HP gets ready to make a cloud ... by boiling the ocean
If only it was Sun, we'd have a headline hat-trick
The goals of Hewlett-Packard's OpenStack cloud are immodest, perhaps even heretical in a Valley where people casually caution you against "boiling the ocean" – or trying to do everything at once.
According to the HP cloud beta announcement page: "HP intends to extend its full spectrum of cloud offerings spanning private, hybrid, and public architectures."
Don't be alarmed, the cloud novice re-assures us: HP's cloud has the "same commitment to leadership" as its existing packaged consulting, software and hardware offerings.
HP is the world's largest maker of personal computers and it only joined the OpenStack community as a cloud recruit in August, but it is already ready to rumble, it told The Reg.
Not only that, but the year before HP had hitched its wagon to an entirely different star: Microsoft. The world's largest maker of software is also turning cloudy with Windows Azure, and in July 2010 HP was teamed with Dell and Fujitsu to test services and deliver Azure in a box for people to buy.
We're still waiting for an HP Windows Azure appliance – also the appliances from Dell and Fujitsu. Meantime, HP has aligned itself with something billed as a the Linux for the cloud, a community effort whose code is available under an open-source license.
"They [HP] are very big company with lots of cloud VPs. Biri at the moment seems to have enough executive buy-in to make this work"
What's behind HP's confidence and reassurances?
Well before it joined OpenStack, HP was ramping up operations in a way that makes membership a mere formality. HP spent its time staffing up and shifting people around in preparation for the cloud in a push that really ramped up this year.
In early 2011, HP snagged IBM's vice president of cloud computing, Zorawar "Biri" Singh, who become senior vice president and general manager of HP cloud services.
One OpenStack community member we spoke to, who wished to remain anonymous, reckoned the wind is at Singh's back. "They [HP] are very big company with lots of cloud VPs. Biri at moment seems to have enough executive buy-in to make this work," we were told.
Singh had been IBM vice president of cloud computing and joined as CEO of single-sign-on startup Encentuate when it was bought by IBM in 2008.
HP has raided Rackspace, which helped launch OpenStack in 2010, and which become a driving force behind the project's governance. HP has hired Rackspace director of software development John Purrier in a role that is not yet clear. Purrier's LinkedIn bio still lists him as a director of OpenStack at Rackspace but he now has an official HP email address. We know because we emailed him. He didn't respond.
HP has also reorganised for cloud. Patrick Scaglia, the vice president and chief technology officer of HP's imaging and printing unit for a period of five years become VP and CTO for HP's cloud services and applications in June.
Leading them all is Shane Robinson, executive vice president and chief strategy and technology officer. A former Compaq man, Robertson was a principal architect of the Compaq-HP merger and went on to lead HP's purchases of Mercury, Opsware, EDS and 3Com along with 30 other companies that helped take HP deeper into software, services and networks.
We've heard of online services
HP has been racing to soak up the kind of brains and talent needed to build compute and storage fabric capable of boiling the ocean.
It has had to: HP is a maker of PCs and servers, meaning that that building and running scalable online services is simply not in its DNA. And while there's a great deal of hype about OpenStack, in its raw, community form OpenStack provides little more than basic building blocks for a potential service. These blocks are also at differing levels of maturity, with the corporate members of the OpenStack community developing the ones that suit their own goals. There's the basic Nova and Swift compute and storage elements, but then you have Quantum for network connectivity with 40 developers from Cisco and Glance for virtual machine images mostly driven by Rackspace.
The really tricky part is not just getting somebody who can build your service based on these blocks, but delivering them on a service-provider-scale architecture capable of scaling to tens of thousands of distributed nodes that deliver reliable, real-time service.