HP snuggles up to OpenStack in cloud embrace

HP's Helion needs to drive the code it creates

It's hard to understand OpenStack's role in the data centres of the future without understanding HP's approach. HP's cloud strategy is named "Helion" and it includes public, private and hybrid elements. HP bought and has incorporated Eucalyptus into Helion, but the core of its cloud bets are on OpenStack.

Helion is vital for two reasons to understanding OpenStack as a whole. The first is that HP is the single largest contributor (by several metrics) to the latest OpenStack release, Liberty.

Put simply: HP's Helion needs to drive the code it creates, and HP is putting more effort into OpenStack than anyone else. (Although NEC currently holds the crown for most lines of code.)

The second reason that Helion is important is that it serves as a model for many other companies. This isn't to say that Helion is unique, or that HP's vision is somehow different from others that have gone before. Quite the opposite: Helion is fairly derivative as far as visions go, and that's why it's such a perfect model.

Know your audience

HP serves the big money. HP's customers are enterprises, governments and various organisations smaller in size, but not in means. HP doesn't do "budget" and so the bottom end of the market simply isn't something they have to think about.

For many reasons, this is a good thing. It means that HP has decades of experience understanding the mindset, foibles and quirks of the organisations that actually bring home the bacon. It also means that HP has internalised the very same risk aversion that are the heart and soul of its clients' corporate IT strategies.

HP does take risks. They are simply carefully considered risks, with every conceivable failsafe, backup and detail planned out. There is a whole ecosystem of venture capitalists, startups and what-have-you out there to do the really crazy stuff: HP will assimilate proven technologies and focus on a slow, methodical evolution.

OpenStack needs this. To be successful, OpenStack has to appeal not only to IT teams that work with open compute servers and compile their own operating systems. It needs to appeal to enterprises, small and midsized service providers and – above all – all these bits need to work together.

HP has generations of experience working on the creation and execution of IT standards. The process is long, tedious and at times outright miserable. Getting standards to work right requires interoperability testing, regression testing and layers upon layers of quality assurance.

Believe it or not, this is something that the OpenStack community is getting quite good at, in no small part because of the guidance of veterans like HP, IBM, Red Hat and NEC. At the recent OpenStack conference in Vancouver I talked to representatives from dozens of companies and the most common comment among them was the importance of interoperability, and the reliance all of that has on careful planning and testing.

Many kings, many castles and a world to conquer

HP's Bill Hilf recently told the The New York Times that it makes no sense for HP to go head to head with Amazon, Microsoft and Google in the public cloud space. Many saw this as HP conceding the public cloud to the Big Three. They couldn't be more wrong.

Hilf tried to explain that HP isn't abandoning the public cloud, but he did not do a great job of articulating what Helion actually is, and what HP's strategy is going to be. That's a shame, because HP's strategy is quite simple.

HP wantsto sell you a Helion cloud that you can run on your own premises. That cloud will be able to send workloads to HP's public Helion cloud. All of this will be supported by HP's enterprise services. These folks will not only help companies understand the tools and technologies in play, but what workloads are more cost effective onsite and which are most cost effective farmed out to HP's public Helion cloud.

This strategy is no different than that of virtually every other major OpenStack contributor on the rolls, as well as several of the implementers who are not contributing back much code.

HP knows that the days of winning accounts with just the hardware are coming to an end. Lots of people make good hardware, and software is integrating with Dell's iDRAC, Supermicro's IPMI and Cisco's UCS just as well as it integrates with HP's iLO. The enterprise management software that HP sells sets it apart for now, but it isn't enough to bet the whole company on.

HP is betting the farm on the total customer experience. HP hardware integrated into HP software (both proprietary and the open source stuff in Helion) supported by HP's services. They'll provide everything an organization could want, including public cloud resources to run workloads, and – in theory at least – the experience will be easier than that which any other vendor can provide.

But you don't have to live with HP forever and ever amen. That's what Helion means. If you feel a burning desire to build an Amazon-compatible hybrid setup, HP owns the only viable alternative: Eucalyptus. If you want to be a little more mainstream and use OpenStack, then you can choice to move workloads to any OpenStack provider on the planet.

You can even have your private clouds delivered by multiple vendors, so long as everyone is running OpenStack. Every vendor in the OpenStack ecosystem understands that organisations of all sizes are sick of lock-in and ability to pick up and move from A to B at a moment's notice is baked right in.

Nobody is going to win the "massive, vertically integrated, locked-in, proprietary stack" game against Microsoft. They have way too far a head start. The end result is an ecosystem of companies big and small that are gearing up to actually compete on merit.

Not a hobby

To understand OpenStack you must understand that it is is not a hobby. It is not a game: it is critical to the survival of so many companies. OpenStack. If OpenStack fails then interoperability between all small and mid-sized cloud providers fails. If OpenStack fails then everyone involved has to go back to trying to build proprietary vertically integrated stacks – including public clouds, as those are now a mandatory part of the complete stack.

Microsoft, Amazon and Google have public clouds that are effectively unchallengeable. IBM might get SoftLayer up there, but they are doing so on the back of a massive OpenStack investment. HP's Helion cloud is probably the next best contender, and Hilf – the guy in charge of HP's cloud – said it's not very likely they'll be directly challenging the Big Three.

So that leaves interoperability. Helion may never directly challenge Amazon, but if you put together every public OpenStack cloud on the planet and make it easy to move workloads between them, that's a whole other story, now isn't it?

Suddenly, OpenStack as an ecosystem can offer me something that Amazon, Microsoft and Google never will: the ability to store my data and run my workloads under any context imaginable.

If I have security, legal or data sovereignty concerns for specific workloads or data I can restrict those to only running on my on-premises OpenStack setup. Or maybe I only run them with OpenStack providers that don't have a foreign legal attack surface.

CloudA, for example, can run my "must live in Canada" stuff, while I farm everything else out to HP's public Helion cloud. Maybe I have some really bursty stuff that is going to be loads cheaper on Amazon, but is isolated enough it doesn't need to move back and forth in traditional hybrid fashion. I can do that too, nothing stops me.

Some variation of "and $vendor can help me with every aspect of this" is what all the OpenStack players are betting on.

But it's even bigger than that. If the OpenStack gambit succeeds then Microsoft, Amazon and Google are the outsiders. The world becomes "OpenStack" and "other".

OpenStack is about building a world where a public cloud is a feature, not a product.

That's pretty brutal for Amazon and Google. They're selling a single feature and not a complete product. VMware is trying to build a complete stack, but they may be too little, too late. They can't ramp to compete directly with the Big Three, and they won't drop prices enough to stimulate global uptake of vSphere by service providers.

That leaves us with Microsoft versus everyone else. Again.

With Microsoft you like what you're served or you leave. With OpenStack, if you don't like the flavour served there are an unlimited number of others to choose from.

That is HP's OpenStack strategy. That is the OpenStack strategy. It's a good one. ®