A common approach to software releases in the cloud and elsewhere is to roll out a series of minor releases that have limited support terms, followed by a major release that has long-term support. That's the model that Red Hat has been using for years with its Red Hat OpenStack Platform, but that is changing following the release of version 15 of the platform on Sept. 19.
There are two OpenStack releases in any given year, and Red Hat had been tracking them with releases of its own platform, providing additional enterprise-grade usability features. Previously, every third Red Hat OpenStack Platform release would be labeled as long-term support, with the intervening releases labeled as minor releases, each getting only one year of support.
"[Red Hat OpenStack Platform 15] is a short-life release, meaning that we will support [OSP15] for a total of one year," Sean Cohen, senior manager of product management for OpenStack at Red Hat, told ITPro Today. "Our next OpenStack 16 release will be a long-life release based on the upstream OpenStack Train release, with support through 2025."
Moving forward, he said, the model will be to have "dot" releases, so the next release will be 16.0, then 16.01, 16.02, etc. Updates will be on the OSP16 version but with functionality added based on customer use cases.
Red Hat OpenStack Platform 15 is based on the OpenStack Stein milestone that was released by the community on April 10. Unlike prior releases, which had a specific Red Hat Enterprise Linux kernel customized for OpenStack, the new release is based on Red Hat Enterprise Linux 8. Cohen said Red Hat Enterprise Linux 8 provides a solid foundation upon which to build OpenStack innovations and provides consistency of operations across enterprise IT footprints.
Building on top of the upstream OpenStack Stein release brings in new features to the platform as well, one being the vGPU feature, Cohen said. The vGPU capability can be used to run workloads, like the TensorFlow open-source machine learning framework, that will comprise multiple virtual machines (VMs) leveraging a single GPU.
"vGPU support allows applications to have managed access to hardware acceleration, enabling near-real-time decision-making and machine learning," he said. "Whether it’s security at a sporting event or quality-control in manufacturing, vGPU support helps make a range of use cases possible."
On the security front, Red Hat added hardware security module (HSM) back-end support for enhanced key generation and management, according to Cohen. In addition, the company introduced efficient kernel tracing monitoring, which allows programs to be injected at different hooks in the kernel to monitor processes, CPU usage and cache usage. Security also benefits from the new crypto offload capability for IPsec VPNs that frees up CPU resources by utilizing NIC resources.
An additional innovation area in the Stein release is storage at the edge. As part of Stein, each cluster at the edge can have its own Ceph object storage cluster, which allows for additional use cases where data locality is required, Cohen said. Another notable Edge compute feature is the new Glance image service caching feature that can reduce boot times and bandwidth usage.
"These features lay the groundwork for future hyperconverged nodes at the edge, which can combine compute and storage into a single appliance," he said. "The size, combined with OpenStack’s scalability and management, can help make edge deployments faster, more simple and simpler to manage."
The next major upstream release will be OpenStack Train, which is scheduled to debut in the middle of October.
Among the new features that will land in OpenStack Train are full IPv6 address support and enhanced migration capabilities for virtual machines, Cohen said.
"From the Train release, we see new additions such as persistent storage at the edge, new Ceph capabilities, performance monitoring, as well as bandwidth management as key functionality that we feel helps our customers not only evolve to new emerging technologies such as edge computing, but to also help in Day Two operations management," he said.