eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
VANCOUVER, British Columbia—Ever since NASA and Rackspace first got together in 2010 to create OpenStack, there has been the concept of an integrated OpenStack release. However, at the OpenStack Summit here this week, developers declared the integrated release model to be dead, being replaced with a new “Big Tent” model that redefines what OpenStack includes as a platform.
The OpenStack integrated release included multiple projects and was growing, but the challenge was that not every new project made sense for every OpenStack cloud and not every new project was actually being deployed. In a session at the OpenStack Summit with press and analysts, Jonathan Bryce, executive director of the OpenStack Foundation, explained that with the new DefCore definition there are projects that are considered to be core and mature that are widely adopted, and then there is everything else. Among the projects that are core are Nova compute, Swift storage, Glance image, Horizon dashboard, Neutron network and Keystone identity.
The recent OpenStack Kilo integrated release also included Heat orchestration, Ironic bare metal service, Trove database and Ceilometer metering projects, which are not part of the DefCore definition.
“The integrated release was a technical management artifact, and it is now going away,” Bryce said. “The ‘integrated release’ term had become overloaded and meant something that it was not intended to mean.”
Some in the community were of the opinion that if a project is part of the integrated release, it needs to be supported and deployed by OpenStack vendors to be considered OpenStack and interoperable with other OpenStack clouds. However, Bryce said the projects that made it to the integrated release really were just the ones that had reached a certain level of development maturity and integration with other components in the OpenStack platform.
In a standing room-only session titled “The Big Tent—A Look at the New OpenStack Projects Governance” at the OpenStack Summit, Thierry Carrez, director of engineering at the OpenStack Foundation, and Sean Dague, software engineer at Hewlett-Packard, detailed the impact of the new approach.
The integrated release model created a lot of confusion for both developers and vendors, Carrez said. Plus, it was becoming increasingly difficult to continue to add new projects into the integrated release, he said.
Dague explained that with the Big Tent model, the OpenStack project process is being refactored in a subtle but very important way.
“We’re changing the question from ‘Is this OpenStack?’ to ‘Are you OpenStack?'” Dague said.
What that means is that rather than having “blessed” projects that graduate to becoming part of an integrated release, projects in the Big Tent are defined by their adherence to using OpenStack processes. That means a project needs to be open-source licensed, have an open community and make use of the OpenStack build model.
What moving to a Big Tent model means for the upcoming OpenStack Liberty release, set to debut in October, is that there will be many more projects that could be part of a coordinated release event. Among the projects that were never part of an integrated release but will now fall under the Big Tent are the Congress policy, Manila file sharing, Rally benchmarking, Magnum containers, Barbican key management and Murano app catalog projects.
Carrez noted that many of the projects in the Big Tent model have been behaving as OpenStack projects for some time, but previously were considered to be second-class citizens and didn’t get as much space at design summits as first-class integrated projects. That said, the OpenStack Foundation Technical Committee will have a focus on core projects, while still enabling a broad community to grow and support diverse user needs, he said.
There are some challenges around moving to the Big Tent model. For one, comparing the upcoming Liberty release to prior integrated releases in terms of lines of code and contributors will be more difficult as the number of included projects will change significantly.
In addition, Carrez admitted that there might be a number of projects that could overlap or potentially compete with each other. While different approaches to solving a problem are welcome in the OpenStack community, there will be some controls.
“The OpenStack Technical Committee still gets a final yes or no about OpenStack projects, and if something would damage OpenStack, we won’t let it live in the community,” Carrez said.
Sean Michael Kerner is a senior editor at eWEEK and InternetNews.com. Follow him on Twitter @TechJournalist.