IBM Weighs Watson for Internal Use
IBM Weighs Watson for Internal Use
At a recent luncheon in New York City to discuss IBM's Smart Computing strategy, IBM Vice President and CIO Jeanette Horan said IBM is doing several things to enable its staff to make better use of unstructured data, and one of them is to use the company's Watson technology. Watson is the computer system IBM designed that won a "Jeopardy!" challenge against some of the game show's biggest winners. IBM calls Watson a workload-optimized system designed for complex analytics, made possible by integrating massively parallel Power7 processors and IBM's DeepQA question answering software. Watson applies advanced natural language processing, information retrieval, knowledge representation and reasoning, and machine learning technologies to answering questions.
"We're looking at a project to do an internal Watson to look at all the information our salespeople need and to take all of that information and build a source of information for our people," Horan said.
Horan noted that IBM has been able to bring together multiple sources of information, such as customer information, market information, pricing and more. "We need to understand how to bring al this information together and to get rid of all the rows and columns and be able to use unstructured data," she said. "We want to have our salespeople ask an unstructured question and get an answer. ... You have to continue to evolve your body of knowledge."
IBM officials have said the first business application of the company's Watson technology will come in the health care field, where Watson could serve as a physician's assistant or as part of evidence-based or collaborative medicine solutions. IBM also identified other potential business applications for Watson, including technical support for help desks, call centers and the like; enterprise knowledge management and business intelligence solutions; and government solutions for improved information sharing and security.
Guru Rao, an IBM Fellow and chief systems engineer in the IBM Systems and Technology group, said, "Watson is a way of not providing just search and retrieval, but a way to create structure."
It is critical, Horan said, that IBM show its customers that it is able to use its own technology and processes to transform itself as evidence that it can do as much and more for them. Prior to becoming IBM's CIO, Horan was vice president of Enterprise Business Transformation at Big Blue. Yet, she still drives IBM's transformation agenda and helps oversee a technology strategy aligned with the business that meets both the growth and productivity commitments of IBM's 2015 road map.
"Within IBM we have a broad application portfolio, and our challenge is to do more with less as our partners are looking for agility and at how IBM is leveraging technology," Horan said. As such, IBM is employing a lot of virtualization technology, but that is not enough, Horan said. "So we are deploying more cloud technology, like our business analytics cloud."
After IBM acquired Cognos, Horan said many internal IBM groups were creating their own Cognos systems, "and we said there's got to be a better way." With so many groups doing their own thing, "we realized we had multiple versions of the truth. But we wanted to get to a single version of the truth."
So IBM built a Cognos-based analytics cloud known as Blue Insight. Announced in 2009, IBM's Blue Insight is the world's largest private cloud computing environment for business analytics, Horan said. Blue Insight can access business intelligence from hundreds of databases-more than a petabyte of information-and make it available to users on their desktops. A petabyte of information is equivalent to 300 billion ATM transactions.
At the end of 2010, Blue Insight had 165,000 users and more than 100 data warehouses ported to it, IBM said. By the end of 2011, Blue Insight will be accessible through IBM's intranet to more than half of the company's 427,000 employees. To date, some of the achievements of Blue Insights include on the spot skills gap analysis by HR professionals, cash position forecasts by IBM treasury analysts, and marketing trends for IBM's marketing teams. And these analyses that used to take weeks or months to perform now only take hours or minutes, IBM said.
Moreover, Horan said IBM also has made use of the company's development and test cloud to provision resources to IBM developers that build and maintain the company's 5,000 internal applications. Because testing needs can vary and fluctuate so broadly, Horan said IBM needed to be able to offer greater flexibility to development teams. With the cloud, IBM can scale up to support large teams or scale back as needed.
"We have a cloud set of systems available for development and test with self-service provisioning," Horan said.
Cloud Computing: Core IBM Growth Area
Meanwhile, focusing on where cloud computing appears to have made the most impact early on, Harvey Koeppel, executive director of the Center for CIO Leadership, said smaller and midsize companies appear to be buying into the cloud more readily than larger companies. "They've made the mental leap, and they understand the potential value of cloud computing," he said. "Look at it in a similar way to how the Internet broke into the enterprise. The cloud is tracking the same way as the Internet did. It started behind the firewall. Public versus private is the wrong conversation to have. It's a surrogate for public Internet versus private intranet. The best cases are hybrids."
Cloud computing has become a key part of IBM's overall strategy and one of the core growth areas in its 2015 road map. Thus the company has deployed considerable resources to cover clients' cloud computing needs.
Providing an example of an IBM customer that initially had reservations about moving to the cloud, Professor Giuseppe Visaggio of the Department of Informatics at Italy's University of Bari, said all reservations went by the wayside when IBM backed the cloud environment with an IBM System z mainframe running Linux.
"The mainframe has a way to work that is more efficient," Visaggio said. He added that IBM's services support was critical to the project's success.
The University of Bari is located on the southern Italian region of Puglia, an area with an economy primarily based on small and medium-sized agriculture and food product businesses. In an effort to sustain local economic development, the university wanted to build a system that would enable local fishermen, wine growers and others to contract for services through a portal that would enable them to decrease time-to-market, reduce transportation costs, reduce the amount of wasted products and improve overall product quality, IBM said.
The university worked with IBM to build a solution that leveraged cloud capabilities, System z, IBM software, storage and Global Technology Services. The system leverages these IBM assets to enable the university to deliver cloud computing services to the local community and allow multiple cloud entities to efficiently tap into heavy-duty computing power at minimal cost, IBM officials said.
However, it is the use of analytics that has been a major hit with users of the University of Bari system, particularly fishermen who are relying on the system's analytics to decide where to sell their fish. The cloud analytics solution analyzes market data and suggests to the fishermen where they should sell their fish that day.
"Before the cloud, when the fish catch was more than the market could take, the [excess] fish were tossed overboard," Visaggio said. "But now, after the cloud, when there are more fish, the thing is to sell it at low cost to a social institution."
So not only does the IBM-enabled solution help the fishermen pick the best market to sell their fish, but when they cannot sell them all, the cloud helps them find alternative ways to distribute the fish rather than to throw them away. That is an example of how IBM is helping to provide for a smarter planet, IBM officials said.
Moreover, IBM maintains that smarter computing enables business innovation, Rao said, citing Citigroup as a reference customer for going smarter. Rao said Citigroup, with 60,000 servers worldwide and 8,500 development servers serving 20,00 internal application developers, had an average server provisioning time of 45 days. "We created a cloud environment for them, and we were able to cut that provisioning time down to 20 minutes," Rao said.