In a rare case of lifting the corporate kimono, HP Labs Tuesday offered some insights into its vision for the future of computing at an event held in Palo Alto, Calif., to celebrate its 35th birthday.
HP chief executive officer Carly Fiorina took time out from defending the planned merger with Compaq Computer Corp. to attend the event. She said the Labs represent “the distilled essence of what we mean when we say invent.”
“I tell the HP Labs people that when I need a fix, some inspiration, I come here,” she said.
The Labs’ mission is to help HP stay ahead of the curve by researching technologies that will carry it years into the future, in areas like computer chip design, data storage, digital photography, printing, and streaming rich content over wireless networks. The group toiled in secrecy for many of its years but recently began offering visitors a peek inside.
While its mission hasn’t changed over the years, the Labs hasn’t escaped the economic downturn. The tough times have created a “sense of urgency” to deliver technologies that can be brought to market more quickly, admitted Dick Lampman, senior vice-president and director of research at HP Labs.
Fiorina alluded to the same theme in her opening remarks. “In some cases HP got a bit blinded by its own inventive riches and didn’t focus on a diamond and pick it up and polish it,” she said.
Researchers here offered an overview of some of the diamonds they’re currently mining.
They included a software application that aims to make better use of the knowledge available inside organizations. It could help identify employees with knowledge in a particular area, and then make that person available – anonymously – to answer questions from other workers, said Bernardo Huberman, an HP fellow and director of its Information Dynamics Research group.
The application, termed “social harvesting of community knowledge,” or SHOCK, uses client software that tracks and records the Web pages, files and e-mails accessed by each user in an organization. When a person asks a question, SHOCK looks at each PC to see which users have accessed related information and then sends them the question in an e-mail.
The key to the system is that it’s anonymous, Huberman said. No one can view the data collected at each client, and no one knows who sends or receives the questions. Researchers here wouldn’t disclose how they achieve the anonymity, saying they’re waiting for a patent on the technology.
The application is ready to be tested across HP Labs, but like everything on show here, researchers wouldn’t say when it might find a place in the market. Most users would likely panic at the idea of software that tracks the files they look at, and one researcher admitted that security on the average PC needs to improve before anyone would use it.
When it comes to understanding how enterprises will store, manage and organize data in the years ahead, researchers here said it all starts with improving the data centre. Assuming that high-bandwidth wireless and wireline networks will become ubiquitous, Rich Friedrich, principle architect at HP’s Internet Systems and Storage Lab, said a race is on to create data centres that can scale to meet demand.
On Nov. 5, HP rolled out its “Utility Data Center” offering, which offers a kind of building block system for creating data centres that can scale to 50,000 nodes or more. The company said that using its OpenView and Netaction platforms, the Utility Data Center allows enterprises to integrate servers, storage and networking gear from almost any vendor.
The data centre ties into HP’s vision of creating an infrastructure that enables a user’s data to follow them as they move around the globe. That long-term vision, explained John Sentag, department manager for the Internet Systems and Storage Lab, will be achieved when storage management is greatly simplified at the user end. Data should be stored in what he called “federated arrays of bricks,” meaning commodity servers that can be easily managed to maintain reliability.
While HP continues to throw its weight behind the physical infrastructure, Friedrich said the key to enabling true data mobility is removing security and economic barriers.
HP Labs is trying to unlock “revolutionary capabilities that will fundamentally change the way security is being done,” he said. The idea is to create a situation where information is constantly moving, removing the possibility that a single disaster could wipe out a company’s data. This in turn builds enterprise confidence in security solutions, he said.
On the economic front, Friedrich said greater emphasis must be put on improving the reliability and scalability of commodity hardware used to build data centres. Creating an abundance of cheaper, more reliable systems would foster greater data mobility, he said.
Computer chips made from molecules are also in the pipeline. Today’s semiconductors use millions of tiny silicon transistors that switch on and off to represent the ones and zeros of binary computer code. As the number of transistors increases, problems of heat and power consumption threaten to stump progress by the end of the decade.
HP Labs has crafted a “molecular switch” from a few dozen atoms that can be charged in a similar way to represent a one or a zero, said Stan Williams, an HP fellow and director of the Quantum Science Research lab. Hybrid chips blending traditional silicon components with molecular electronics could allow chips to continue their performance gains for the next 50 years, he said.
HP Labs isn’t alone in pursuing molecular electronics, but claims to have built the smallest molecular switch using a production method that could be replicated on a large scale. Prototype switches developed in the lab measure about 50 nanometers across, or about 100 times thinner than a human hair, but the lab expects to shrink them much further.
The switches can be knitted together using “nanowires,” or fine threads crafted from platinum, silicon, carbon or other materials. Using nanowires that are as wide as only six atoms, HP has used the switches to create simple memory devices, Williams said. “The next step is to literally link them into an integrated circuit,” he said. He predicted that the first molecular switches would find their way into commercial use within 10 years, perhaps five.