What Happens Next As OpenStack's Hype Bubble Bursts?

OpenStack, the cloud-building technology that won concerted industry backing to challenge proprietary virtualization vendors like VMware, and public cloud operators like AWS, is now fighting for survival in a brutal and rapidly evolving IT market.

Boosters and critics agree that early exuberance around the project has given way to questions of long-term relevance in the face of rapid public cloud maturation.

"As far as the excitement around OpenStack that was present two, three years ago, that's done," Boris Renski, a founder of Mirantis, the largest pure-play OpenStack vendor in the industry, told CRN.

[Related: The Great OpenStack Debate: A Powerful Platform That Will Revolutionize The Data Center? Or A Dud?]

Now that the hype has evaporated, what remains is a highly capable technology looking for a suitable deployment model and market niche, said Renski, a director on the board of the OpenStack Foundation, the organization that guides the project.

A joint effort launched seven years ago by Rackspace Hosting and NASA, OpenStack shot to prominence with the promise of a robust, comprehensive cloud technology supported by a vibrant corporate community building its many interconnected components. But critics say the sprawling nature of the project, and diverse interests of its contributors cooled actual market enthusiasm.

Looking back, many in the OpenStack community peg the six months between the Austin OpenStack Summit of April 2016 and the following conference in Barcelona as the period in which a hype bubble, inflating for years around the open source cloud technology, finally burst.

The Texas-sized enthusiasm that characterized Austin gave way to a subdued affair the following October in Spain, Renski said.

Vendors like Mirantis have since recalibrated their strategies to capitalize on the inherent benefits of OpenStack while mitigating the deficiencies that leveled the adoption curve.

First, they're trying to ease delivery of the notoriously difficult system, recognizing operational challenges sidelined many enterprise OpenStack initiatives and prevented early proof of concepts from converting to production deployments.

Companies including Mirantis, Cisco, Rackspace, and Canonical, as well as a recent crop of startups like Platform9 and ZeroStack, are promoting a new paradigm: call it Managed OpenStack or OpenStack-as-a-Service.

"Interest from enterprise customers has never been higher, but that has been limited by the fact that OpenStack has been really, really hard. And the only way to make it easy is to deliver it as a service," said Sirish Raghuram, CEO and founder of Platform9.

"There are exceptions, but enterprises should not be in the business of trying to run OpenStack. They're just not skilled for that," Raghuram told CRN.

Canonical, a major player in the Linux market with its Ubuntu distribution, for its part is focusing on automating the upgrade process for customers that have become "stuck" in outdated releases they've extensively customized for their own environments, said Mark Baker, product manager for Ubuntu OpenStack.

"We believe that OpenStack can power efficient, scalable private clouds if they’re built and managed well," Baker told CRN.

The channel has seen up close customer hesitance resulting from OpenStack's complexity and the persistent scarcity of skilled IT professionals to operate it, David Linthicum, senior vice president of Cloud Technology Partners, a solution provider based in Boston, told CRN.

Those factors made the open source software too costly to implement for most businesses, he told CRN.

"We're just not seeing it that much anymore," Linthicum said of OpenStack. "May have seen it a couple years ago, got involved with some private implementations. The reality is the customers aren't as interested as they used to be."

Partners noticed when the hype bubble popped, Linthicum said, and have interpreted recent actions by name-brand players, primarily HPE and Cisco Systems, as indicators of weakness in the market.

In 2014, Hewlett Packard Enterprise, at the time the most-prolific contributor to the project, pledged $1 billion in spending over two years on OpenStack products and services. But earlier this year the tech giant sold off its OpenStack assets to German software developer SUSE.

HPE still offers its Helion-branded OpenStack solution through an OEM deal with SUSE, now a 'preferred partner,' and has also invested in Platform9's latest funding round.

SUSE's own research shows the SUSE and HPE Helion install base accounts for more than 22 percent of OpenStack deployments, according to Peter Chadwick, SUSE's director of product management.

Cisco, at the start of this year, shuttered its Intercloud public cloud, which ran on OpenStack. But the networking giant remains committed to OpenStack through its Metacloud brand, a managed on-premises solution, Lew Tucker, Cisco's vice president and CTO for cloud computing, told CRN.

"We see OpenStack continuing to evolve into an open source platform for a multi-cloud world as cloud-native applications and services expand across virtual machines, containers, and bare metal," Tucker, vice chairman of the OpenStack Foundation, said.

To power its UCS line of converged servers, Cisco also offers Red Hat's OpenStack distribution, he added.

Red Hat has mastered jumping into open source projects and imparting its imprimatur for enterprise consumption. The company has bet on OpenStack's future, ascending to become the largest project contributor, as well as a market leader.

But even Red Hat CEO James Whitehurst told CRN the technology had entered the "trough of disillusionment"—a stage of Gartner's Hype Cycle describing the adoption of new technologies.

"I think a lot of the really big players, the large OEM types, thought OpenStack is my solution to public cloud," Whitehurst told CRN in March.

"They thought, if I'm going to lose infrastructure to public cloud, this is what I'm going be able to say is an alternative. And OpenStack is not an alternative to the public cloud."

Into The Trough

"When it first came out, a lot of people jumped on that bandwagon," said Kyle Betts, director of cloud integration at Technologent, a solution provider headquartered in Irvine, Calif.

"We thought it would be a great idea, something we want to leverage," Betts said of OpenStack.

As an open source alternative to VMware, one with a pedigree from NASA and serious buy-in from industry giants, OpenStack looked like a unique opportunity for the channel, he said.

Dennis Smith, a cloud analyst at Gartner, told CRN that IT professionals immediately recognized OpenStack was a technology that would require extensive engineering talent to leverage.

"But that meant transferring the capital costs of a VMware license, per se, to the operating expenses of paying engineers to manage the environment," Smith said. Many customers and channel partners were happy to see such a shift.

Mirantis and a host of other pioneering vendors, especially Nebula, a company founded by former NASA CTO Chris Kemp, immediately became industry darlings.

"I mean three years ago, we, Mirantis, would make a press release with the word OpenStack in it," Renski said, and "journalists would be chomping at the bit."

"Now the world has moved on, and it's no longer very interesting." But neither is a new Linux customer or version, Renski added, which doesn't reflect on the value of that technology.

In those hype-filled days when OpenStack was taking off, proof-of-concept clouds abounded across enterprise data centers. Big wins, like PayPal deploying the software, got widespread public attention.

Later it became evident most of the proof of concepts were on small clusters that never materialized into production deployments. Many early OpenStack initiatives lacked business value and substance, Renski said.

When in 2015 Nebula went out of business, while other OpenStack startups got acquired for modest sums, it became clear OpenStack had lost its luster.

"When you're flying high, and everybody is excited and wants you, then in a year-and-a-half you become not a cool thing anymore, it is disappointing. It carries emotional notes and propagates into some of the things that happen in the field," Renski said.

OpenStack's well-known technical challenges did a lot to hinder the transition from hyped darling to viable business solution, said Smith, of Gartner. Smith once tried to stand up several open source clouds on his own. Eucalyptus and CloudStack, two IaaS solutions that once looked like serious rivals to OpenStack, he had up and running over a long weekend. He couldn't do the same with OpenStack, he said.

Those competing solutions were easier to deploy because of their monolithic architecture, Smith said, as opposed to OpenStack's makeup as a series of projects developed by a vast ecosystem. And a shortage of professionals with the unique skills necessary to deploy and manage the unruly cloud operating system became a perennial problem, Smith told CRN.

Ariel Maislos, CEO of Stratoscale, a company that develops rival cloud infrastructure software, said OpenStack simply was not able to deliver on its initial promise.

"It became this bucket of technology and not a coherent product," Maislos told CRN. "It's designed for developers, not designed for IT managers."

That's anathema to the channel, he said, which thrives on repeatable transactions, not custom engagements.

Early implementers did see some consulting revenue. "If you were lucky you were able to extract professional services dollars," Maislos said. "Someone wants to install OpenStack, but they don’t have the knowledge, you bring in the cavalry."

But many of those vanguard customers became disillusioned, Maislos said, and either went back to VMware or migrated to public cloud. Despite the exuberance of early adopters, OpenStack never got beyond a core group of users. "The channel is looking for transactions, repeatability. [OpenStack] became this monster that's difficult to deliver and not a scalable strategy," Maislos said.

The rapid maturation of public cloud over that same period further stymied market penetration. OpenStack was promoted as an alternative to Amazon Web Services and other providers, but those adversaries became much more imposing faster than anyone expected, according to Linthicum, of Cloud Technology Partners.

"If we price the two out, we can always find a better deal on public cloud than on private cloud," Linthicum said. Implementing any private infrastructure these days "becomes the path of most resistance." Interest in OpenStack has waned "because AWS, Google, and Microsoft are eating their lunches now," Linthicum said of the three hyper-scale cloud leaders.

Industry giants like Intel, IBM, HPE and Cisco that invested in OpenStack as a vehicle to collectively push back against the hegemony of those public cloud juggernauts, and limit their buying power in the hardware market, eventually recognized the weakness of that strategy, said Red Hat's CEO.

"I think the large hardware vendors who saw a lot of market share shift to Amazon, which was using commodity stuff, saw that as their savior," Whitehurst told CRN. "And it's not. It's not."

Path To Enlightenment

The OpenStack community has learned some hard lessons over the last 18 months. Now, more grounded and focused, OpenStack enthusiasts are trying to progress the technology toward the Slope Of Enlightenment – the next phase of Gartner's Hype Cycle.

"When [technologies] come out of the Trough [of disillusionment], then people are much more open-eyed about them," said Smith, the Gartner analyst. "They view where they can use them."

While OpenStack is "unlikely to become the de facto IaaS private cloud like many people thought four, five years ago," Smith said, it also doesn't look destined for a dead end. Instead, the technology will likely find its footing as a private cloud solution well-suited for certain industries and use cases.

Powering the data centers of telecommunications providers is one niche that's caught traction, Smith added.

Telecoms, often running Network Functions Virtualization (NFV) environments, have found OpenStack to be a solution sophisticated enough to enable modernization of their legacy footprints, Smith told CRN.

Mirantis, in particular, has seen success addressing that market, most notably a large deal with AT&T it began implementing in earnest last year.

OpenStack might also emerge as the preferred environment for hosting on-premises Docker container deployments, an increasingly popular architecture for cloud-native applications, Smith said.

The technological complexities aren't going away anytime soon, Smith said, but there are still compelling motivations for OpenStack adoption, starting with cost.

Organizations looking for "run-of-the-mill IaaS" are likely to find cheaper options, he said. But the price of the actual software—free—means at least a subset of enterprises with advanced cloud needs could pay a lot more to IT staff and solution providers capable of installing and maintaining OpenStack and still see overall savings.

A large OpenStack user recently told Betts, of Technologent, if he quadrupled his current managed services bill, he'd still be paying only a third of the cost of a comparable VMware license.

For telcos and other scaled services providers that rigorously scrutinize hosting costs to remain competitive, OpenStack presents an attractive option, Betts said. Large enterprises, less likely to scrape savings at the expense of incurring operational headaches, "just need enabling technologies."

"My customers who know how to use it still don’t want to operate it. They don’t want to invest the time or energy. They're literally ready to just consume," Betts told CRN.

"If they can just use it like a platform, it’s a win-win for all of us. Managed OpenStack. It’s the only way I'm seeing people be successful with it right now," he said.

Mirantis has transformed its business around that very same premise, Renski said.

The company will no longer even sell a subscription for patches and new releases alone, having learned the hard way that road too often leads to customer failure.

"The model of selling the pieces has been put to death by the success of public cloud," Renski said.

Mirantis now insists on "taking ownership of the operations piece" with an extended contract for management services. Customers need to focus on the business problem that brought them to OpenStack before worrying about hiring professionals to self-service their infrastructure, Renski said.

While Mirantis has stared down "a large degree of disillusionment and unpleasantness" over the last 18 months, the company's founders believe those tribulations have led them to discover the right business model for the product, Renski said.

"People don’t really want to buy software building blocks and stitch together," Renski said. "They'd much rather have somebody else just take care of that for them, in an Amazon data center or their data center, and they just consume it at higher levels of the stack."

It's important for the industry to figure out the best delivery model, he said, because OpenStack, for all the snags encountered of late, as an open cloud standard is still unique in the market, he said.

That realization is fueling a fresh wave of optimism, free of hype.

In May, at the latest OpenStack Summit in Boston, the mood of the community again had shifted.

"Austin was the coolest thing ever, and Barcelona was the dreariest," Renski said. "But Boston, which just happened, was a meaningful uptick in my view."