Special Features:

GGF'S WALTER STEWART LAYS OUT SOLID CASE FOR GRID

Recently, GRIDtoday had an opportunity, on the eve of GGF12, to speak to Walter Stewart, SGI's global coordinator of Grid strategy and co-chair of the Plenary Program at GGF, on his views on a variety of Grid issues.

GRIDtoday: Vendors talk a lot about Grid, but do you see real uptake of Grid for production use rather than just Grid research today? Who's building and using Grid?

WALTER STEWART: Grid in the past two years has become a serious tool for production in a variety of industries and within the research community. Two years ago, most Grids for research were Grids for research on Grids.

Today, there are researchers in cosmology, bio-informatics, chemistry, physics, engineering, seismology, medicine and a host of other disciplines gaining vital access to compute power, visualization and data management using the Grid. Many of those same researchers generate their data on highly specialized equipment such as cyclotrons, telescopes, and shake tables and move that data to the compute resources required to deal with it on the Grid.

In business and industry, Grids are making major inroads in the work of pharmaceutical, aeronautical, automobile, financial and energy companies. The Global Grid Forum 12 meeting this week in Brussels has two full days of plenary sessions on industry experience in the use of Grid. Companies on the program include Novartis, Wachovia Bank, Intel (not as a supplier to the Grid, but as a serious user in its own internal processes), Magna Steyr, Danske Bank, Boeing, BAE, The Toronto Dominion Bank, British Telecom, Johnson & Johnson and Lion BioScience.

Gt: Do you see Grid moving beyond a niche use by scientists and researchers?

STEWART: The Grid has definitely moved beyond the world of science and research. Again, this week's GGF12 is testimony enough to the move. The move is going to accelerate over the next several years, and it's going to have broad industry impact.

Gt: What's driving the accelerating demand for Grid solutions in industry?

STEWART: Most of the world is sitting on IT infrastructure installed in 1999 to cope with the Y2K problem. Well, that infrastructure is getting past its sell-by date. More and more organizations -- both commercial and not-for-profit -- are beginning to recognize that they are not locked into replacing their '99 infrastructure with the 2004/5/6 versions of the same thing with the only expectation that they will be able to do what they've always done faster, cheaper and with a smaller machine.

Some organizations are recognizing that there are whole new orders of possibility for doing vital and valuable things we have never been able to do because, up until now, we've not been able to truly handle and profit from big data. Those organizations are beginning the process of building infrastructure for the knowledge economy -- an infrastructure that will give us the facility in generating, processing and interpreting big data. That is the promise of the Grid.

In the 19th and 20th centuries, we built elaborate distribution systems for the movement of raw materials and finished goods. Those systems were the keystone of the industrial economy. In the early days of the 21st century, we are building the same kinds of distribution systems, but this time it is for generating, processing, managing, collaborating and visualizing data.

A mere update to the infrastructure of '99 will not "infrastructure for a knowledge economy" make. That is why SGI is completely focused on ensuring that all our technology in computation, visualization and data management is completely Grid enabled and completely client neutral.

Gt: The concept of making computing resources available anywhere seamlessly is an attractive one, but also one that's been around for a while. What's so different about the Grid approach?

STEWART: You are right that the idea has been around for a long time in one form or another, but let's be clear: it has never been seamless.

The Grid is a metaphor that owes its inspiration to the electricity Grid. When we turn on a light or plug in a kettle, we don't give a nanosecond's thought as to where the electricity was generated. So we are working to ensure that same "not a nanosecond's thought" facility will make it possible to access the resources we need to work and the data we require for our work -- whether we are a brain researcher needing to compare hundreds of gigabyte-sized brain scans, a financial analyst needing to work through a petabyte-sized data set to assess risk, or a drilling engineer needing to evaluate drilling opportunities with a team distributed half way around the world.

The Grid is a product of extraordinary convergence of four key technologies. First, broadband networks. Second, extraordinary compute power. Third, applications built for distributed use -- such as SGI's CXFS File System over WAN and SGI's Open GL Viz Server for Visual Area Networking. And fourth, middleware that adds ever stronger functionality in scheduling, security and inter-operability, etc.

Never have the four been more aligned. More importantly, they bring the promise of even greater alignment to deliver a real infrastructure for knowledge -- Grid.

Gt: So, what's the real value to Grid users?

STEWART: Hyper-functionality with data. Any organization that is content with the way it deals with data today and sees little need to grow its ability to deal with data can afford to ignore Grids -- particularly because they are likely to be out of business in three years.

The Grid promises to provide a cost effective way of ensuring that the location of user, data, and compute resources becomes irrelevant in an organization's operation. The Grid also provides for the scalability in resources that will be critical to support the near perpendicular growth in data that we see today.

Many early Grid enthusiasts were completely fixated on distributed resources. At SGI, we are completely focused on distributed users who must collaborate with each other using big data to move their organizations ahead. The configuration and location of the resources should be a minor consideration. The Grid holds the promise of ubiquitous access to these resources.

Gt: But, the level of complexity required to implement a Grid solution today means that Grid is still inaccessible to any but the most sophisticated users, doesn't it?

STEWART: Absolutely not. True, a full-blown production Grid with complete beyond state-of-the-art functionality may not be available to all but the most sophisticated user. But Grids are built, not bought. Organizations can make a decision to take their IT infrastructure "Gridwards". They can consolidate their storage in SANs over WAN and avoid proliferation of data copies as well as proliferation of storage devices. They can make all their resources clients on those SANs and ensure that all have access to their most robust compute and visualization resources, regardless of OS. They can develop a plan which can be implemented incrementally to build an IT infrastructure equipped to meet the data challenges of 2010 rather than building a low-budget version of their 1999 infrastructure.

Gt: What role do vendors such as SGI play in Grid?

STEWART: SGI does three things for the Grid.

First, SGI builds Data Management systems that are fully Grid enabled. Using CXFS over WAN in alliance with Yotta Yotta, SGI makes Data Grid real today. Seamless access to data over tens of thousands of kilometers without making file copies. CXFS over WAN is client OS neutral, which is completely in keeping with the heterogeneous nature of the Grid.

In addition, SGI provides heterogeneous, collaborative, distributed access to high performance visualization of data over the Grid using OpenGL VizServer.

Finally, SGI delivers the world's most powerful capability and capacity computers fully configured as major compute engines on the Grid.

SGI ensures all its developments in hardware and software remain completely compatible with the latest efforts in middleware development for the Grid, and supports the wider Grid community through its engagement with GGF and with multiple Grids around the world.

Gt: Finally, Walter, how do you see Grid evolving over the next 5 years?

STEWART: Infrastructure. As we come to the end of the first decade of this century, we'll look back and wonder what all the fuss was about. It will seem so "of course."

Of course we can't have a knowledge-based economy without an infrastructure that will allow us to deal with the data -- the raw materials behind the knowledge on which we base the economy.

Of course the Grid is heterogeneous -- and heterogeneous way beyond just a question of operating systems. The Grid will be heterogeneous for multiple devices: sensing networks, data ingest systems, data management systems, visualization systems, compute systems and myriads of user devices most of which will bear little resemblance to today's PC.

Of course the Grid will be ubiquitous -- it can't be infrastructure if it isn't ubiquitous. We will stay a mobile people and we will collaborate more and more with one another as we deal with the big data challenges, so of course we need the power of the Grid, ubiquitously available.