OpenStack Summit Tokyo recap: containers, cloud native, and a dash of Cinder
The OpenStack Summit in Tokyo has come and gone. If the topic of containers was a loud buzz at the spring summit in Vancouver, by the time everyone had arrived in Japan it was a loud roar.
Over the course of three days, from keynotes to breakout sessions, OpenStack’s interoperability and integration with new technologies (I’m looking at you, containers) was on display, alongside maturing community offerings, and the evolution of core projects as well as “big tent” ones. Check out our latest Elements of SolidFire community podcast for some highlights of Tokyo, or continue reading if the printed word is more your speed.
OpenStack Foundation’s chief operating officer Mark Collier called OpenStack an integration engine. “It’s about looking at the technologies emerging over the next ten years and building a platform that can take advantage of those, can bring them to market, bring them into the data center under one roof, one API.”
Two hot areas in OpenStack are networking (Neutron) and container (Magnum), both getting a lot of airtime during the conference.
- Magnum, OpenStack’s containers-as-a-service project had its first release in January and officially joined the Foundation in March. Magnum’s Liberty release expanded the list of container orchestration engines (COE) supported to include Apache Mesos, alongside Docker Swarm and Google Kubernetes, giving users greater choice in how they deploy containers.
- COEs, as an alternative to traditional VMs, are poised to be disruptive. However, the PTL for Magnum, Adrian Otto, characterized the relationship between OpenStack and containers as complementary. COEs address orchestration of the software layer, OpenStack addresses orchestration of the infrastructure layer. Magnum creates a “bay” for COEs that integrates into Nova, streamlining deploying and managing COEs in OpenStack.
- For the first time in OpenStack’s history, Neutron was the most active project during the current release cycle. Collier characterized Networking as “hot,” and said the market was growing serious about software-defined networking. Telcos, in particular, are embracing OpenStack for its Network Functions Virtualization (NFV) capabilities, and how NFV is changing how those telcos handle network traffic.
- Neutron’s usage is growing, up from 68% at the fall 2014 Paris Summit to 89% as of Tokyo. Networking services such as load balancing-, VPN-, and Firewall-as-a-service are helping to advance Neutron.
Core projects vs. Big Tent
The Foundation reorganized the ecosystem of OpenStack projects into Core – six main projects critical to all cloud deployments (Nova, Neutron, Keystone, Glance, Cinder, and Swift) – and what they call “Big Tent” – a wide variety of additional services, but ones that not every cloud needs (Trove, Heat, Magnum, etc.)
- Key new Cinder features in Liberty include a generic image caching solution, so popular VM images can be cached and copied-on-write to a new volume, as well as non-disruptive backups. Tokyo is the design summit for Mitaka, the next OpenStack release. Included with features such as improved replication, testing of rolling upgrades, and third-party CI, is how to address high availability, particularly how failover should occur if the Cinder service were to go down. Given containers’ rise in popularity, figuring out ways to integrate persistent block storage with COEs and enable easy volume provisioning and modification is a focus of Cinder – and SolidFire – as we look toward Mitaka.
- The Foundation highlighted two new community-related items. First up was the Project Navigator on www.openstack.org, which lists Core and Big Tent projects, their adoption, maturity, and age, helping users quickly gauge the status of their chosen OpenStack projects online. Jonathan Bryce, Executive Director of the OpenStack Foundation, also announced OpenStack administrator training that will commence in 2016. OpenStack is still often characterized as “difficult” to deploy and manage, and standardizing what it means to be an “expert” will help those who are skilled in using OpenStack better communicate their skill level, as well as those companies looking to deploy OpenStack scout for the necessary talent.
- Ironic, the bare metal provisioning service in OpenStack, also had several sessions, including one on suggested Cinder integration. Through Ironic’s API, users can manage bare metal nodes the same way they would virtual machines. The Project Navigator lists Ironic’s maturity as two out of five, lacking critical features, such as supporting external storage, like SolidFire’s Cinder block storage. Orchestrated infrastructure may lean heavily toward the virtualized and container-ized, but bare metal still has a place (I’m looking at you, databases).
SolidFire continues to be uniquely positioned, with our deep Cinder integration, and next generation architecture that can handle enterprise, virtualized needs of today alongside the self-service cloud needs of the future. Bimodal IT exists for a reason, and as more users move toward or outright embrace cloud-native applications (while evolving and migrating their legacy ones) they’ll need an agile storage platform that can accommodate their shifting strategies and goals.
Three out of the four Superuser Award finalists in Tokyo, and five out of the last eight (Tokyo + Vancouver), are SolidFire customers, validating just how well our storage can meet those needs. Given the every-six-month cadence of OpenStack releases, the sea change currently happening around data center strategy and infrastructure deployment should make for a very interesting summit in Austin next April.
| 1 comments
Posted in OpenStack.