Rackspace Technology has admittedly been relatively quiet in recent years when it’s come to OpenStack, the open source cloud infrastructure platform that was born in 2010 out of the collaboration between the cloud computing company and NASA.
It’s not that Rackspace turned its back on OpenStack. It’s contributed more than 5.6 million lines of code to OpenStack, is among the largest OpenStack cloud providers, has more than a billion server hours of experience operating production-ready OpenStack clouds at scale, and has more than 150 OpenStack experts and 50 Kubernetes administrators.
However, in recent years had withdrawn to some extent from its commitments to the OpenStack Foundation, according to Kevin Carter, product director at Rackspace. But that’s changing. Rackspace this week is reaffirming its commitment to OpenStack, illustrated by the launch of OpenStack Enterprise, a fully managed cloud offering aimed at critical workloads that run at scale and that brings enhanced security and efficiency.
After co-founding OpenStack and the OpenStack Foundation – which took over management of OpenStack in 2012 and evolved into the OpenInfra Foundation – Rackspace was all in on the technology, Carter tells The Next Platform.
“We did push really, really heavily into the upstream codebase,” he says. “And we also pushed into our own internal systems and platform for our public cloud. As a first mover, we had a lot of problems and we had to solve them upfront and then work in some of the community spaces and all of those different things. When we say that we’re recommitting to it, we’re really recommitting to that upstream. We have been quiet for a few years in terms of our upstream commitments and we’re now coming back to it.”
That includes “recommitting to the Foundation, recommitting to working upstream and recommitting to getting our developers and engineers now comfortable working in that open upstream space again, where they have been kind of isolated for a few years.”
Carter has a seen the Rackspace-OpenStack relationship from both sides. This is his fourth turn with Rackspace that, added up, amounts to more than eight years with the company in total. He also did two tours with OpenStack that spanned a cumulative eight-plus years.
OpenStack’s Rollercoaster Ride
OpenStack has seen its share of ups and downs in an industry that demands technologies adapt or get left behind, and there have been times during the past 14 years that OpenStack seemed to be headed to the glue factory. During the heady days of the cloud, enterprises rushed to move their workloads into the public clouds from the likes of Amazon Web Services, Microsoft Azure, Google Cloud Platform, and other public cloud providers, lifting the responsibility for managing the actual infrastructure off the organizations.
And then those cloud providers, seeing an opportunity of play on both sides of the hybrid cloud fence, began offering products and services – AWS’s Outposts, Azure Stack, and Google’s Anthos, for example – for on-premises datacenters. Where was the room for OpenStack?
Despite this, the OpenInfra Foundation has continued to evolve OpenStack. It’s now based on Kubernetes, opening it up to containers, and the infrastructure is being modernized to enable the infrastructure to run generative AI workloads. Dawn, the UK’s fastest supercomputer, will run an OpenStack variant, Scientific OpenStack, developed by StackHPC.
The Changing Cloud Environment
In addition, enterprises are adopting multi-cloud strategies, with 89 percent embracing it – up from 87 percent last year – according to Flexera’s annual State of the Cloud report, which means OpenStack may get more looks as the provider of on-prem cloud infrastructures. In the same vein, cloud repatriation is on the rise as organizations try to get a handle on costs associated with the cloud. Citrix noted in a study that 42 percent of US organizations have moved or are considering moving at least half of their cloud-based workloads back into on-prem infrastructures.
Rackspace’s Carter says he remains bullish about OpenStack, adding that the company is “very confident in its ability to serve our workloads” and that the technology’s mission has changed with the times.
“A long time ago it wanted to be the end-all and be-all and now it’s more of an integrated platform into a lot of different solutions,” he says. “It’s always been a pluggable back end. It’s always worked with enterprise storage solutions and things like that. It’s always had very robust networking capabilities in the L3 space, especially early on in Quantum and then later Neutron. There were millions of drivers all the time.”
However, that has changed as the open source space has changed.
“It’s more stable state now,” he says. “There are a few different drivers. The nuanced things that were being experimented with and tried with back in the day, they’re not happening in the open tech space anymore because it’s more or less stabilized. You are seeing those things still being tried and kicked around. And just other technology streams, be it Kubernetes or whatever the flavor of the day is. We still are seeing a ton of innovation in the open tech space, but it is different. They’re iterating on how they can make the services better, more robust, how they can extend. They’re trying to build up that core technology.”
One OpenStack to Rule Them All
Rackspace is looking to help. The company points to recent upstream code contributions to the OpenStack community that dovetail with its One OpenStack strategy, which is the idea of creating a single technology stream for OpenStack products to make it easier for enterprises to embrace each offering without having to address technical hurdles between them.
That’s part of the idea behind Rackspace’s OpenStack Enterprise cloud offering, which is in some ways a rebranding of its Public Cloud and Private Cloud platforms to address an increasingly multi-cloud world. More features and capabilities will be folded into OpenStack Enterprise in the coming months.
“The way we’re positioning this, you’ll see this throughout the year where we’ll have a few more launches of new products and things that are coming out,” Carter says. “We have a ton of stuff in our roadmap. Enterprise is really the foundation for it, where we’re taking our enterprise foundation and building out all of our different offerings off of that. We color that as our One OpenStack strategy. We’re really pivoting and pushing OpenStack as a multi-cloud solution, giving people that single-tenant, multi-tenant environment as well as hybrid, where you can pivot back and forth between them.”
Coming with OpenStack Enterprise are a range of add-on services, including data protection and virtual machine management, and being more descriptive about how systems and services run, he says. OpenStack historically has covered core components, but enterprises have long demanded more capabilities and flexibility. OpenStack Enterprise embraces some features and functionalities, such as Octavia for load balancing and BBQ for key management.
“There are a lot of magnets for Kubernetes,” Carter says. “While we may offer the Rackspace managed platform for Kubernetes as a product offering that we can support on top of it, if customers want to use that open source Kubernetes tool as well, we want to be able to give them that choice. ‘I want managed services on this one. I’m going to run it myself over here on this other one.’ It’s really leaning into the flexibility. There are a lot of new things – using Skyline [an OpenStack dashboard] instead of Horizon for just as a better user experience overall.”
It’s giving organizations the chance to use services that “are not otherwise foundational components. It’s no longer just a virtualization platform. It really is a datacenter management suite as a resource manager,” he says.
Well, IMHO, if it were a beer, it might be called the Bengali-Dane Amber-Stout (SNL/NNSA CTS-2, Top 500 #151-130-98-107), a perfect complement to Manzano olive Magma appetizer (CTS-1, #195-269) (historically: https://www.nextplatform.com/2017/08/10/look-inside-u-s-nuclear-securitys-commodity-technology-systems/ , and more recently: https://www.nextplatform.com/2023/10/02/the-first-peeks-at-the-doe-post-exascale-supercomputers/ ). I guess the spaghetti was on the wall all along in this pastafarian plot of disaggregated heterogeneously composable commodity supercomputing beer ingredients, and the related rise of tentacular software defined cephalopods, both heeding the ineluctable multi-cloud ascent of the ZettaCthulhus! It could be a new Dawn in HPC/AI fermentation technology, with Turtle-powered Sharks all the way down! 8^p ( https://www.stackhpc.com/magnum-cluster-api-helm-deep-dive.html )
Deftly so, I reckon … no horsing around here, with MS Azure Eagle leading the rodeo (#3), Meta’s Grand Tetons piercing through the AI clouds, and herds of compute machinery ranched at Google, AWS, xAI … plenty of opportunities for Rackspace to excel-in at this here AI County Fair Bonanza: Tame the wild Enterprise multi-cloud longhorn, steer the OpenStack sirloin to Octavia’s BBQ, ride into the Skyline sunset … just got to get up one more time than you’re knocked down! (famous quote)
Archer2 at EPCC is the UK’s fastest Supercomputer: #49 in the latest top500 list, as opposed to #51 for Dawn. Dawn was procured specifically as part a new AI Research Resource as opposed to HPC, thus potentially making it the “fastest” AI supercomputer in the UK https://www.hpc.cam.ac.uk/d-w-n)?