four giant elephants who in turn carry the Discworld. Allegedly, it is “the only turtle ever to feature on the Hertzsprung–Russell diagram.”
The OpenNebula team is now set to bug-fixing mode. Note that this is a beta release aimed at testers and developers to try these new features (not production environments) and send a more than welcomed feedback for the final release.
Several organizations have sponsored the project through the Fund a Feature Program:
- Disk snapshots with Ceph backend was funded by Unity.
- Qcow2 snapshots implementation was funded by BIT.nl.
- GPU devices support was funded by SURFsara.
- Flexible network attributes definition in contextualization was funded by Université Catholique de Louvain.
The OpenNebula project would like to thank these organizations and the community members and users who have contributed to this software release by being active with the discussions, answering user questions, or providing patches for bugfixes, features and documentation.