Seven Questions for Prith Banerjee, Hewlett-Packard's Head of Research
It’s been about two months since Hewlett-Packard’s new CEO Léo Apotheker put the company on a new cloud-centric path as part of a big speech laying out a new strategy. But there haven’t been a lot of specific announcements concerning how HP is going to get there.
It turns out there’s a lot of work going on at HP that’s taking place with the cloud–and the new strategy in mind. I recently caught up with Prith Banerjee, HP’s Senior Vice President for Research and director of HP Labs. We talked about how the research going on in the labs fits with the new strategy, or as Banerjee would put it, how the work in the labs is actually driving the corporate strategy. He also talks about some of the moves HP is planning to make in the cloud in the coming months.
NewEnterprise: Prith, When Léo made his presentation, everyone focused on his comments around the cloud, but the expectation coming in had been that his comments would focus primarily on software. He really touched on both, but the cloud was the bit that everyone remembers. At a high level, what do his comments mean for you at HP Labs?
Banerjee: HP’s strategy is around seamless, secure, context-aware experiences. It’s around cloud and connectivity. And again software is the glue that brings them all together. What we believe we are going to bring to our customers is a seamless view of the world and a seamless way of accessing information. The driving problem that we are trying to solve is around the explosion of information both in the consumer world and in the enterprise world. And what HP wants to bring to the table is a way to help customers seamlessly migrate from the consumer world to the enterprise world and back and forth, and in a secure manner. First of all, connectivity over the cloud, while being secure all the time.
And what’s new at HP Labs since Léo has been on the job?
We have eight teams at HP labs, and in their own way each is working on different aspects of things that Léo talked about. Print and content delivery, that’s around the document ecosystem, if you look at mobile and immersive experiences, it’s directly tied to the future of connectivity and the connected world. Our work on cloud security is clearly tied to Leo’s strategy around cloud. Look at information analytics, that’s the software that’s going to take all the information and drive better insights. Work on intelligent infrastructure and networking will be the foundational layer of the cloud as people try to reach out to the cloud, both as consumers and enterprises we have to have robust, scalable high performance infrastructure. On services Leo talked about solutions for various vertical fields, financial, health care, transportation and so on. So that’s the grand context. So HP Lab’s strategy has been forward-looking. Our role has been to provide guidance three to five to 10 years into the future. And we’re delighted to be in a position of driving the company’s strategy. It’s not like we’re following the company’s strategy. The company’s strategy is just well-aligned with the kinds of things that we were doing at HP Labs to begin with.
Let’s talk about some of the specific things you see coming out of HP Labs three and five and 10 years out.
We are particularly excited about some near-term work. Leo talked about the cloud, and when we talk about the cloud, there’s the public cloud. Shane Robison [HP's executive VP and chief strategy and technology officer] and Léo talked about HP coming into the public cloud arena. There’s some core intellectual property that HP Labs has developed that is core to the strategy. When you come out with a public cloud effort you need to have some foundational things like storage as a service. The key technology there came out of our key-value storage technology at HP Labs. That’s a scalable, reliable, low-cost way of providing tremendous amounts of storage to our customers. That should come out within a few months. That will be the first thing that comes out. Immediately following that there will be a compute-as-a-service play. And that is based on the Cirious platform work that is going on in our cloud and security lab. It’s a compute-as-a-service play that we call internally cells-as-a-service. These are virtual compute cells and memory cells and networking, sort of virtualizations, which allow people to create these multi-tenant applications that are incredibly secure so that there’s no way for one customer to impact the application of another customer on the cloud. You should see that in the hands of customers in the next six to eight months. Longer-term we are moving up the stack. We have significant work going on at the higher level around platform-as-a-service. We have IP built in the Intelligent Infrastructure Lab to provide future Web application developers a platform which will be highly scalable, available and secure. On top of that will be services we provide on the cloud. That’s where the services lab comes in. We have hired a new person, Jamie Erbes, to lead that effort. Jamie and her team are trying to develop a set of cloud services aimed at verticals like health care and transportation and financials. We’re coming up with some very unique IP in these solution areas. And our delivery mechanism will be over the cloud. As you create the interesting applications you need some deep analytics to see how you can improve services to customers.
These sound like things that we’ll see in fairly short order. What are some of the longer-term things you’re working on?
Our networking and intelligent infrastructure labs are working on some very novel ways of building out future networks for enterprises that will be for a cloud setting. And there’s the work on the data centers and servers of the future. The future server architectures will be the foundation of the cloud in the future. Those will come out in a little longer time frame. The server architecture work will come out maybe two to five years in the future.
So we still haven’t talked much about software yet. Where does software fit in all this?
We’ve created a new information analytics lab. It’s led by Laura Hill, who came from Sun Labs. What they’re trying to do is look at all the structured and unstructured data. About 80 to 90 percent of it is going to be unstructured. The key is tying it together with the structured data and do really deep analytics on that. The work that is going there is being leveraged with a new company that we acquired, called Vertica. It’s now sitting as an incubator within Shane’s organization [Office of Strategy and Technology]. And so Vertica and the information and analytics lab are working very closely together on some of the future analytics opportunities for both consumers and enterprises.
Having seen some of the troubles that have taken place at Amazon recently, in your view and HP’s view what does the cloud need?
I don’t want to comment specifically on anything taking place at another company. But when we started our work on scalable storage, there were three goals toward that project. One was tremendous scale. We defined that as hundreds or thousands of petabytes, even exabytes of information. That’s a tremendous amount of information that you want to store, not in one data center or even two but in hundreds of data centers. And the reason we wanted to do it that way was to address some of the issues that Amazon has faced. Amazon currently has about six data centers around the world. We are trying to design something for a hundred around the world. The second reason is availability. We want to design a system that is 99.999 percent available. Whatever happens you should always be able to access your data. That is a core tenet to doing these things. And the third is low cost. The world knows how to do two of these things simultaneously but not three. You can do large scale and high availability, but it will cost you a ton of money. The world also knows how to do large scale, low cost, but it fails all the time. The world does not yet know how to do large scale, highly available storage at low cost. And that is a key bit of IP that was created at HP Labs. It’s based on erasure coding, which is a coding technique that’s been around for many years. The innovation is how we applied it in a system-level setting using software, using these low-cost disks that we get from the hardware world. And that innovation allows us to have high availability at low cost. Once HP has this kind of public cloud storage, we think we’ll have a service that is very resilient.
The other fundamental concern in the cloud is around security. What can HP offer on the cloud security front?
One of our labs is devoted just to cloud security. We have two things cooking related to cloud security. One thread is security analytics. When you have a large complex enterprise and IT organization, often CIOs have no clue how secure they are. They think they’re secure. They patch their systems regularly and they hope to God they are secure. But the security threats keep changing, and so there’s a constant battle for resources. Currently at a very frivolous level, the strategy of many CIOs is to keep adding security until you run out of budget. There’s either a cap on dollars, or people or time. Somehow you have to cap the number of security dollars. We’ve done some deep research on determining how secure you really are. We came up with a demonstration this year called G-Cloud or Government Cloud. We will provide a dashboard regarding how secure your cloud infrastructure is as the world around you is trying to create new threats. The other thread that is going is around trusted virtualization that is highly relevant to what Leo talked about. What we are doing is to provide this seamless view from the consumer view to the enterprise view. Well, what does that mean? Today you can talk to large organizations like HP, and our CIO will not allow random devices on the system. It has to be an HP-approved device. And here there are these young people trying to bring these gadgets like the iPhone and the CIOs will not let you do that. However in the consumer world, people love the touch interfaces you get on those devices. But you can’t use the same device in the enterprise world. The technology we’ve come up with is trusted virtualization. We provide a consumer view and an enterprise view. And these two worlds will not clash. We keep them completely secure, and running on the same device. It will run on either a Web OS device or on a Windows PC. It will provide a completely bulletproof consumer and enterprise experience.