HP gets Cirious about cloud research

Prith Bannerjee, head of HP Labs, talks about how the company will use open innovation in its cloud-computing research

HP has opened a new research lab in Singapore dedicated to cloud computing and cloud datacentre design. ZDNet UK spoke to Prith Banerjee, worldwide director of HP Labs, to find out more about the project.

HP Labs Singapore is the company's first new lab since Banerjee took control of its research in 2007 and set out a five-year strategic plan that made cloud computing one of the eight pillars of its research focus.

However, it is not HP's first cloud research lab, or its first cloud investment in Singapore. In 2008, the company teamed up with Intel, Yahoo and a number of academic sites (including the Infocomm Development Authority of Singapore) to create Open Cirrus, an experimental, open-source cloud platform, which was based on x86 hardware and Yahoo's Hadoop supercomputing project. Open Cirrus already has 10 testbed sites around the world for researchers to develop on.

We asked Banerjee what the new lab will concentrate its research on, how the facility fits in with Open Cirrus and HP's other labs, and how it will feed into the company's cloud services for enterprises.

Q: HP already does cloud research in the Service Automation and Integration Lab in Palo Alto and the Automated Infrastructure Lab in Bristol. What will the researchers in the new lab concentrate on?
A: All three will work to develop an enterprise software cloud platform that we refer to as 'Cirious'. The Cirious vision is to develop a set of cloud capabilities that enable service providers such Verizon, Vodafone and others to take full advantage of the power of the cloud.

Our work will enable service providers of all types to tap into foundational cloud capabilities — into general-purpose horizontal capabilities as well as those tailored for specific industry verticals such as telcos, entertainment, healthcare and others.

We are working on a platform that will make it easy for them to develop, host and manage their services, to deliver value from the cloud, and to integrate with an ecosystem of services from other providers.

Cirious is an enterprise-grade software platform, and the vision of this is to enable service providers [with] access to cloud capabilities within an equal system of other services. There are cloud services available from other companies; however, those are mostly available for more consumer apps, where you really don't require as much scale, as much security, as much flexibility and availability.

Our vision is to provide a cloud platform that will be at an enterprise scale. We believe enterprises will not embrace the cloud until you can prove to them that it is highly scalable — and 'scalable' means scaling up and scaling down. It also means highly flexible, it means highly secure, so we can trust our data on remote sites.

It has to be highly available; we cannot afford to have a cloud services datacentre go down. These are very complex challenges.

How does HP Labs Singapore help you deliver that?
The future [is] billions of users accessing millions of services though thousands of service providers over millions of servers. Where those services are housed, how the data is processed or stored what the network traffic looks like — those will be irrelevant to the user.

But to our enterprise customers and service partners, the advancement in technology architecture driven out of this lab will be paramount.

HP Labs Singapore will research and develop a set of cloud-based apps in partnership with our customers in the Asia Pacific region. They'll be built on top of technologies...

... developed out of cloud-based research at the other two labs, in addition to new technologies they develop themselves.

The second area of research is into datacentre design, to determine what an ideal datacentre for the cloud would look like.

What about projects like Open Cirrus; how does that sit with HP's plans for Cirious?
Open innovation is such an important innovation of our overall research and business growth strategy. We will continue to invent heavily to ensure we maintain our technology advantage and align our portfolio to lead the evolution of the marketplace, but we cannot do it in a vacuum.

The next big shift in computing will be far more global in nature and will require a far more collaborative spirit when it comes to driving innovation.

It's not about how much research we as HP alone put behind innovation anymore; that alone will not drive progress. Open innovation is absolutely critical.

How are you planning to offer cloud security that will satisfy the enterprise?/>
We will allow enterprise customers to have part of their computation within their own private enterprise datacentre. When they need to flex up their computing or storage capabilities, they will just flex up to the cloud — so they will go across an external network and run some of their computations on external facilities.

The way we do this is we create a concept of virtual cells. Instead of actually running your computations on physical processors or physical storage devices or physical networks, we create a virtualisation abstraction and essentially run your computations on processing cells and memory cells and networking cells and storage cells.

We provide that encapsulation in such a secure manner that you don't have to worry about the security. That is the key 'secret sauce'.

Will the new lab add to HP's research on sustainability and reducing costs?
We have a 'big bet' project on the sustainable datacentre. The vision is how you can come up with an end-to-end design with a supply chain and demand-based complete view of the world.

Datacentres today can consume [up to] 200 megawatts of power; that is not good for the environment. We are looking at these datacentres being powered by multiple energy sources: power form the grid, power from alternative energy sources such as solar, such as biofuel, such as wind.

And then we're also trying to adjust the demand, so when different computations come onto your datacentres, they can be placed on specific physical processors where we can dynamically control the air flow [and] the air conditioning, so that only those specific areas will be cooled. It's a very ambitious project whose goal is to reduce the TCO [total cost of ownership] by 75 percent and the carbon emission by 50 percent.

TCO has a materials cost and an operational cost and just acquiring the hardware... We have done some analysis, and the operational cost is actually quite significant, in terms of the human cost — we can relieve that though automation — and the energy consumption.

These are the environmental costs people don't normally look at from a dollar perspective, and we think it important to do that. We are looking at the total cost of ownership — the private cost and the social cost.

And is Singapore a useful place for looking at those costs?
Singapore is home to one of largest concentrations of datacentres outside the US. A number of technology companies have or plan to open a datacentre here. Conservation and optimisation are key points for a cost-effective service model — but Singapore is hot, it's humid and land comes at a premium. Natural resources here are scarce and most resources — for power production, for example — are shipped in from other countries. This provides us with a living laboratory; if we can make a services datacentre cost effective here, then we can do it anywhere.