IBMs Systems Strategy

 
 
By Darryl K. Taft  |  Posted 2010-03-30
 
 
 

Why Hardware Still Makes IBM Tick


Rod Adkins is senior vice president in charge of IBM's Systems and Technology Group,  the company's $19 billion semiconductor, server, storage and systems software business. He was named to the job in October 2009. Adkins has held a number of product development, business operations and general management positions since joining IBM in 1981, including stints as general manager for desktop systems in IBM's Personal Systems Group, for UNIX systems in Server Group and for Pervasive Computing in Software Group. In 2002, Fortune magazine named Adkins one of the 50 most powerful black executives in America.

For this interview, Adkins spoke with eWEEK Senior Editor Darryl K. Taft on a variety of subjects, including competition with Oracle and Hewlett-Packard, how systems play into the overall IBM strategy and topics such as the IBM Smarter Planet initiative.

IBM has been experiencing share gains across the systems and storage markets, according to the major analyst tracking reports. What do you think is going on in the market?

I think the market is responding to the business value it sees in our capabilities for addressing their data center requirements. IBM was the world's No. 1 overall server vendor in 2009, according to Gartner and IDC, and we gained share points in both their tracking reports. In Unix, as the leader in the market, we added about 4.5 points of revenue share while HP and Sun [Microsystems] lost share. In x86, IBM experienced the largest revenue growth of any vendor in the fourth quarter-3.5 percent. It was IBM's fourth consecutive quarter of revenue share gains in x86. In storage, we outgrew the overall external disk segment in Q4 and passed HP to take the No. 2 position in the market. That's all according to IDC.

Why do you think IBM is taking share from the competition?

The world is changing. As the planet becomes more instrumented and interconnected, the amount of data and the number of transactions will continue to increase-driving the need for more real-time analytics and prediction capabilities. Every piece of data generated is a building block of insight that can predict what happens next. This is impacting virtually every system inside a company, a city, the world. This applies to health care systems, retail systems, financial systems, energy grids, supply chains-and to the world's data centers. Opportunities are unlimited, but resources are not. As a result, IT departments will process more transactions and manage exponentially more data. Yet they will need to do it more efficiently than ever, consume less energy and make the most of the existing floor space.

The need for a new definition of performance is becoming increasingly clear. And that's what I think is behind the momentum we're seeing in our systems business. We're not pursuing a strategy to drive more common PC parts into enterprise servers to gain our own operational advantages, like some vendors. We take a very client-centric approach. We believe it's the innovation you wrap around those parts and the deep integration and optimization between the hardware and software that can attack the pain points enterprises are seeing and their need for lower costs and better management. It boils down to three distinct advantages we have in our systems business: the combination of our deep understanding of client needs, a comprehensive systems portfolio and technology leadership.

Take our recently announced eX5 systems on the x86 platform. We had engineers at work for years to come up with an industry first-decoupling memory from its usual, tightly bound place alongside the server's processor so you can dramatically scale memory in new ways and get much more out of the individual system. That is changing the economics of x86-based computing as we lead the way in re-engineering the 30-year-old PC server architecture. In addition to more cost-efficient use of memory, we've achieved technology breakthroughs in virtualization and energy savings. You heard these themes not just in the eX5 announcement, but also in February when we introduced new Power7 systems. Our Power7 systems ushered in a new category of performance for our clients. We call it "intelligent performance," rather than just raw speed, to manage billions of transactions in real time and analyze the data on the fly. This is key for emerging workloads like smart utility grids.

All this makes us unique in the marketplace. IBM's long-term strategy has been driven by our investments focused on innovation and integration. This is much more than being a vendor declaring a strategy based on a recent acquisition or declaring a strategy based on delivering commodity hardware through supply chain management. We've been at this for decades, and we're designing new integrated systems-from chip to end product to everything in between-optimized for a new generation of demanding workloads and to help our clients reduce their IT costs for existing workloads.

IBMs Systems Strategy


 

As you mentioned, IBM made a couple of major systems announcements in the first quarter of 2010-the new Power7 servers and eX5 on the x86 platform. Workload optimization was a major theme in both. Can you describe IBM's strategy there?

Workload optimization reflects how clients run IT to support their businesses. Sure, technically any platform can be configured to run any workload, but in reality workloads influence platform choices. For example, a typical transaction processing workload like core banking is characterized by a high rate of transactions. Scale matters and quality of service is extremely important. The Bank of China, for example, runs 380 million accounts on a single System z that processes 10,000 transactions a second with near zero down time per year.

For all the reasons I laid out about the way the world is changing, workload optimization will become even more important. Delivering greater value to CIOs means making this concept of intelligent performance easier to implement and manage. We're not asking them to standardize on a single platform, but we're doing the heavy lifting-the innovation, integration and optimization-for them. Hardware alone can't provide the breakthroughs. Consider that historically computational power doubles every two years-a factor of 32 over a 10-year period. We believe that to address the demanding requirements of emerging, data-intensive workloads, what's needed is a fourfold increase of computational power every two years. That means over 10 years-it's an exponential increase-a factor of 1,000. That's a really significant engineering challenge.

Delivering that kind of exponential improvement will require optimization at every level of the system. For example, Power7 isn't just about a new chip and new hardware-it's about integrated hardware and software to manage millions of concurrent transactions. We dramatically increased the parallel processing capabilities of middleware such as WebSphere, DB2 InfoSphere Warehouse and Cognos for managing data, transactions and analytics to support Power7 systems. That means clients or ISVs don't have to rewrite existing applications to take advantage of Power7's advances. IBM is leading this shift-leveraging the breadth of our portfolio of systems and software-and our unmatched expertise to deliver new levels of both innovation and integration.

Despite the company's history, the common assumption the last few years has been that IBM was becoming mainly a software and services company. IBM's recent aggressiveness and momentum in systems seem to contrast with that. How does IBM see systems fitting into its overall strategy?

IBM invests over $3 billion a year in systems and technology R&D for a reason. Our clients need innovative systems to help them reduce costs and support new business models. And IBM is innovating at every level: semiconductor, processor design, hardware design and architecture, operating system, systems management software. We work very hard at delivering highly optimized systems for transaction processing, analytics, business applications and Web collaboration. We know what our clients are looking for: innovation across a range of systems; high performance for ever-more-challenging workloads; extremely high levels of security and high availability; energy conservation; and, of course, low operating costs.

And that is driving our momentum. IBM has logged more than 2,100 customer wins in our Power servers from 2006 thru 2009, including nearly 1,050 versus Sun and nearly 825 versus HP. That has resulted in more than $2.15 billion in revenues. It's an average of more than one customer per day moving to IBM. Or I could talk about how we're building an impressive list of System z wins in emerging markets-such as First National Bank of Namibia, which purchased that African nation's first-ever mainframe last year. I have a pretty exciting job these days-it's an especially fun time to be working on systems at IBM.

IBM Rivals and Clients


 

Oracle also has been talking about how they're looking at vertically integrated optimized systems. It sounds like IBM's strategy-they even said they want to be like IBM of the '60s. How is Oracle's approach different, or what sets IBM apart?

They do talk a lot. Just one problem: When Oracle talks about Oracle software and Sun hardware being "vertically integrated systems," in reality it's just packaging a bag of parts and putting a label on it. Workload-optimized systems require years of R&D and close work with clients. Optimized systems require long-term investments, deep integration, industry knowledge ... partnerships with clients and business partners. It's a long list. We've invested billions. We've worked with clients, the world's top banks for example, for decades on the integrated systems that run today's global banking transactions. And now we're widening the gap with most innovative and integrated systems, optimized for the most demanding workloads. You can't do that with a bag of parts.

What about HP?

Leaders in this market innovate. They attack the real pains their customers are experiencing. But even today, some in our industry, such as HP, still seem to believe that information technology has become a commodity. If you really understand where the technology is going-that it is moving out of the back office and into the front lines and into the core business of every organization-you see the idea of "smarter systems" is resonating with decision makers. These decision makers have real business needs. They want an infrastructure that is not only as secure as a laptop PC, they need to process exabytes of data, in real time. They do not want to revert to the inefficient sprawl of the 1990s. They know what that's like when they pour all their resources into operating and maintaining the mess.

We are accelerating our market position through continued innovation at every point across the stack, delivering what I call intelligent performance versus raw performance. We can definitely scale the technology to get the raw performance, but given the complexity of today's IT infrastructure environments, IT managers also need to see improvement and simplification in how they manage that environment: virtualization, consolidation, systems management and workload optimization. So that's what we're focused on and, ultimately, what differentiates IBM from our competitors.

Can you give me a few examples of IBM clients that are using IBM systems in innovative ways?

Rice University is collaborating with Texas Medical Center on a research project to better understand root causes of cancer and other diseases. Rice told us that memory was a huge issue for them. Existing systems didn't provide enough memory for data-intensive work such as genomic sequencing, protein folding, drug modeling and simulating a multitude of molecular interactions in normal and unhealthy tissues. IBM responded to this feedback when we built innovative memory-doubling features into Power7. Rice also told us that fast parallel processing is very important to its medical research-that is, being able to process multiple jobs in real time so that teams of researchers in disparate locations are working off the latest information. This feedback played a part in IBM designing Power7 with strong parallel processing capabilities.

Another good example is smart electrical grids. eMeter provides software for smart electrical grids, and the company is a great example of an ISV choosing IBM technology for emerging, data-intensive business models. What does this look like in the real world? One eMeter client is CenterPoint Energy in Houston, an $8 billion energy wholesaler with more than 5 million electric and gas customers across six states. CenterPoint Energy has deployed more than 100,000 smart meters and expects to install more than 2 million meters by 2014. They're capturing power usage data from homes and businesses every 15 minutes. In late March, eMeter and IBM announced a unique bundled software package, available preloaded on Power7, that helps utilities customers do Smart Grid implementations out of the box. This integrated bundle can help utilities cut Smart grid implementation and test time from a year to six months, and drop 60 percent off the implementation cost.

One more example is Acxiom Corp., a leader in interactive marketing services and early user of eX5 systems. Acxiom analyzes massive amounts of rapidly ballooning Web-based consumer data on behalf of its clients-including seven of the top 10 retail banks and nine of the top 10 auto makers. They've gone from 4 petabytes one year ago to 7 petabytes just six months ago to more than 10 petabytes of data today. Their CIO calls IBM's new eX5 systems 'game changers' because of their ability to let Acxiom double their virtualization capacity, dropping software licensing costs.

Rocket Fuel