Red Hat CEO Whitehurst on innovation, OpenStack, Hadoop

As computing systems become commoditized, the "profit pools are going to evaporate" for enterprise software vendors, said Whitehurst.
Written by Larry Dignan, Contributor

Red Hat CEO Jim Whitehurst argued that enterprise software vendors are at an inflection point where they'll adapt or falter, noted OpenStack is keeper but needs enterprise support and Hadoop has become a strong open source project that's becoming commercially fragmented. 

I caught up with Whitehurst, who has led Red Hat since December 2007 after being chief operating officer at Delta Airlines, a few weeks ago in San Francisco. Here's a look at the highlights:

What's the state of open source today? Whitehurst said open source has evolved to become a leading innovation engine. "Open source is how innovation is happening. Open source started off making copies of other things, but now it's leading. Big data and software defined networking are open source," explained Whitehurst. "The new stuff is happening in open source."

What's the future of enterprise software? The enterprise software market is in flux and it's unclear how applications will be consumed in the future and how companies will react, said Whitehurst. Today companies such as Facebook, Amazon and Google are ahead of where the vendors are. "So much of enterprise software was due to identifying a problem and then building a custom solution," said Whitehurst. "Increasingly we use common systems."


Whitehurst likens common systems, which are commoditized via open source and cheap hardware, to the standard parts that eventually led to the combustion engine. Once parts were standardized there was a new wave of innovation. As computing systems become commoditized, the "profit pools are going to evaporate" for enterprise software vendors, said Whitehurst. "It's a huge challenge for most infrastructure software players."

At the very least, Whitehurst expects the enterprise software vendor landscape to radically change. The other thread for enterprise software is that it may be embedded into service providers. For instance, the functionality provided by SAP and Oracle could just become part of a broader logistical service from a company like UPS. "The line where ISVs (independent software vendors) and companies begin are blurring," said Whitehurst. "Many companies will be software companies. GE has massive amounts of software but isn't a software company."

Among the current field of software and hardware infrastructure players, only IBM has navigated this level of change before. IBM is the master of moving up the stack. Whitehurst noted that when he was at Delta Airlines, IBM was the only IT vendor who came into a meeting talking about the airline industry and not technology.

The other thread about enterprise software is the landscape is more diverse. "There is a whole breed of software companies developing based on an idea and they build cheaply and deploy on AWS (Amazon Web Services)," said Whitehurst. "These companies don't need much capital, a sales force and distribution because they have taken all the cost out. It's a brilliant model, but enterprise software will be more fragmented."

On Hadoop's future, Whitehurst likens the big data engine to Linux. He also sees similar opportunities for Red Hat. "Hadoop was never written for sale. Its innovation wasn't targeted for the enterprise customer," explained Whitehurst. As a result, Hadoop is commercially splintering as various parties create distributions to better target corporations. One key thing to note, however, is that Hadoop is splintering commercially, but the open source project isn't. "Hadoop will be massively used," said Whitehurst. "I could see Hortonworks and Cloudera becoming the Oracle and IBM DB2 of Hadoop. I can also see Hadoop being embedded everywhere and the value is upstream."


Would Red Hat create a Hadoop distribution? Whitehurst would never say never, but it doesn't make sense for Red Hat to create a Hadoop distribution. "We don't have any intentions right now," said Whitehurst, who noted the Hadoop distribution game is crowded. Whitehurst questioned whether various Hadoop distributions that aim to meld proprietary and open source technologies will be viable in the long run. "To take open source and proprietary stuff is hard. The way you drive a proprietary roadmap is very different than open source. It's hard to do both," said Whitehurst. "You would lose clarity in marketing and the project would be hard to drive the roadmap and work with community." Why would a player like EMC and VMware create a Hadoop distribution via Pivotal then? "Data is where all the value is," said Whitehurst, noting that having the data gives you a say on where it is stored. "EMC sees that value," he said.

Learnings from OpenShift,  Red Hat's platform as a service effort. OpenShift has 250,000 running apps and runs on Amazon's AWS. The biggest surprise so far? Enterprise customers want OpenShift on premise for private clouds. "A lot of customers wanted tighter integration between the hardware and PaaS," said Whitehurst. Red Hat's auto scaling platform is a bet that private and public platform as a service will be an enterprise hit.

OpenStack. Red Hat is also a big supporter of OpenStack, a cloud operating system that was started by NASA and Rackspace. On the surface, Red Hat appeared to be late to the game. Whitehurst said there's a role for integration with OpenShift and OpenStack. He also added that Red Hat was lukewarm on OpenStack 18 months ago. "To Rackspace's credit it created a neutral foundation," said Whitehurst. "OpenStack isn't the first effort around cloud infrastructure, but it has broad support and reference standards."

The promise of OpenStack is that it can abstract the infrastructure layer so companies get control over costs and the hardware used. It's likely that OpenStack will ultimately enable customers to automatically swap between providers such as Rackspace and AWS and their own infrastructure. "The theory is that in the next several years that the application layer will be agnostic," said Whitehurst.

Whitehurst sees Red Hat's role in OpenStack similar to its position in Linux. Red Hat will provide support and stability. "We will support an enterprise version and commit to it for 10 years," said Whitehurst, noting the exact timeline may be different. The reality is that OpenStack has new functionality and releases at a rapid pace. If you're an enterprise, you're not going to backport into every old version. That's where Red Hat comes in," said Whitehurst. "We have the credibility and knowhow to support OpenStack at scale. We'll win there," said Whitehurst. "Our biggest allure is long-term stability. We won't break the interface and won't break binary compatibility."

Does that make Red Hat a legacy vendor sans innovation? When asked if Red Hat is now part of the establishment, Whitehurst chuckled. "Well you don't get fired for buying Red Hat and we're growing solidly," said Whitehurst. However, Red Hat can't be considered a legacy player when it is contributing to efforts like OpenStack that will redefine the way IT runs. Add it up and it's a bit hard to define Red Hat. If open source is leading the way, Red Hat is a key player driving the revolution to commoditize technology. On the other side, Red Hat bring stability that enables CIOs to plan ahead. My verdict is that Red Hat is a tweener. Whitehurst seems pretty comfortable with that role.

Editorial standards