Marten Mickos's Challenge


I’m in the lobby of the San Francisco office of Benchmark Capital, the venture capital firm, and the lobby guard is speaking to me in apologetic tones. He says he’s sorry he had to so carefully inspect me before allowing me on the elevator. He had checked my name on the visitor list, peered at my ID, given me the once over.

The problem, he explains, is that people walk in and want to give their pitch for a startup. Benchmark – the firm’s $6.7 million investment in eBay in 1997 earned it $5 billion by 1999 – inspires plenty of dreamers with business plans. 

My time with the guard is brief yet Benchmark is central to this story. I’m here to interview Marten Mickos, CEO of Eucalyptus, an open source cloud vendor funded by Benchmark and other investors. As Mickos explains, “We’ve raised a ton of capital, so we have time to sit here and improve the product and wait for the market to be ready for what we have for them.”

The time that this funding buys Eucalyptus is precious, indeed essential. Because Mickos’s challenges are numerous and, by anyone’s estimation, substantial.

First, a challenge that may be purely speculative. Eucalyptus is a private cloud vendor whose top selling point is its compatibility with Amazon Web Services. That’s positive – AWS is far and away the big dog. But Amazon, a public cloud vendor, makes sneering noises about the private cloud, its partnership with Eucalyptus notwithstanding. In the Amazon worldview, there’s only one cloud, the public cloud.

That is, until Amazon itself scored a $600 million contract to build a private cloud for the CIA. Does Amazon’s foray into private cloud mean that, long term, it will expand into private cloud – creating a roadblock for private cloud vendors like Eucalyptus? Amazon has given no sign that it will, so it’s pure guesswork.

There’s a challenge that’s far more real: businesses have been slow to embrace the hybrid cloud. Eucalyptus offers a private cloud that pairs with a larger public cloud. The ideal is to merge on-premise and off-premise into a seamless environment that can be managed as a single entity. That’s a forward-looking stance and will eventually see major adoption. But as of late 2013, hybrid cloud was still “rare,” noted Gartner analyst Tom Bittman.

Here’s the toughest challenge of all: in the race for private cloud market presence, Eucalyptus is running behind its open source compatriot, OpenStack, which boasts a roster of gold-plated corporate sponsors. In fact Eucalyptus is running behind an entire bevy of private cloud vendors, from leaders like VMware and IBM to contenders like Citrix and CA Technologies. In reality these vendors can’t be ranked in a straight line. Amid the confusion of an emerging market, “private cloud” refers to various services sold to different sectors. But no matter how you slice it, Eucalyptus is one of the smaller players.  

Mickos, who had a long and successful run as CEO of open source database vendor MySQL before Sun acquired it for $1 billion, has a strategy to surmount each of these hurdles. “We are in this to win,” he says. “We are not in this to be an also-ran.” He plans to push up Eucalyptus to be “the best vendor in this space” – a climb he concedes will be a steep one.

The Open Source Cloud Landscape

The relationship between cloud computing and open source has its pluses and minuses, Mickos notes.

“Five years ago, we saw [cloud] as…a potential risk for open source,” he says. The worry was that outfits building cloud services wouldn’t share code because they’re not actually shipping software.

In some cases those worries have come true. “You go and look at fantastic companies like Box, Dropbox, Zendesk, Salesforce, Workday, NetSuite…Are they sharing their code with the world? No. So, you could say, all these brave new companies with the best software engineers, how can they stay so closed? And that’s sort of a negative.”

On the upside, the sheer magnitude of new cloud services companies is fueling open source. “It turns out they need so much software that there is a huge opportunity for database vendors, cloud platform vendors, networking software vendors, and storage software vendors to provide and sell their open source software. So, I do think it’s strengthening open source.”

As for private cloud adoption, Mickos is frank about the fact that it’s faced resistance among businesses.

“We thought that private cloud would be a massive market, and so far we’ve been wrong. And I believe it’s a traditional S-curve. Initially, you have this exuberance and excitement, everybody thinks it’s a big, big business coming, and then actually the curve goes down before it starts growing.”

This sense of a market that has yet to fully emerge runs throughout the industry, he says. “Everybody believed in it. Everybody’s throwing billions into it. Where’s the business? I mean, there’s no business yet that would warrant a billion dollar investment in private cloud. Yet three companies [IBM, Cisco, HP] have already committed a billion.”

Part of the reason it’s taken time for the market to arrive has been the vendors themselves, he says. “I think that many of us early vendors, we had products that didn’t…weren’t good enough. Ours wasn’t good enough. OpenStack isn’t good enough. VMware’s isn’t good enough. BMC doesn’t really have a cloud platform. IBM’s SmartCloud isn’t good enough.”

The Road to Ease of Use

To address this early product immaturity, two years ago Eucalyptus took a hard look in the mirror.

“In 2012 we did a bunch of proof of concepts, and most of them failed,” Mickos says. “Test after test came back and the customers said, ‘No. It’s not working for us.’”

It was time for a change. “We said, ‘Okay. Let’s stop this. Let’s go back, do our homework, focus on the product, make sure that the product is really solid. And then we go back to customers.’”

The technical changes were numerous. Eucalyptus consists of a series of independent Web services that communicate with each other; each was deeply reviewed. “So, an example, we just rewrote our networking. We had functional networking, so it didn’t stop customers from deploying, but it was complex and you had to make really hard configuration choices up front and customers weren’t sure which ones to choose and small changes would have big effects. Now we’ve fixed that so we have re-architected how networking is configured.”

At the time, Eucalyptus had optimistically staffed up to 95 employees. Along with the technical changes came an organizational change – life as a startup – and head count was slashed to 70, with some churn. Over the last year the company has been hiring.

These days, Eucalyptus touts ease of use as a key selling point, alongside its AWS compatibility. The most recent release, Eucalyptus 4.0, includes enhanced features for operations managers, those who maintain the cloud. Networking, as Mickos tells it, is now easier to install and scale. Eucalyptus has its own S3 implementation, called Walrus, which now has links out to third-party solutions like Swift, RiakCS and Ceph. The cloud controller, CLC, has been beefed up to enable faster scalability.

Eucalyptus’s customer base leans toward small and medium-sized tech companies that want to cut costs with open source but don’t have the staff for a do-it-yourself toolkit like OpenStack. “We go to the tech companies that say, ‘Marten, I don’t want more than three guys to be working on this. I really want it to function so that our engineers can do other stuff.’ So that’s a distinction between OpenStack and us.”

The client list includes the gaming outfits Riot Games and Ultimate Gaming, tech firms AppDynamics, Codenvy and MemSQL, government agencies NASA and NIST, financial regulator FINRA, Cornell University, OneHealth and retailer Puma.

To grow its market share as an open source cloud vendor, Eucalyptus will need to overcome the shadow of OpenStack. Or, maybe not.

“Over time everybody will be an OpenStack product,” Mickos says, though he stresses this isn’t a roadmap statement for Eucalyptus itself. His prediction: “OpenStack will be a set of projects, it will be a set of hardware certification, and many people will build different projects out of them. Already today you cannot find two OpenStack products that are mutually compatible.” He points to OpenStack projects by Piston, Nebula, Cloudscaling, HP, Rackspace, Red Hat, and IBM that are not compatible with one another.

“So, in the long run I think we will all use OpenStack components, and therefore you could argue we will all be OpenStack products.”

What then is the differentiation? “I would say CloudStack is selling to service providers. Eucalyptus is Amazon compatible. And then it’s up to the OpenStack vendors to differentiate. I don’t know how to differentiate it, that’s their problem.”

The centrality of OpenStack prompted a recent firestorm in the tech industry when Red Hat announced that it will not support customers that deploy to any version of OpenStack other than its own. Some open source advocates cried foul, claiming the move wasn’t in the spirit of sharing that’s so key to open source.

“I’m not saying I agree with them and I’m not saying I would do it that way,” Mickos says. “But if you just ask ‘will it work’? Yes, it will work. And it will be painful for other OpenStack vendors who have been pushing and investing and contributing and building, and then Red Hat comes in and says, ‘Thank you, it’s now ours.’”

Private Cloud, Hybrid Cloud

Eucalyptus’s competitive strategy is based on its Amazon compatibility. That’s a convincing selling point, but what if AWS, inspired by building a private cloud for the CIA, moves into the private cloud business itself?

It won’t do that, Mickos says. “I believe that what they are doing for the CIA is completely custom, hand-built, manual labor. There’s no scale. They can’t replicate it.” In fact, AWS’s efforts with the CIA will move it in the opposite direction. It will finally convince Amazon of the importance of the private cloud.

AWS will also come to understand the difficulty of building a private cloud platform. “So, when they see that, they’ll say, ‘Oh! Now we understand why Eucalyptus is doing what they do. Now we understand why Eucalyptus is a product. Now we understand why they are shipping software. Now we understand why installation and configuration must be automated.’ And that insight will drive them to pull us into business deals. That’s my prediction.”

Apart from other competitive concerns, Eucalyptus’s long-term fortunes are tied to the growth of the hybrid cloud. Businesses must embrace this model wholeheartedly for the company to truly flourish.

There’s certainly debate on this topic. “People say, ‘Oh, it’s one or the other [public or private]. One is threatening the other.’”

He disagrees. “They don’t know it’s a symbiosis. A public cloud, ultimately, cannot win if it doesn’t have on-premise satellites. And on-premise environments cannot be really powerful unless they are connected to the public cloud. It’s not two competing worlds. It’s two dimensions of the same world.”

Whether businesses will eventually see these two dimensions as a single unified cloud platform is Eucalyptus’s big bet. Only the future can tell.

Here's Mickos on the future of cloud computing and Eucalyptus: 

This is a single player video template with no brightcove branding and autoplay.

James Maguire is Datamation's managing editor. Follow him on Twitter: @JamesMaguire

Image courtesy of Shutterstock.