HP talks clouds, diamonds and tiaras
Drawing up magical kingdoms in old Vienna
Comment For HP the "big switch" to cloud computing is a big opportunity but is not really a switch at all.
Here at HP's Software Universe event in Vienna the cloud is mentioned a lot but you can't for a moment think that HP's head is in the clouds. It's not. The company does not believe in Nick Carr's Big Switch premise - that there will be a huge switch-over from customer-operated data centres and that, ultimately, computing services will be delivered across a network as a utility.
"There'll be no big switch," said David Gee, an HP marketing VP for software and solutions. "It's not all or nothing. There'll be convergence." But no wipeout of in-house computing. HP is anticipating a mix of in-premise computing, outsourcing, private and public clouds.
Not everything can become a cloud-delivered service in Gee's view. Some software just doesn't offer itself as a service. A case in point is network management software for equipment you own and operate. We might assume that any software used to manage an in-house resource isn't a good candidate for delivery as a service. But if the in-house resource is itself delivered as a service then you don't need the software to manage it.
Much of HP's software is concerned with managing infrastructure resources and optimising their use. If these resources are delivered as a service then customers would no longer need the HP software. A partial cloud switchover will radically reduce HP's infrastructure software golden goose. Doesn't that mean that, to compensate, HP needs to build out its own cloud data centre infrastructure in order not to be left behind by Google, Amazon, and Microsoft, who are all active in cloud data centre build-outs?
Diamonds in the sand
There is no sign of HP doing this - of it building a world-wide set of interlinked data centres. Tasked with this, Gee said:" I'll say three letters: E, D, S." That big HP services acquisition operates lots of data centres under outsourcing contracts and "it's a good place to learn about cloud delivery".
He says HP has many cloud initiatives under the hood - more than we think. Two thirds of HP's software portfolio is already deliverable as a service. The cloud may look like one thing but we should not be confused. Behind the front, the delivery side of the cloud, there will be a complex infrastructure with lots of layers of hardware and software and many vertical niches.
Gee says HP can go through the cloud to the far side and there will be this whole new developing complex infrastructure for it to play in, for it to manage and optimise. The further back in the cloud you look the more dots there are to connect and manage. HP he says, is being very deliberate about this and it has lots of separate initiatives: "We've got a lot of diamonds in the sand that we'll show up with in a tiara some time soon."
The pitch is that HP software, used by customers for their own infrastructure, will smoothly transition to being used by cloud service providers. One example cited is HP's Quality Center v10 software, based on the technology acquired with Mercury, which is used to manage and test software applications and has now been developed to support web 2.0 and service-oriented application delivery, suitable for cloud applications.
A second example is software based on the Opsware acquisition, now called the Universal Configuration Management Database (UCMDB). This, according to Steen Lomholt-Thomden, a marketing veep at HP "is the secret sauce that links everything in IT to how it impacts the business - one view of the truth". It provides automated discovery of infrastructure elements and applications, change management, system health monitoring, automated release management and incident and problem management. It's an IT operations management suite and HP is pushing its use as an efficiency raiser and cost-saver for IT, and as another infrastructure tool for cloud service providers.
No glamour, no sexing up
This software is not glamorous or sexy, but every IT department developing applications and running a data centre is a potential customer for it. As they transition to providing cloud services themselves or as new cloud service providers spring up, the requirements on the cloud infrastructure to be reliable, robust and deliver a predictable quality of service will make this infrastructure software even more necessary. That's the HP view anyway, the HP way to cloud services.
We should envisage this software being used by cloud service providers as a horizontal layer above their physical data centre servers, storage and network boxes. If HP does begin to use its own EDS data centre infrastructure to offer cloud services then there's this software ready to be used to manage and optimise the processes, there's the EDS data centre operating expertise, and there's the whole panoply of HP servers, blade servers, storage kit and ProCurve networking gear that can be used.
It's an organised and pretty - and pretty solid - Lego building block approach to cloud computing services. There's going to be no big bang approach, no all-or-nothing big switch, and no dramatic Microsoft Azure-like catch-up race, cloud data centre hardware and software build-out. Instead HP sees a controlled and executed evolution or transition. No Google-like rush to hyperscale data centres, no Amazon-scale cloud unreliability, instead ordinary and undramatic enterprise data centre predictability and reliability on the far side, behind the cloud's presentation layer.
This is HP's feet-on-the-ground approach to cloud computing and it might, as the fabled tortoise did, pad patiently past the hares sprinting into the distance at the moment. HP will be hoping that customers like this story, are reassured by it, and go with the Hurd instinct. ®