The latest edition of the OpenStack framework released this week adds an ability to reserve instances of graphical processor units (GPUs) along with a revamped user interface (UI) that promises to make the cloud management framework more accessible.

In addition, release 2024.2 Dalmatian of OpenStack Ironic adds a runbooks application programming interface (API) through which a series of curated workflows can be defined, along with any ability to manage updates to storage resources more granularly.

Finally, Ironic, a tool for provisioning bare metal servers has been further hardened to better secure passwords. Additionally, Neutron, a network virtualization module, provides access to a manager role to its application programming interface to enable role-based access controls (RBAC), while Nova controller fabric now automatically detects virtual Trusted Platform Module (vTPM) and requires Transport Layer Security (TLS) connections when using SPICE consoles.

Thierry Carrez, general manager of the OpenInfra Foundation, said an OpenStack resurgence that began at the beginning of this year is being driven by three separate trends. The first is the rise of generative artificial intelligence (AI) organizations are looking for frameworks that make it simpler to manage compute resources across a hybrid IT environment. The extension to the Blazer resource management tool added with this update will make it simpler for IT teams running AI models to manage the allocation of limited GPU resources, noted Carrez.

At the same time, data sovereignty laws are driving more organizations to deploy workloads in on-premises IT environments.

Finally, the changes to VMware licensing terms that Broadcom has enacted have driven more organizations to explore alternatives.

According to Open Infra, there are also more than 45 million OpenStack cores running in production environments, and 487 contributors submitted 7,640 changes over the past six months alone.

It’s not clear to what extent many IT teams are dependent on OpenStack, but as IT environments become more distributed many organizations are looking for ways to centralize the management of IT. OpenStack creates an opportunity to manage multiple platforms using a single team rather than having to hire and retain separate IT staff for every platform employed. While OpenStack itself requires expertise to deploy and maintain, that approach can reduce the total labor cost by reducing the total amount of labor required to manage a distributed computing environment.

One way or another, a reckoning with IT complexity is on the horizon. In theory, advances in AI should make it simpler to manage frameworks such as OpenStack. The issue now is finding a way to make that transition over time. It’s not feasible to switch from one management framework to another overnight.

Of course, there are also new workloads being regularly deployed. Some IT organizations may decide to draw a line in the sand that determines from here on out all new workloads will be managed via OpenStack.

Regardless of approach, organizations are more sensitive to being locked into specific platforms either because of proprietary APIs or onerous contract terms. As such, many decisions made years ago are now being revisited with a more jaundiced eye.