By Jean S. Bozman
The Open Stack Foundation (OSF) is changing its structure by adopting a new framework that will allow it to support open-source projects that did not originate with OpenStack. This move will allow OpenStack to bring in dozens of new projects for incubation within the OpenStack Foundation incubation within the OpenStack Foundation, and testing for end-to-end operations. By doing so, the OpenStack Foundation plans to make its cloud infrastructure building blocks easier to integrate with non-OpenStack projects, easier to consume – and faster to deploy.
As announced at the OpenStack Berlin conference, the changes will result in cloud infrastructure that will be leveraged by enterprise customers, telcos, CSPs – and by software ISVs that integrate OpenStack software in their products.
The worldwide momentum in hybrid cloud brought this challenge to a decision point – as private clouds built with OpenStack software meet public clouds built by CSPs (e.g., AWS, Azure, GCP, Baidu, Alibaba). Cloud service delivery now depends on containers and orchestration – and hybrid clouds must connect smoothly, via standard APIs, to ease migration of workloads that span public and private clouds. OpenStack will continue to leverage an existing four-step testing/certification cycle to ensure easier integration for hybrid cloud software layers.
After eight years of building open source for IaaS (compute, storage, networking), OpenStack plans to address a broader array of cloud infrastructure for enterprise deployments. As announced at OpenStack Berlin, four new pilot projects are addressing DevOps, deployment scalability, security and project management. There will be even more support for the Kubernetes orchestration layer – which is already widely used by most OpenStack users – and more support for Ceph projects that originated with the Linux Foundation.
OpenStack had faced a tough dilemma: Focus on developers’ enthusiasm for branching innovation and starting new projects, while listening to business customers that wanted to make OpenStack layers easier to use and deploy for production workloads.
Steps Taken by OpenStack
To address these goals, OpenStack has taken the following steps:
- Renamed its semi-annual conference as the Open Infrastructure Summit. This broader framing of cloud-building categories will make it easier to apply the OpenStack open-source community’s projects to a wider range of technologies for hybrid cloud. As the OpenStack Summit web page describes it: “The open infrastructure landscape is changing and so is the Summit. Now that users are integrating dozens of open source tools into a modern stack that reaches well beyond the scope of OpenStack, we re-organized the [Summit] event to focus on specific problem domains.” The domains include: CI/CD, containers, Edge, GPU, and HPC.
- Updated four major pilot projects that will be managed by the OpenStack Foundation, along with the OpenStack open-source community projects. The new projects, launched since December 2017, are: Airship for lifecycle management; Kata Containers for secure containers; StarlingX for development of Edge applications; and Zuul, a project gating system for CI/CD continuous development teams and deployment scalability. See these links for more details:
- Reduced the number of small projects under development. It combined emerging technologies into fewer group projects – ending the “Big Tent” of dozens of smaller projects via streamlining and simplification.
- Increased its focus on China and Asia, given high levels of interest in open-source in Asia/Pacific. In OpenStack’s most recent customer survey, Asia respondents accounted for 48% of nearly 1,500 respondents in the survey. That’s up from 24% of survey respondents in 2016. Importantly, OpenStack Foundation executives plan to have the 2H19 conference in China – the first time it’s been hosted there.
- Placed more emphasis on orchestration, management and automation. This addresses Opex costs related to developer and sysadmin’s time and effort, including platforms, DevOps, management, and automation software.
- Building for the edge. Edge computing has a higher profile in OpenStack projects this year – reflecting growing customer investment in edge-based sensor data for business analytics (e.g. telecom, manufacturing, oil/gas production, and health care). This emphasis on the edge is expected to grow, as AI/ML analysis of data generated at the edge is driving new forms of business competition.
The very nature of open-source communities, with constant interaction between thousands of developers, led to a rapid expansion in the total number of projects managed by the OpenStack Foundation. It appears that the OpenStack board, comprised of 24 board members, is preserving the innovation of open-source community development, while applying more structure, streamlining overlapping efforts, and simplifying DevOps and deployment for customers. The board meets monthly – and the changes seen at this OpenStack meeting are clearly evident.
The Berlin conference left no doubt that there had been deep, behind-the-scenes discussion about a new trajectory for the organization. Some might once have said that OpenStack technologies were directed at the very largest companies, such as AT&T and Verizon and Walmart, to implement. However, it is becoming clear that OpenStack executives believe that the technologies must become more widely accessible and easy to deploy for mid-size and smaller companies.
Wider acceptance and adoption of OpenStack are vital to its future. Up to now, OpenStack customers have looked to major releases twice a year. That may still be the case, but introduction of new functionality within the projects may proceed more quickly, because it is now possible to upgrade across multiple OpenStack releases. Next year – 2019 – will be the proving ground for the recent changes in OpenStack’s approach to building –and delivering – open source infrastructure for enterprise customers, telcos building cloud infrastructure, and software ISVs.