BlueMatter, a new algorithm
Computing: What’s Next?
June / July 2010
Volume 8 Number 3
As computing progresses, technology increasingly is infused into our daily lives. We see it in consumer products such as mobile phones and personal computers. But its impact in these devices pales in comparison to how the instrumentation of physical infrastructures and natural systems will change our lives over the next few years.
Major systems—electric grids, for example—are becoming digitized, opening the door to far greater efficiency in power distribution. Sensors are being built into roadways to help predict traffic jams before they happen, food containers are tracked to help reduce the chance of contamination and massive, aging sewer pipes running beneath a city can be monitored to enable them to be repaired before they break.
This tremendous progress raises new questions for leaders across every industry about where computing can go, where it can take business and society, and how to prepare for the opportunities it will provide.
The same questions are asked by IBM scientists and engineers around the world as they focus on making discoveries, achieving breakthroughs and creating new technologies. They ask these questions not only because they are looking for the next best thing, but they also know the projects they work on can have a major impact in improving the way the world works.
We're watching it take shape already as advances in analytics, cloud computing, high performance computing and the convergence of information technology and wireless infrastructures reshape what we can see, understand, predict and improve in areas ranging from cities to financial services to energy to water resources.
The amount of data that exists in the world continues to explode. Every day, 15 petabytes (15 quadrillion bytes) of new data is generated by businesses and individuals—eight times more than the total amount in all U.S. libraries, according to the International Data Corporation.
But data is not information or intelligence. According to the recent IBM CIO study, one in three business leaders make critical decisions without the information they need. To be successful today, companies must be fast—but they also must be right. It no longer is sufficient to base business decisions on simply prior experience or raw data.
Key decisions today need to be based on analytics—advanced algorithms that give us the ability to see patterns in vast data sets. Analytics can turn data into information and intelligence, allowing us to model and predict the outcomes of business decisions.
With analytics, organizations can move beyond the traditional notion of sense and respond; they actually can predict the outcomes of certain business decisions. The ability to speed decisions, understand consequences and predict outcomes never has been more available—or more important —than it is now. It soon will become foundational to the way companies and perhaps even individuals operate.
Because of this, IBM Research is pursuing leading-edge research to extend the ability for analytics to help predict outcomes and improve the speed and quality of business. Just last year, IBM announced the creation of an entirely new kind of consulting organization dedicated to applying analytics to help companies improve the speed and quality of business decisions and understand the outcomes.
Our more than 200 mathematicians and 4,000 analytics-focused business consultants are unlocking new possibilities for organizations in fields as diverse as traffic congestion, oil exploration, customer churn and retention, supply chain efficiency, financial intelligence, risk management, reputation and customer sentiment analysis, disease modeling, fraud detection and many more.
When we consider that the proliferation of sensors, digital communications and other forms of digital data collection has led to a projected tenfold growth in digital data between 2007 and 2011, organizations must find new ways to manage their diverse computing workloads.
Given these pressures on current IT environments and data centers, IT executives are looking for a new model—and cloud computing has emerged as an important option.
But the buzz surrounding cloud computing is based mostly on a new kind of user experience—particularly in the consumer web space—for search, social networking and retail. From the consumer perspective, cloud computing is a means of acquiring services without needing to understand the underlying technology.
From a technology perspective, cloud is both a new consumption and delivery model. Cloud represents a massively scalable, self-service delivery model where processing, storage, networking and applications can be accessed as services over the Internet. And, enterprises are expected to adopt cloud models to help improve employee productivity, deploy new products and services faster and reduce operating costs —starting with workloads that are ripe for this environment, including: development and test, virtual desktop, collaboration and analytics.
In practice, cloud services will be deployed in two modes, public and private. While each will leverage similar technical attributes and focus on the simplified cloud delivery architecture, they represent different business and operational models with different goals. Each approach strikes a balance. Public clouds may provide a greater degree of standardization for functions such as search, but carry a higher security risk for mission critical or personal data. Private clouds enable greater customization, while enabling client control of data and more robust security and privacy. Many companies are looking for a hybrid cloud environment that exploits the best of both worlds and will do well to look for vendors who can help manage hybrid cloud environments.
Populations across the globe are using wireless devices to stay connected. According to the International Telecommunications Union, there will be 5 billion mobile phone subscriptions worldwide at the end of 2010. That number will continue to increase as mobile devices increasingly become computing platforms in many emerging countries where PCs are not affordable or where literacy is not pervasive.
IBM estimates that over the next five years, mobile data traffic will increase by more than ten times the amount we see on networks today. As traffic increases, so does the need for more efficient use of radio spectrum and wireless infrastructure capacity.
Additionally, as more organizations utilize sensors and analytics capabilities to drive more intelligence and predictability into their operations, traditional cellular mobile-based monitoring and tracking functions will give way to more complex communications, creating even greater demand for wireless capacity.
This will all take place while users raise their expectations for services and applications on their mobile devices to mimic computers and workstations. And as users’ requirements grow, so will the demands of companies who want to give customers the ability to wirelessly carry out tasks such as paying bills, tracking their water and electricity use, and mobile shopping. All of this will catalyze the convergence of IT and wireless infrastructure. Over the next few years, wireless operators will attempt to reduce backhaul by deploying technology at base stations and core networks to bring services closer to end users. Additionally, as these base stations become more complex, there will be a need for new computing architectures that enable more powerful wireless services, while computing clouds will be built specifically for wireless traffic.
Given the anticipated growth in wireless bandwidth requirements and infrastructure constraints, there will be a major opportunity for companies to create and engage in new computing systems and solutions.
In its Strategy for American innovation, the Obama administration listed exascale computing (1,000 times greater than petascale) among the grand challenges of the 21st century in science, technology and innovation. Meeting these challenges the administration said “will allow the nation to set and meet ambitious goals that will improve our quality of life and establish the foundation for the industries and jobs of the future.”
The Department of Energy has found that exascale computing will revolutionize the way we approach challenges in energy research, environmental sustainability and national security. Exascale computing will bring us to a tipping point in predictive science.
Exascale computing has the potential to address a class of highly complex problems that have been beyond our reach, not just due to their sheer size, but because of their inherent uncertainties and unpredictability—problems like climate change or customized medicine. The way to deal with such uncertainty is to simultaneously run multiple ensembles or copies of the same applications, using many different combinations of parameters, and thus be able to explore otherwise unpredictable problems.
For example, exascale could significantly boost the success rate of oil recovery techniques Exascale systems also might help doctors turn around disease prediction, identification and cures closer to the point-of-care, which could allow civilization to outrun the future epidemics.
But the challenges of exascale are great. The degree of challenge in this next big step is much more significant than the leap from teraflop to petaflop, which largely was contingent on optimizing processor technology. If we continue along with current technology, an exascale system would require a tenfold increase in power consumption.
Additionally, because the way to now increase performance toward an exascale system is massive parallelism, an exaflop supercomputer might have hundreds of millions of processing elements or cores, and the industry will need to figure out how to program that many processors. Such massive parallelism will require major innovations in the architecture, software and applications for exascale systems.
Achieving these grand challenges will help us solve major business and societal problems. At IBM Research, we are committed to collaborating with others to be the first to achieve a computing system that operates in the exascale regime.
As we focus on building a smarter planet, we are exploring ways to help solve some of the big issues of our time. Often, this involves collaborating with other leading institutions. We are inspired by the possibilities of what we can do when we look at a problem with new lenses and think of new ways to solve it. And we are excited and optimistic about how computing will improve the way the world works.
Dr. John E. Kelly III is senior vice president and director, IBM Research.
Copyright © 2012 | Innovation America