It has been claimed that “IT doesn't matter,” with the implication that IT is now so commoditized that it can no longer be a significant source of competitive advantage.

Conversations with senior executives across many Fortune 2000 companies lead me to disagree with this assessment. True, companies are concerned with controlling IT costs. However, I also find a growing recognition that competitiveness depends on a company's ability to innovate (Steve Jobs says simply: “Innovation distinguishes between a leader and a follower.”). I argue here that there are important strategic opportunities in improving enterprise IT infrastructure to accelerate innovation.

Others have written at length on the importance of innovation, so I will not revisit that topic here. Suffice it to say that in today's increasingly global, hyper-accelerated, and winner-take-all markets, a company's ability to deliver superior products and services before the competition can often make the difference between success and failure. Thus, the ability to innovate consistently and rapidly has emerged as a key differentiator. Successful companies are increasingly focusing attention on the process of innovation and on the empowerment of the knowledge worker — the innovator.

Such competitive pressures are particularly acute in industries where product design and development require the use of sophisticated IT for complex and computationally intensive simulation, design optimization and data analysis (e.g., aerospace and defense, semiconductors and electronics manufacturing, oil and gas, automotive, pharmaceuticals, entertainment and digital media, and financial services, etc.).

In these industries, the knowledge worker needs more than a workstation and an Internet connection; developing a new product or service requires the manipulation and management of large quantities of data, access to large-scale computing and, in many instances, extensive collaboration within distributed teams. The following scenario (based on a real example, but with details changed) introduces key themes:

Strong competitive pressures demand that BestWidget Inc. reduce the time to develop the next version of its best-selling product by half — while also improving quality, reducing manufacturing costs and ensuring adherence to environmental standards. Achieving this goal requires that the design team, spanning five locations across the globe, turn around design revisions four times faster — while also performing an order of magnitude more testing and verification to increase product quality.

While this task is not expressed as an IT challenge, its accomplishment founders on IT issues. Paradoxically, the chief difficulty was not a lack of needed IT capacity and services, but an absence of efficient delivery mechanisms. Dramatic IT improvements over the past decade had given designers their own computers, and each workgroup its own cluster and storage system. These developments had freed designers and workgroups from the limitations of the central mainframe. However, with increasing design complexity, these developments had also become significant barriers to innovation. The ability to manage one's own data had become the burden of managing one's own data (in another industry, chip designers can spend 25 percent of their time managing data!). Convenient access to local data had become inconvenient access to other data — when the data required to complete a design was located at a dozen sites worldwide. The power of a dedicated workstation had become the limited capacity of a single machine — when design goals required thousands of computers. The consequence is that BestWidget designers produced an inferior product, behind schedule and over budget. Not because they are bad designers — on the contrary — but because they just couldn't get access to the data, computing and other resources they needed to do their jobs.

These complaints are all symptoms of the “distributed computing hangover,” a situation where completely decentralized management makes it impractical to allocate resources in alignment with overall objectives. Such difficulties are becoming increasingly widespread and urgent as the importance of continuous, distributed and dynamic innovation grows. However, inherent in such difficulties is also an opportunity for IT to deliver significant competitive advantage.

The key is that knowledge workers should not have to wait for resources or have to adapt their work processes to the peculiarities of available resources. To this end, we must break down barriers that constrain both collaboration among team members and access by team members to needed resources. We must make it possible for innovators to pull needed computing, application and data resources into the innovation process, on a schedule that meets their needs. Furthermore, as multiple design teams are typically active, we must enable the innovative enterprise to balance competing demands for fixed resources, by expressing and enforcing policies that reflect the respective priorities of different design team activities.

In short, we must deliver to innovators:

• What they need from the IT infrastructure (data, software, computing resources, licenses, etc.) to accomplish necessary tasks.

• Where they need it (in terms of accessibility to the innovation team), regardless of the location of team members and required resources within or outside the enterprise.

• When they need it, so the environment is matched to the lifecycle requirements of the innovation cycle.

• Why they need it, meaning that innovation team activities are consistent with the overall business objectives of the enterprise.

If we think of the innovative enterprise as a high-performance automobile, then our goal in addressing what, where, when and why is to ensure that fuel (computing, data and other resources) is delivered to its engine (the innovators) when needed — not in a best-effort fashion, or after a multi-week manual provisioning process.

In this way, we can ensure that BestWidget designers can access and share data resources quickly, perform computations rapidly, and above all count on the availability of resources as they schedule their work. The company itself can create the highest quality products and services consistent with business priorities and objectives (and given available resources) across all competing tasks.

Enabling this agility requires new capabilities. It requires capacity planning mechanisms for matching supply and demand while taking into account constraints specified as business policies at each level of the infrastructure (ultimately, as in manufacturing supply chains, demand should drive resource planning and scheduling, within policy constraints, to deliver optimal service levels). It requires resource configuration, allocation and scheduling mechanisms to ensure that diverse and distributed assets throughout the enterprise are delivered as and when needed. It also requires monitoring and management mechanisms to track usage, to ensure that demands are met, and to diagnose and correct problems as they occur. Finally, these different mechanisms need to be integrated with enterprise IT infrastructure and tools.

No existing technology addresses all these needs. Product lifecycle management tools address information management requirements, but not the delivery of the computing environments needed to generate or process data. Cluster management tools and workflow tools address elements of workgroup operation and process, but not the larger questions of information delivery and computation scheduling across concurrent activities. Virtualization tools address the configuration of computational environments, but not other aspects of the physical IT infrastructure. Thus, enterprises are left attempting to support the innovation lifecycle by cobbling together disconnected proprietary tools in an ad hoc fashion. The result is non-standard, non-scalable, difficult-to-replicate and difficult-to-manage solutions with limited ability to respond to dynamic business conditions.

Where then should we look for solutions? I believe that Grid technologies have an important role to play. This claim should not be surprising. After all, members of the Grid community have been working for close to a decade on precisely the issues discussed here, with considerable success. For example, the LIGO astronomical observatory delivers 1TB of data a day to eight sites around the world, creating more than 120 million file replicas to date; the U.S. TeraGrid national infrastructure enables flexible, policy-driven access to computing and storage resources at eight science data centers; and the National Cancer Institute's Cancer Bioinformatics Grid provides access to data and services at 60 cancer centers. In each case, Grid technology (specifically, open source Globus software in these examples) is being used to accelerate the pace of innovation.

In the next year or two, I expect that we will see significant progress in the creation and application of IT infrastructures architected specifically to facilitate innovation, and a shift from thinking of IT solely as a cost center to recognizing IT as a value enabler. In the process, we will also see a significant change in how we think about the role of Grid technologies in creating robust, scalable, and adaptive enterprise IT infrastructures.

About Ian Foster

Dr. Ian Foster is associate director of the mathematics and computer science division of Argonne National Laboratory and the Arthur Holly Compton Professor of Computer Science at the University of Chicago. He created the Distributed Systems Lab at both institutions, which has pioneered key Grid concepts, developed Globus software, the most widely deployed Grid software, and led the development of successful Grid applications across the sciences. Foster is also the chief open source strategist and a board member of Univa.