Gluster was founded in 2005 to productize their eponymous global distributed filesystem, GlusterFS. As an all-software solution for storing immense quantities of distributed and replicated data, it quickly caught the eye of many working with big data including Red Hat, who purchased the company late last year.
Yesterday at Red Hat Summit, Red Hat officially announced their branded solution based upon the Gluster foundation: Red Hat Storage. As an all-software storage solution, Red Hat Storage lives atop the Red Hat Enterprise Linux operating system and provides robust scale-out storage for both block and file level volumes.
Prior to its acquisition, Gluster had raised over $8 million in funding. Red Hat bought them for $138 million. That’s a pretty remarkable jump. I spoke with a number of Red Hat executives about the acquisition, specifically asking if they felt they’d gotten their money’s worth.
Ranga Rangachari, VP and general manager of Red Hat’s storage division, observed that the commoditization of servers — from expensive proprietary UNIX systems to low-cost nearly disposable x86 systems — set a precedent, and that the same transition was happening in the field of storage. Expensive proprietary storage solutions were about to be replaced by commodity, standardized storage options like Red Hat Storage. “Standards drive volume, and volume drives standards,” Rangachari said.
Rangachari also told me that the acquisition of Gluster was only partially about the technology. It was also about the community that had formed around the GlusterFS technology. As GPL-licensed code, Red Hat couldn’t easily prevent continued open source development after their acquisition, even if they had wanted to. Rather, Rangachari said, Red Hat wanted to leverage their experience with stewardship and leadership within open source communities to ensure that the GlusterFS technology continued to flourish, while simultaneously developing an enterprise product that Red Hat could market to customers.
Rangachari shared his understanding of the three pillars of Red Hat’s strength when it comes to acquisitions: “unify, simplify, and certify.” Unification means, really, integration with the broader spectrum of Red Hat’s enterprise products. Certify, in the specific case of Red Hat Storage, really means to set customer SLA expectations for the product. As an all-software solution running atop Red Hat Enterprise Linux, there’s little for Red Hat Storage to require, other than a supported hardware platform for the underlying OS. Red Hat has already paved this road with their Hardware Compatibility List for RHEL, which includes all the usual suspects (HP, IBM, Dell), along with newly certified SuperMicro.
I next asked Brian Stevens, Red Hat CTO, about the acquisition. According to him, Red Hat had been pursuing the Gluster purchase for nearly a year. The company had a low cost profile: they had focused on the technology, which was validated by the robust community that developed around it, but had failed to build a terribly successful business.
Stevens shared with me that Red Hat was working on a Plan B, should the Gluster purchase have fallen through, but that plan was a very distant second choice. Ceph, another all-software distributed filesystem recently integrated into the mainline Linux kernel, was, in Steven’s opinion, several years away from having a viable product for enterprise customers. The purchase of Gluster allowed Red Hat to quickly get to market with their storage solution.
As with just about every Red Hat product, the core technology is open source and available for free to interested developers and users. Gluster remains a standalone open source product. The difference between Gluster and Red Hat Storage is primarily one of support: customers of the latter get the full breadth of Red Hat engineering support services to resolve any problems they might experience.
Red Hat as a company continues to encourage choice for their customers. Unlike, say, Oracle who is trying to drive all their database customers to use Oracle Enterprise Linux as the only supported platform, Stevens told me that Red Hat would not gain dominance by removing customer choice. They would win over their competitors by offering technically superior solutions at lower cost, rather than by locking customers into an all-Red Hat software stack.
“Disrupting legacy business models is a fun mission,” Stevens said to me with obvious enthusiasm.
A History Lesson
All of this reverberates from the highest levels of the company. During his opening keynote, Red Hat CEO Jim Whitehurst gave attendees a bit of a history lesson about the auto-lathe, and how the introduction of commodity components really fueled the innovations of the industrial revolution. Prior to things like the auto-lathe, nuts and bolts needed to be mostly manually machined, and the bolts created by one machinist probably wouldn’t work with nuts from another machinist. Standardization on commodity nuts and bolts allowed for the automobile and the jet engine to be developed. “If nuts and bolts had been patented, the jet engine would not exist today,” Whitehurst said.
Specifically, according to Whitehurst, the Industrial Revolution allowed individuals to stop thinking about the constituent parts and instead think about what you could make with those parts. So too is the same thing happening in today’s Information Revolution. No longer is their as much value in the ownership of physical assets. The real value is in the information assets.
To make the analogy explicit, tight control of the Gluster technology is insufficient for companies like Red Hat to maintain competitive advantage. What’s important is the integration of these technologies in ways that demonstrably benefit their customers.
This, too, is becoming an increasingly interesting proposition. As Whitehurst observed during his keynote, the historical model of IT vendors developing solutions for customers is shifting toward customers solving their own problems using commodity parts. As an example, he pointed to the explosion in Big Data. There was no Big Data vendor, originally; instead, customers set out to address their own Big Data problems, whether on their own or through participation in community projects. Big Data solutions aren’t a competitive advantage in most cases.
As ZettaSet founder and CTO Brian Christian told me, companies don’t have Big Data problems; they have information problems. Big Data solutions are one avenue for these customers to try to address their information problems.
“How do you respond to customer needs when you don’t fully control the product roadmap?” Whitehurst asked, only somewhat rhetorically, during his keynote. Red Hat continues to try to provide meaningful solutions for their customers through collaboration with community initiatives.