Can Marten Mickos make 'Linux for the cloud' work for HP?

The 'not-another-Unix play' play


Hewlett-Packard didn’t just buy cloudy startup Eucalyptus Systems to build its fledgling OpenStack cloud biz, it also bought Marten Mickos, the firm’s Finnish CEO.

HP isn’t the first to pay for Mickos' expertise - that was Sun Microsystems, when it acquired his venture previous venture, MySQL AB, for $1bn in 2008.

Just who is this Mickos bloke and why do big systems companies like him and what he has to offer?

HP announced two weeks ago it is buying Mickos’ Eucalyptus cloud spinner to accelerate the addition of hybrid clouds in the enterprise.

Hybrid is the phrase du jour of tech firms to describe customers that use both hosted systems and on-prem facilities to run various bits of their IT.

HP isn’t the first big tech firm to strap a cloud firm to its mast. Cisco has also been busy, last week buying cloud management startup Metacloud under John Chambers’ two-year, $1bn cloud plan.

But HP’s deal is a real eyebrow-raiser. Instead of following Cisco down the rather obvious path of buying an OpenStack startup, HP – which along with Cisco is a member of the OpenStack Project – bought a company that spins up Amazon-compatible clouds instead.

Eucalyptus lets you build clouds using APIs compatible with those of Amazon Web Services – both for EC2 on compute and S3 on storage. OpenStack was spun up in 2010 to provide a set of open-source APIs for those who didn’t wish to use AWS.

HP has also given Mickos a seat at the top table: he's been made general manager of HP’s Cloud organisation, tasked with building HP’s OpenStack-based Helion cloud. Helion is HP’s supported spin of OpenStack code. Mickos is reporting straight to the queen herself, HP CEO Meg Whitman.

According to HP, Mickos will help build out the HP Helion portfolio.

Mickos replaces HP old-school corporate man Martin Fink as cloud chief. Fink remains HP’s chief technology officer and director of HP Labs. The CTO was not a man to grow a cloud business, but Mickos might be.

Tellingly, the HP announcement pointed out how Mickos had taken MySQL AB, his company before Eucalyptus, from garage startup to “a company providing the second most widely used open-source software in the world.”

Anybody can make and peddle open-source code and achieve downloads, but it’s another thing to make money – which is what Mickos’ MySQL did.

Mickos was named CEO of MySQL AB CEO in 2001, just six years after it was founded 1995 by Monty Widenius and Co. Under his leadership, the company rode a growing wave of open-source database popularity to hit annual revenue in the range of $50m. Such was its rapid rate of growth, some thought it would surpass world number-one Linux business Red Hat for sales. Under Mickos, MySQL AB developed paid subscriptions and support and led the company through three rounds of funding from Intel, Red Hat, SAP and others.

Mickos proved MySQL could make money – to the point where he convinced Sun, desperate to add a software string to its bow, to pay $1bn for his company.

Along with Java, MySQL was the only growing part of Sun’s business.

From open-source databases, and following a brief stint as an entrepreneur in residence at Benchmark Capital and Index Ventures, Mickos joined Eucalyptus as CEO in 2010 – the year after it was founded. Eucalyptus received venture funding from Benchmark.

It’s tempting to see HP's acquisition of Eucalyptus and hiring of Mickos as a modern-day version of Sun's MySQL AB venture. But history doesn’t repeat itself. Certainly there are echoes, but Mickos’ pony was Amazon-compatible APIs while HP’s horse is OpenStack. It doesn’t look like HP hired Mickos to graft his Amazon biz to the HP biz.

But in August, the former MySQL CEO began to talk like he was embracing OpenStack. Clearly, something was in the wind and the Finn was preparing for his coming out in September.

Now HP has a real business person heading up its cloud business rather than technologist or a corporate vet. And, certainly, the opportunity is there for the taking – just as it was on open-source databases last decade.

But remember that bit about history not repeating itself?

The growth and the money in cloud is going the way of proprietary options. On databases, open source was helped by its low price versus the expense of proprietary combined with the growth in web-based development.

Four years after OpenStack was born as an alternative to proprietary Amazon, it's becoming the last word when it comes to various proprietary software - for example, Microsoft's Windows Azure - that provide alternatives to Jeff Bezos juggernaut. AWS growth is slowing while Windows Azure is growing in triple figures – albeit because it’s catching up.

Those trying to build a platform business with OpenStack are struggling.

RackSpace was a prime OpenStack mover in 2010 but has already tried - and failed - to get out. Its new hope is a new chief executive and activist investor. Citrix, another top dog early on, bailed a while back, spinning up its own CloudStack - which has proved just as frustrating.

The problem isn't just the economics - the fact the big companies can price their kit lower thanks to their bigger businesses - it's the technology.

OpenStack has plenty of contributors, but they are pulling in different directions according to their own priorities and preferences.

The upshot is OpenStack overall remains difficult to implement. You still require a team of rocket scientists on staff – like at eBay – or must stump up cash to hire from a growing army of OpenStack consultants.

Further, OpenStack is in danger of fragmenting as tech firms begin to build or bundle their own OpenStack distros – take HP and Oracle's attempts, for one.

OpenStack is starting to look dangerously like Unix or CORBA - as Mickos noted in August.

“It’s difficult to produce technically brilliant products when governance is shared among very large corporations, each one with their own agenda,” Mickos said. "In an all-embracing collective it is difficult to say no to new ideas, but 'no' is a vital component of designs that win.

"OpenStack must also pay close attention to fragmentation tendencies. Certain degrees and forms of fragmentation are not harmful (they may even be useful), but others can slow down adoption and confuse users,” he added.

Mickos will have been brought in not just to "do a MySQL" and grow HP's Helion open-stack cloud business. He'll also be expected to help keep HP honest, by building Helion while also making sure it doesn't slip outside the OpenStack mainstream – all the while keeping the community onside.

Failure in this regard would be death for HP's Helion, rather as if HP were left building and maintaining its own Unix. Imagine that. ®


Other stories you might like

Biting the hand that feeds IT © 1998–2022