Perceptive Information Strategies

When Graeme McDowell tapped home his putt to seal a championship at the U.S. Open on Sunday (June 20), spectators who packed the stands and stood shoulder-to-shoulder around the green roared their approval. Tiger Woods and Phil Mickelson, the Open’s superstars, were humbled by the Pebble Beach course and its famously changeable weather. The little-known McDowell “survived,” as several commentators put it. But that doesn’t really give him enough credit. He played a smart, safe game that adapted well to course conditions.

(Photo credit: Lance Iversen, The Chronicle)

The same might be said about IBM’s technology operations, which in partnership with the U.S. Golf Association’s Digital Media team stood the test of a massive number of virtual fans visiting online and mobile U.S. Open sites. IBM and the USGA said that over four million visitors came to the U.S. Open’s Web site, about 8 percent more than last year. This was the first big year for the mobile site, which had nearly two million visits. A major attraction was the “Playtracker” application, which enabled users to fly over the course and get visualizations of how the course was playing through heat maps based on scoring feeds. You can imagine the potential for future data-driven visualizations based on historical data about courses, players, pin positions on the greens and much more.

IBM’s technology management of the U.S. Open site offered a case example of how virtualization and workload management are becoming the essential ingredients of scalability, availability and agility, certainly for consumer Web sites like the Open’s. The USGA is no stranger to IBM’s virtualization technology; IBM has a close services partnership with the USGA, which includes running a variety of cloud services for the Association from its data center in North Carolina. When I visited the trailer near the Pebble Beach course where Web site and scoring services technicians were holed up, I couldn’t help but be amazed at the simplicity of the dashboards that offered real-time views of workload performance on a virtual platform of servers located across the country.

As John J. Kent, IBM Program Manager for Worldwide Sponsorship Marketing explained, virtualization is critical to utilization efficiency, enabling IBM to combine several workloads onto a single platform. “Virtualization basically makes the distributed environment into a mainframe, which has had this virtualization capability forever,” he said. Kent heads up IBM’s technology partnerships with other events, including this week’s Wimbledon Championships tennis event. Kent said that tennis is actually the more data-rich game, with fans already interested in analysis of “all the potential data points – such as unforced errors and rally counts – that can help you understand the strength of a player’s performance.”

In distributed environments, scaling up has always meant adding more hardware; with virtualization and cloud computing, organizations can avoid the long “cap x” procurement process and simply request more of what they need, and it can be made available rapidly over the network. What’s key, then, is to understand and monitor their workloads so that they can be optimized as demand rises and falls; then, organizations don’t have to spend on procuring enough servers to match peak workloads – but otherwise let them sit idle.

The other performance throttle IBM needed during the Open was to regulate content flow. Bandwidth is now the chief bottleneck; the explosion of advanced mobile devices in particular has moved users ahead of what networking providers are able to offer. IBM and the USGA’s Digital Media team needed the ability to make dynamic decisions about regulating content flow. “We needed to understand content demand well,” said Kent. “We were able to slow scoring updates, for example, if we were reaching a threshold in demand for content access and live streaming.” Thus, workload intelligence is critical to managing unstructured content as much as it is for data.

The USGA needs to provide a rich virtual experience on mobile devices to capture a younger demographic, which is important not only for the continued success of professional golf but also for attracting advertising on its Web site. However, as fans grow more dependent on the experience delivered by their mobile devices, it will be interesting to see if the USGA responds to pressure to allow those who attend the Open to bring them, which they are currently prohibited from doing. While there are good reasons not to have onsite fans working their mobile devices and interrupting the lovely hush before a player takes a swing, I wonder if the USGA will have to bow to the inevitable. Otherwise, fans might prefer to stay outside, where they can enjoy a rich, virtual experience.

But in any case, from an IT perspective, the key to victory in the U.S. Open and similar high performance events is clear: Know the workload and optimize it through the virtualized infrastructure. The victorious Graeme McDowell set a good example.

Oracle’s acquisition of Sun Microsystems continues to reshape the competitive landscape in the software and IT industry. Perhaps nowhere else is this more apparent than in the long-running battle between Oracle and IBM. By packaging together database software and systems in Exadata – especially with storage and server technology from Sun, which was already an arch IBM competitor – Oracle has ignited a “stack” war with Big Blue. Pick up the Financial Times or other business press, and you are likely to see one of Oracle’s trademark no-frills advertisements claiming Exadata’s benchmark performance superiority to IBM database systems.

IBM has hardly taken Oracle’s jabs lightly; for example, it has responded recently with benchmark results that assert lower overall database system costs compared with Oracle/Sun systems. I’m not going to write about benchmark wars here, but I will say that as stacks turn into pre-configured appliances, apples-to-apples comparisons of price and performance get tougher. It is important to examine closely the specifics of what the marketed benchmark results are reporting: in other words, whether the price-performance numbers account for all software and hardware costs, just hardware or whatever. This is especially true for organizations evaluating database appliances that offer pre-configured systems that integrate software, storage and server technology.

My focus here is on how IBM, with its recent software and systems announcements to support its “smarter systems for a smarter planet” strategy, is aimed at changing the basis of competition. To be sure, IBM has been traveling in this direction for a lot longer than just since Oracle’s acquisition of Sun. However, at the launch event (April 7, IBM Almaden Research Center, San Jose), you could feel the temperature in the room rise whenever Oracle came up. IBM needed to respond to Oracle, but there’s a lot more going on than a battle of database machines.

The April announcements brought technology substance to IBM’s long-running campaign to educate the market about why “the planet” needs to be smarter. In short, the context IBM has been articulating is that public and private organizations in all industries are growing increasingly dependent on the flow of data for everything they do. This includes the data tsunami arriving in the form of sensor data, online clicks and comments, surveillance and more. If they wish to improve processes, performance, customer service, market intelligence and innovation, they need to use data effectively and be “smarter.” This must happen with all information activities, including analytics and transaction processing.

The significance from a technology perspective is that software and systems can’t be part of the problem. Organizations need technology that does not simply add to the headaches of poorly integrated information silos, no “end-to-end” view of performance and prohibitive costs for scalability and speed. It’s not good enough just to deliver a souped-up database machine; the technology must offer something more, so that the organization can become smarter, not dumber.

Each of the announced systems (preconfigured for x86, Unix/Linux and System Z platforms) has distinctive features; those based on the POWER7 processor were the most impressive. However, the unifying theme for all was workload optimization. Arvind Krishna, general manager for Information Management in the IBM Software Group contrasted “closed” appliances with IBM’s workload optimized systems, which he said are designed to take on additional capabilities and flex to the workload demand so that the system remains efficient, cost-effective and scalable. IBM Research’s Bijan Davari, IBM Fellow and vice president for Next Generation Computing touted IBM’s workload optimization leadership “spanning decades,” particularly in mainframes, and described ongoing research. IBM demonstrated how the DB2 pureScale Application systems optimize workloads for transaction-intensive systems, and the Smart Analytics systems do so for BI and analytics.

Pardon me for using political consultant James Carville’s ugly phrase, but to have smarter systems, it’s the workload, stupid. Organizations must understand their current and anticipated workload, and then choose the technology platform (or external service) appropriately. They must ensure that the systems they choose enable continuous analysis, management and optimization of workloads from the end-to-end, user experience perspective, so that as demand intensifies and more data flows through, performance does not suffer and costs don’t skyrocket. Move over, speeds and feeds: workload is what matters.