Whatever else has ever been said about OpenStack, no one has ever said the open-source Infrastructure-as-a-Service (IaaS) cloud was easy to deploy or update. With the release of Pike, the 16th version of OpenStack, that's changing.
Pike, and the two updates, Queens and Rocky, to follow it, won't bring major new features or changes. Instead, each will build on the Ocata release. Ocata, too, was focused on improving stability, scalability, and performance of the core services.
Specifically, Pike offers new capabilities that improve manageability and provide greater flexibility and scale. These include:
- Nova Cells v2: The Nova Cells compute architecture supports large deployments and scaling the compute service. Version 2 allows operators to shard their deployments to help with scaling the database and message queue, as well as segregate failure domains and help eliminate single points of failure.
- Python 3.5 upgrade: Working across all projects, the community introduced support for Python 3.5 to be ready for Python 2.x end-of-life in 2020 and to take advantage of its new features and increased performance. OpenStack is written in Python and all its components expose Python application programming interfaces (API)s. This change will eventually speed up all aspects of OpenStack.
- Leveraging etcd: CoreOS's Etcd is is a distributed key value store that provides a reliable way to store data across a cluster of machines. Starting with this release, etcd v3 will be used for distributed lock management.
- Ironic bare metal service matures: Ironic continues to mature in the Pike release, with the ability to plug into Neutron networks for true multi-tenancy. Ironic also joins Cinder, Neutron, Nova, and Swift as projects that support rolling upgrades, letting operators roll out new code without interrupting service. This will make updating OpenStack much easier.
- Cinder, OpenStack's block storage: Now comes with 'revert to snapshot' and ability to extend volumes. Revert to snapshot enables users to recover from events such as data corruption, or to reset after running tests. Users can also now extend storage volumes without shutting down virtual machines (VM)s, keeping applications online during extensions.
- Improved OpenStack Kolla lifecycle management tools. Kolla provides production-ready containers and deployment tools for operating OpenStack clouds. These make it easier to manage and upgrade OpenStack using DevOps programs such as Kubernetes and Ansible.
- Swift object storage incorporates globally distributed erasure codes: Even if the cross-region network is down, individual regions can still function, and failures in one region can use the remote region to recover. Swift also added performance improvements by enabling users to run multiple concurrent processes per server.
- Zun container management added: With this release, OpenStack's own native container management service has finally been added.
OpenStack's modular architecture also enables you to pick the functionality you need -- whether that's bare metal or block storage provisioning -- to plug into your infrastructure stack. This composability makes it easy to install just as much OpenStack as you need for a specific job. For example, OpenStack Ironic bare metal service now features enhanced integration for Cinder block storage and Neutron networking, and Cinder can now act as a standalone storage service for virtual machines, bare metal, or containers using Docker or Kubernetes.
"The features and upgrades that Pike brings are the lessons of experience you get from enabling thousands of public and private clouds, large and small, for seven-plus years," said Jonathan Bryce, the OpenStack Foundation's executive director in a statement. "The rise of composable services and simpler consumption options are part of that maturation process. Our community is now focused on eliminating future technical debt as well as growing OpenStack's capabilities to support ever-expanding use cases."
Pike jumps into existence as more OpenStack users are adopting a multi-cloud strategy and placing workloads across public and private cloud environments. According to the April 2017 OpenStack user survey, vendor lock-in was the number one business driver for OpenStack clouds, and 38 percent of OpenStack deployments interact with at least one other public or private cloud environment.
OpenStack is continuing to experience strong growth, with the April 2017 user survey reporting 44 percent more deployments compared to the previous year and new at-scale production deployments in Europe and China at China UnionPay, Paddy Power Betfair, and Tencent, which uses OpenStack to power WeChat. The software now powers 60 public cloud availability zones and over a thousand private clouds running across more than five million physical cores.
Want to try it? You can download OpenStack Pike and read the release notes starting on Aug. 30.
PREVIOUS AND RELATED COVERAGE
OpenStack continues to grow across businesses and grows ever stronger on the edge of the internet with its telecomm partners.
At the VMworld conference in Las Vegas, VMware is demonstrating how its strategy to help hybrid cloud deployments extends to multiple public clouds. Stephanie Condon
Besides managing OpenStack clouds, Mirantis is adding cloud container management to its skillset with Kubernetes.